case study of customer segmentation

August 27, 2021

Customer segmentation in retail: 6 powerful client case studies, are you still talking to all of your customers the same way in today’s hyper-competitive retail environment, that just won’t cut it. you need to use customer segmentation to send each customer unique communications and offers. here are 6 case studies demonstrating the value of customer segmentation..

Customer insights and segmentation can help you unlock a new competitive advantage, identify opportunities to grow customer lifetime value, and optimize campaign performance.

By employing data-driven customer segmentation, you can improve your performance across every sales channel and customer touchpoint. Customer data platforms (CDPs) like Lexer can help you manage your data effectively, create valuable customer segments, and automatically update audiences across other retail systems.

In fact, Lexer is the CDP of choice for leading brands like Quiksilver, Igloo, Nine West, Rip Curl, Supergoop!, and more. Here are 6 case studies from brands and retailers who have used Lexer's customer segmentation tools to implement data-driven retail strategies and drive results.

Customer segmentation case studies for acquisition

Black diamond.

An excellent customer segmentation example as it pertains to customer acquisition in the retail space is the case of Black Diamond. The business aimed at growing its direct-to-consumer business to improve personalization, acquisition, and retention. This is with a backdrop of a healthy wholesale business and a small DTC team without a dedicated IT team that could provide actionable customer insight.

Black Diamond enlisted the help of the Lexer team to overcome these challenges. The team was in charge of providing customer data and gathering insights into their behaviors. The insights helped the brand develop an agile strategy for customer acquisition and retention campaigns across all its channels.

Using the Lexer CDP, Black Diamond was able to cut their cost-per-acquisition (CPA) in half and double their return on ad spend (ROAS) . Additionally, there was a 1,101% increase in the revenue per email when targeting lapsed customers. All of this was achieved using a 5-phased process which included collecting and analyzing historical data, targeted lead generation, and using Lexer's high-value lookalike audiences to improve customer acquisition.

Brand Collective

With the advent of Covid-19, Brand Collective was looking for a way to drive online sales as the performance of their traditional brick-and-mortar stores had significantly been affected. The brand wanted data on their customer base as they looked for new ways to engage these new customers who were increasingly digital-first shoppers.

Using the Lexer CDP, Brand Collective was able to gain holistic customer data in real-time. The easy-to-use Lexer platform built targeted segments across all marketing channels, including their email, mobile, and search. These yielded an action plan that helped the brand take on new opportunities and avoid the risks of the ever-evolving marketplace.

The Lexer team enabled Brand Collective to customize their digital campaigns and messages sent to their segmented audiences. This drove a 220% increase in return-on-ad-spend, a 2x increase in new customer acquisition, and a 5x increase in revenue from paid channels.

Sign up for our monthly newsletter

Customer segmentation case studies for lifetime value growth.

The global surfing brand Rip Curl needed intelligent segmentation to help them identify high-level customers. Additionally, the team wanted to see an increase in impact while still minimizing its digital marketing campaign budget.

The brand decided on Lexer’s CDP to help it gain insights, perform advanced segmentation, and target customers. Additionally, Lexer helped them orchestrate omnichannel campaigns.

By working with the Lexer team, Rip Curl achieved an in-depth understanding of its customer, which would give the brand the insights it needed for high-value customer acquisition. Additionally, due to the advanced audience segmentation and automation, the business could now benefit from customer lifetime value growth.

Specific results included the achievement of 93% more revenue per segmented campaign in August and 15x higher income than the benchmark for Lexer segments.

PAS Group wanted to significantly reduce ad wastage, re-engage lapsed customers, and create unique customer experiences . Additionally, the group wanted its brands to stand out and grow revenue within the highly competitive fashion and apparel industry. All of these would be made possible by linking all customer data to help with data-centric decision-making.

Using Lexer's CDP, the brand was able to segment its customer audiences and deliver targeted campaigns to recent and lapsed customers on paid social and email. This resulted in a 4x return on their advertisement spending and an 18x overall return on investment. These were achieved through the unification of all online and offline purchase data with loyalty and engagement data, all of which provided a holistic view of PAS Group customer data.

Customer segmentation case studies for retention

Wondercide wanted to rely on the traditional direct mail in conjunction with digital campaigns to help with re-engaging high-value customers. By measuring key customer retention metrics and understanding the factors driving retention in their business, they were able to improve retention rates significantly.

Using the Lexer CDP, Wondercide sent out personalized direct mail postcards that drove an ROI of 600%. The direct mail reengagement campaign targeted lapsed and opted-out customers whose last order was within the previous year. It also targeted inactive customers who hadn’t interacted with the business within two years and lapsed customers whose previous orders had been more than two years past. As a result, the business experienced a 310% ROI for the opted-out segment, 203% ROI for the inactive segment, and 155% for the lapsed segment.

Mountain Khakis

In a bid to increase its holiday seasons sales, Mountain Khakis used the real-time insights provided by Lexer's CDP to activate segmented campaigns. Specifically, the brand was able to retarget its female gift-buyers with a "treat yourself” campaign that saw a 7.1x increase in sales 2-3 weeks post the campaign.

Additionally, the campaign resulted in a 5x return on ad spend from female customers. This translated to a 49% boost in sales just in time for Christmas and a 47% boost in total customers.

Effective customer segmentation begins with mastering your data

As a business, you need to lean on customer intelligence to orchestrate specific high-value customer segmentation.

Lexer’s customer data and experience platform provides you with customer insights tools , data enrichment tools , segmentation tools, and predictive analytics tools that helps your business identify and target the right audience. As the only CDP built for retail with native tools to support every customer touchpoint, we are well equipped to help you drive incremental sales from improved customer engagement.

Book a demo today to see how Lexer's powerful segmentation and personalization tools can help you drive incremental sales growth.

Speak with our retail experts.

The Leading Source of Insights On Business Model Strategy & Tech Business Models

Customer Segmentation: Types, Examples And Case Studies

Customer segmentation is a marketing method that divides the customers in sub-groups, that share similar characteristics. Thus, product, marketing and engineering teams can center the strategy from go-to-market to product development and communication around each sub-group. Customer segments can be broken down is several ways, such as demographics, geography, psychographics and more.

Table of Contents

Why customer segmentation matters

No matter how niche your brand may be, it is important to keep in mind that every customer is, in fact, an individual. What’s more, they deserve to be treated as such.

Of course, most businesses will not have the resources to cater to every customer on an individual basis.

They can, however, more broadly assess the needs of their customers according to certain metrics.

Customer segmentation in a nutshell

Customer segmentation is the process of separating your customers into groups according to certain traits (e.g. personality or interests) and factors (age or income level). 

So why should customers be segmented? There are several important reasons:

• It allows businesses to tailor marketing strategies and ad campaigns according to particular groups of people.
• It enables businesses to learn about their consumers on a deeper level. And with this increased understanding, to create better products that resonate with consumer needs.
• Enhanced customer support – since businesses with customer segments are better able to predict problems ahead of time.
• Conversely, segmentation may also identify groups of consumers previously unknown to the business – allowing marketing resources to be directed toward these untapped groups.

Now that we have a basic understanding of customer segmentation and why it should be implemented, let’s look at some common customer segment types.

Demographics 

Demographic data is relatively straightforward and includes information on age, gender, marital status, income, and education level.

It is perhaps the most well-known and well utilized of all customer segments because demographic data is easy to obtain through market research.

A simple example of demographic customer segmentation might involve the marketing of a high-end sports car.

The manufacturer may want to target consumers that are unmarried or divorced, have a high income, and be at or approaching retirement age.

While the above examples deal with business-to-consumer marketing , demographic segmentation can also be used in business-to-business marketing .

In this case, businesses may target the industry, job function, or company size as part of their marketing efforts.

Geographical segments detail such parameters as climate, zip code, land use (urban or rural), and the radius around a particular point of interest. But it also concerns the scope and extent of potential marketing efforts.

Smaller organizations, for example, may target consumers living in specific towns or cities. Larger organizations may target consumers according to their country or continent of residence.

If we return to the sports car example, let’s assume that the car is marketed primarily as a convertible.

As a result, the manufacturer may choose to target specific countries (or geographic areas) with sunny climates that are conducive to driving with the top down, so to speak. 

Public transport operators could also use geographic segments to target commuters living within 15 minutes of a train station.

They could use this information to develop a marketing campaign to convince commuters to leave the car at home and take the train instead.

Psychographics 

Psychographic segments include such things as socioeconomic class, lifestyle, and personality traits.

They also include factors that are big drivers of buying decisions, such as values, motivations, attitudes, and conscious or subconscious beliefs. 

However, psychographic data is more difficult to collect than demographic data. Why? Because it is more subjective and requires deeper research to unearth.

Psychographic segments and the information that comprises them are also more fluid because motivations, beliefs and values can change over time.

The luxury sports car manufacturer may target consumers whose values and motivations relate to status, freedom, and fine craftsmanship.

But if, for example, the consumer who bought a 2-seater convertible suddenly welcomed grandchildren into his life, he may then prioritize safety and reliability over status and freedom.

Of course, marketing departments cannot plan for every contingency. But they must be aware that psychographic customer segmentation is fluid and has the potential to shift over time.

Behavioural 

Behavioral segments include a consumer’s direct interactions with a business.

In other words, behavior dictates how they act according to their demographic and psychographic attributes. 

The behavioral segment encompasses spending habits, product/service usage, and the perceived or actual benefits of such usage.

Behavioral segments are derived from internal data that is collected by the business itself.

It may include data on how consumers use a product and the frequency with which they do so.

Furthermore, information may also include the specific benefits that the consumer is after, such as a time or money saving or loyalty status. 

Perhaps most importantly, behavioral segments clarify a consumer’s willingness to purchase.

If a typical sports-car driver likes to upgrade to the new model every three years, then it is the marketing team’s priority to understand this cycle and market to this segment accordingly.

Similar predictive behavioral learning is also utilized by Netflix, who segment their users according to their content preferences and then recommend content in similar genres.

Technographic 

Technographic segmentation is segmentation according to a consumer’s preferred choice of technology.

Think smartphones, software, operating systems, desktops, and apps.

As technology becomes increasingly prevalent in the lives of consumers, technographic segmentation has never been more important to marketing departments. 

Business-to-consumer marketing can also use technographic segmentation to target consumers according to their social media use.

In their Harvard Business School published book Groundswell , authors Li and Bernoff suggest that marketing teams further divide their technographic segments according to social media use.

Each “sub-division” requires a different marketing strategy . Some of the more common sub-divisions include:

• Creators – who maintain a blog or website or upload music or videos.
• Critics – who post reviews of products or services or who like to contribute to forums or blog posts.
• Joiners – who maintain active social media accounts.
• Spectators – who read blogs, listen to podcasts, or watch video content without contributing or participating. 

Business to business (B2B) also stands to benefit by technographic segmentation. Specific parameters in the B2B sphere include network and storage capabilities, cloud utilization, and big data technologies.

All B2B interactions should segment businesses according to the prevalence of their technological capabilities before the marketing strategy is developed.

Target market examples

To recap, a target market is a segment of customers most likely to purchase a company’s products or services.

While the two terms have some overlap, it’s important to first make the distinction between a target market and a target audience.

The target market is the end consumer who will use the product.

The target audience, on the other hand, is the focus of the brand’s promotional efforts. 

To illustrate this difference, consider the McDonald’s Happy Meal. The product itself is obviously consumed by children, but it is the parents who control the finances and what the child eats.

As a result, McDonald’s may promote the Happy Meal’s nutritional value or low cost – factors that appeal to the parents but which the child cares very little about.

To solidify the concept of a target market further, read through the following examples.

Nike started out marketing to professional athletes and then expanded its business model to incorporate “everyday” athletes and sports enthusiasts.

As part of its rebranding effort, the company analyzed the benefits of owning its apparel, shoes, equipment, and accessories.

From this, Nike defined a target market of mostly younger consumers who were interested in fitness and possessed the disposal income to invest in equipment and achieve their goals.

Today, most of Nike’s promotional efforts focus on aspiring athletes and runners in a way that is motivational and inclusive.

Vans is an American shoe manufacturer founded in 1966 that made the bold decision to champion alternative subcultures such as skateboarding and bicycle motocross (BMX).

The brand appealed to so-called “misfits and rebels” who saw these sports as not only a hobby or passion but a lifestyle choice.

Vans is now taking advantage of the athleisure trend target market and has a much broader appeal, but the company’s stores continue their retro, skateboarding vibe.

In a Manhattan store, for example, vintage posters of skateboarders adorn the walls with industry slogans and skateboards from popular brands.

Next to skateboard accessories such as wheels and trucks is apparel more reminiscent of earlier decades with muted colors and oversized logos.

Dior is a French luxury fashion house founded by Christian Dior in 1946.

The company primarily targets the so-called “Chardonnay Girls” target market which consists of confident, optimistic, fashion-conscious women in the 18-32 age bracket.

Perhaps unsurprisingly, this target market tends to live in world cities such as Moscow, New York, and Milan with above-average salaries and career prospects.

They have also a propensity to shop offline, but having said that, Chardonnay Girls are consumers that are more likely to become advocates for a brand and share their experiences with friends.

Thus, reducing marketing costs through efficient, customer-focused communication.

Customer segmentation examples

In this section, we’ll delve into some additional customer segmentation examples.

Region and culture

With more than 36,000 restaurants in over 119 countries , McDonald’s uses a subset of geographic customer segmentation to promote menu items to users from various cultures.

In India, for example, ads show McSpicy Paneer alongside Green Chili Naan-Aloo. 

Another region and culture-specific advertisement promotes the Maharaja Mac – better known as the Big Mac – which is “ made with handpicked ingredients from across India” and features the #TrulyIndianBurger hashtag. 

Customer segmentation based on the forecast weather conditions enables the company to predict the moods, needs, and purchase behavior of its customers.

This is usually achieved via the integration of real-time weather data into an existing personalization platform.

Segmentation based on the weather is especially important for retail brands whose products are highly seasonal.

A clothing brand based in the United States, for example, can segment its users based on location and direct those living in the colder northern states to a page promoting scarves, jackets, and gloves.

An undisclosed football club – but one of the largest in England – used weather targeting to recommend merchandise to fans based on their location which is positioned on a Google Maps image in the background.

Some airlines are also using the approach to promote destinations with warmer or sunnier weather than the customer’s home conditions.

Home Chef is a food delivery company that segments its customers based on their profession.

In one email campaign aimed at the healthcare and education industries, the company referenced the upcoming National Teachers and Nurses Day and took the opportunity to thank these individuals for their service.

For those that could verify their teaching or nursing credentials, Home Chef offered 50% off the cost of their first box of food.

Cart abandonment

Almost 70% of desktop users and 86% of those on mobile abandon their cart before finalizing the purchase.

This represents a major source of lost income that can at least be partly recovered with laser-focused customer segmentation.

To encourage users to complete their purchases, companies can create a series of drip campaigns or emails based on metrics such as product type or customer activity level.

Google’s approach for abandoned items in its Google Store is to send users an email with personalization, excellent copywriting, and a clear call to action.

This is normally accompanied by a message that creates urgency such as “ Our popular items sell fast ” and “ Going, going, (almost) gone ”.

Politics is a divisive issue that can easily result in negative publicity for a brand. But rather than shy away from the topic, some brave companies use it as a tool for advanced and highly targeted customer segmentation.

Ben & Jerry’s is one brand that uses political segmentation to sell different flavors of ice cream across the United States.

In the democratic state of Vermont, for example, it released an “Empower Mint” ice cream with a slogan that read “ Democracy is in your hands” .

Key takeaways

Customer segmentation is a crucial part of any marketing strategy , but some businesses may be daunted by the initial investment of time and money. 

However, customer segmentation concerns serving customers and serving them well. Those who do not invest in segmentation run the risk of losing their customers to a competitor.

Accurate and detailed segmentation allows businesses to understand their customers on a deeper level and increases the probability of retaining them.

For the business, this increases conversion rates and drives down costs.

• In essence, a target market is a segment of customers most likely to purchase a company’s products or services. A target market should not be confused with a target audience, which is the focus of the brand’s promotional efforts.
• Nike’s target market consists of younger consumers who are interested in fitness and possess the disposable income to invest in equipment and achieve their goals. 
• Vans once appealed to smaller alternative subcultures such as skateboarding and BMX. Today, the company’s target market has broadened to include athleisure wearers.

Key Highlights:

• Customer Segmentation Overview: Customer segmentation is a marketing technique that divides customers into sub-groups based on similar characteristics. This allows businesses to tailor their strategies to specific groups and understand customer needs better.
• Importance of Customer Segmentation: Market segmentation helps businesses understand customer preferences, locations, and communication preferences. Treating each customer as an individual is essential, even if catering to every customer individually isn’t feasible.
• Demographics: Segmenting by age, gender, income, education, etc., allows businesses to target specific customer groups effectively.
• Geography: Targeting customers based on location, climate, and proximity to certain points of interest.
• Psychographics: Segmenting based on lifestyle, values, motivations, attitudes, and beliefs.
• Behavioral: Segmenting by customer behavior, including spending habits, product usage, and benefits sought.
• Technographic: Segmenting based on preferred technology, such as devices, software, and social media usage.
• Target Market vs. Target Audience: A target market is the end consumer of a product, while the target audience is the focus of promotional efforts. For example, McDonald’s targets parents as the target audience for Happy Meals, even though children consume the product.
• Nike: Initially targeting professional athletes, expanded to include everyday athletes with a focus on fitness enthusiasts.
• Vans: Initially targeted skateboarding and BMX subcultures, now appealing to athleisure wearers.
• Dior: Targets confident, fashion-conscious women aged 18-32 with above-average salaries.
• Region and Culture: McDonald’s tailors ads and menu items based on cultural preferences in different countries.
• Weather: Brands use real-time weather data to personalize offers based on weather conditions.
• Profession: Home Chef offers discounts to customers in specific professions, like teachers and nurses.
• Cart Abandonment: Brands send targeted emails to customers who abandon their shopping carts.
• Politics: Ben & Jerry’s uses political segmentation to offer ice cream flavors tied to political messages.
• Key Benefits: Customer segmentation leads to better understanding, tailored marketing, improved product development, enhanced customer support, cost efficiency, and increased conversion rates.
• Risk of Not Segmenting: Businesses that don’t invest in customer segmentation risk losing customers to competitors and missing out on opportunities to connect on a deeper level.
• Bottom Line: Accurate customer segmentation leads to higher retention rates, improved conversion rates, reduced costs, and overall business success. It’s a crucial part of any effective marketing strategy .

Visual Marketing Glossary

Account-Based Marketing

AARRR Funnel

Affinity Marketing

Ambush Marketing

Affiliate Marketing

Bullseye Framework

• Brand Building

Brand Dilution

Brand Essence Wheel

• Brand Equity

Brand Positioning

Business Storytelling

Content Marketing

Customer Lifetime Value

• Customer Segmentation

Developer Marketing

Digital Marketing Channels

Field Marketing

Funnel Marketing

Go-To-Market Strategy

Greenwashing

Grassroots Marketing

Growth Marketing

Guerrilla Marketing

Hunger Marketing

Integrated Communication

Inbound Marketing

Integrated Marketing

Marketing Mix

Marketing Myopia

Marketing Personas

Meme Marketing

Microtargeting

Multi-Channel Marketing

Multi-Level Marketing

Net Promoter Score

Neuromarketing

Newsjacking

Niche Marketing

Push vs. Pull Marketing

Real-Time Marketing

Relationship Marketing

Reverse Marketing

Remarketing

Sensory Marketing

Services Marketing

Sustainable Marketing

Word-of-Mouth Marketing

360 Marketing

• Market Segmentation
• Brand Awareness
• Types of Business Models You Need to Know
• Business Strategy: Definition, Examples, And Case Studies
• Marketing Strategy: Definition, Types, And Examples
• Platform Business Models In A Nutshell
• Network Effects In A Nutshell
• Gross Margin In A Nutshell

More Resources

About The Author

Gennaro Cuofano

Discover more from fourweekmba.

Subscribe now to keep reading and get access to the full archive.

Type your email…

Continue reading

• 70+ Business Models
• Airbnb Business Model
• Amazon Business Model
• Apple Business Model
• Google Business Model
• Facebook [Meta] Business Model
• Microsoft Business Model
• Netflix Business Model
• Uber Business Model

• For Beginners
• For Experts
• Email All you need to know about email: strategy, solutions, etc.
• SMS What makes brilliant text message campaigns.
• Viber Viber messages that help reach out to your customers.
• Web Push Wield the power of web push notifications.
• Mobile Push Why mobile pushes matter for your business.
• App Inbox How an App Inbox can increase your marketing campaigns
• Omnichannel Introduce omnichannel solutions to your marketing strategy.
• Segmentation How to segment your customers to grow revenue.
• Recommendations + AI AI-powered product recommendations and how to use them for ecommerce.
• Automation Best marketing automation practices to transform your campaigns.
• Analysis How to analyse and improve marketing metrics.
• Data Collection (CDP) Data sources and collection methods.
• Copywriting Content your subscribers actually want to read.
• Design Awesome email marketing designs and how to create them.
• Calculators With our calculators, you’ll save lots of time.
• Cases Brilliant marketing case studies to learn from.
• Ideas Marketing ideas to boost your sales.
• Strategies Proven marketing tactics that generate more revenue.
• How To Marketing campaigns and analytics easily explained.
• Comparison Best email marketing platforms: review & analysis.
• News Features, news stories and fresh updates on the marketing trends.
• Video Marketing life caught on camera.
• Interviews Digital marketing explained by the industry leaders.
• Conferences Email marketing events you simply can’t leave unnoticed.
• What's New Introduce new features to your marketing strategy.
• Glossary See the comprehensive list of basic marketing terms, collected to help you create and improve your marketing strategy.
• Webinars See, Explore, Apply
• Email Examples Inspire your email marketing campaigns with our huge collection of email designs from a variety of industries and occasions, with comments by our experts.

Alex Danchenko

COO, co-founder Yespo

Maya Skidanova

Content Writer

Valeriia Shudryk

Content Marketer

30 January 2024

Customer Segmentation: Key Role, Types, Usage, and Case Studies

Irrelevant marketing campaigns that don't address customers' interests and pains can quickly turn them away from a company. According to the study , 49% of consumers are annoyed by such messages, and 42% feel frustrated when they receive messages that do not meet their needs.

Audience segmentation helps to personalize marketing campaigns by focusing on separate groups of contacts that share common needs and characteristics. By understanding the wishes and demands of segment participants, marketers can find the most relevant offers for them. This focused approach drives customer conversion and improves customer experience. As a result, it increases the return on marketing investment.

Understanding Customer Segmentation

First of all, let's take a look at the definitions.

A segment is a group of contacts united by a certain characteristic, condition, or set of specific characteristics, conditions, and criteria.

Customer segmentation is

a marketing strategy that divides consumers into separate segments. Members of each group share commonalities, such as interests, needs, goals, and buying behavior patterns. The goal of customer segmentation strategy is to better understand and meet the needs of specific segments and the audience as a whole. In the context of Customer Data Platform (CDP), segmentation is the process of dividing a company's clients into separate groups based on certain criteria or attributes.

The criteria for segmenting the customer base depend on:

• the industry in which the business operates;
• its target market;
• the quantity and quality of data;
• its marketing goals.

CDP helps marketers segment the database and manage these groups within a single system: run triggered marketing campaigns, analyze the groups, and perform A/B testing.

How to Approach Marketing Segmentation: Principles and Creation

Three main types of segments are key to the business, and the marketing strategy should be based on them. Let's explore their specifics in more detail.

Strategic Segments

By strategic segments, marketers mean customer groups that are most important to the company from a financial, conceptual, and communication perspective. The participants in these major segments are categorized according to:

• Life cycle (from the phase of acquaintance with the company to the level of a long-term loyal customer);
• Activity (sales, views – any actions or lack thereof);
• The product or service used by the consumer;
• Basic characteristics (gender, language, location, etc.);
• By the source of customer acquisition.

Several signs help detect a strategic segment. First, you can identify it based on specific numbers you can trust (e.g., the average check amount, the number of items in it, the NPS score, etc.). Second, the segment size is enough to use for tests and experiments. Third, there is a specific way to motivate people from this group to take your desired actions. For example, newcomers need an explanation of the benefits of buying from you, VIP customers need special service conditions and personal offers, etc.

Derived Segments

Derived segments are based on strategic groups. They are a combination of existing strategic segments but with additional clarifications and characteristics. Let’s say if we do a segmentation based on gender alone, we’ll get 4 segments: men, women, them, and ‘not specified.’ The same can be said for language, which is especially important for companies operating in a global market. By combining, adding, or extracting some characteristics, a marketer creates derived segments. For example, from the group of customers who rate our product highly, we can identify residents of a particular city whose average purchase check exceeds $250. In addition, we can remove from this subgroup those who have received marketing messages from the company in the past week. This gives us a separate segment to work with.

Microsegments

Microsegments are the solution for automation, such as trigger activation. To create them, marketers set complex conditions based on the behavioral characteristics of customers, which are constantly changing .

For example, you can add to a single segment all customers who viewed an out-of-stock product that is now available. This way, you can send a highly targeted campaign ‘Back in stock.’ Such segmentation depends on the quantity and quality of customer data. The more data you have, the more targeted trigger marketing campaigns you can set up.

The Purpose of Segmentation and the Problems It Solves

As a result, segments are needed to effectively achieve several key objectives:

• Run marketing campaigns.
• Conduct research (for example, how different segments respond to a particular message. Let's say the OR may be significantly different in the group of active readers, new subscribers, those who read the message but don't click on the link, etc;).
• Run tests (split tests, create control groups).
• Monitor (for example, set up group size tracking and automatically send a notification to the marketer if the size drops to a critical level).
• Automate processes.

One of the main benefits of segmentation is that it helps businesses solve common problems. Let's explore how.

1. Non-personalized marketing – due to the lack of information about customers' personality and their needs, companies send marketing messages that do not resonate with their targeted audience. At the same time, statistics reveal that:

• 76% of customers admitted personalized communication as the major factor making them look closely at a brand.
• 78% said personalized content in marketing messages makes them more likely to buy something again.

Customer segmentation allows you to gain valuable insights into different groups. By analyzing clients' demographics, behaviors, and preferences, you can uncover patterns and trends that help you understand their unique desires. Then, you can make informed decisions and develop communication strategies that resonate with each segment by delivering relevant messages through preferred channels at the right time.

2. Waste of time and money – without understanding your target audience and their pain points, your company produces untargeted and ineffective marketing activities. As a result, the business wastes money and team efforts. Consumer segmentation allows to allocate your resources strategically. By identifying the most profitable client segments, you can focus your efforts on channels and tactics that resonate with them the most.

3. Customer churn – when clients are faced with irrelevant experiences, they become disengaged and dissatisfied. This drives them to seek alternatives that can meet their individualized expectations. It reduces revenue as lost customers stop buying. Segmentation lets you personalize products, services, and interactions for a more satisfying customer experience. This fosters loyalty and even helps win back 74% of former customers with tailored offers.

4. The difficulty of attracting new customers – getting a new customer isn't cheap. For example, the average cost of acquiring in ecommerce is $45-50 . Generic messages that lack personalization can easily go unnoticed or be dismissed. When your communication is relevant and personalized, it fosters a positive brand perception, positioning your business as customer-centric and attentive to their demands. It establishes trust and credibility for the brand.

Data for Segmentation

The quality and completeness of your data play a key role in the success of the customer segmentation strategy. To send the most fitting message to the right audience at the ideal moment, businesses require Big Data. Therefore, having access to accurate and up-to-date user information is essential. The CDP can make this process possible. Let's explore the data Yespo CDP can collect, manage, and utilize for segmentation:

1. Static data – any information a customer shares with you (name, gender, birthdate, email, push token, phone number, location, marital status, children, etc.).

2. Activity in campaigns with the specified details:

    a. Time (how much activity there was during a specific period).

    b. Interaction type (promo or trigger, daytime, nighttime messages, etc.).

    c. Direct channels in which the customer is active (email, mob push, web push, app inbox, in-app, widgets, SMS, messengers).

3. Purchase history – what products does the customer buy, from what category, at what frequency, what is the average check size, how much money is spent, and so on.

4. User events – all data coming into the CDP about the customer's actions on the website, in the app, in channels, or offline. To track the behavior of web visitors, you need to implement web tracking. To set up the sending of events from the app, use the Yespo SDK .

Let’s use an example of how it works in e-commerce. A person visits your website and fills out a subscription form, leaving their name, email, and phone number. Yespo creates a user profile card and automatically collects data about the subscriber's behavior on the website and in direct channels. When the customer receives a promotional email and follows a link to the page of the product of interest, the system will recognize the person even without authorization on the site. Information about this visit and online actions will automatically update the data of the unified customer profile. This ensures that the data is complete, which is necessary for high-quality segmentation.

Types of Customer Segmentation in Yespo

Based on the intensity of exploration, segmentation is divided into basic and advanced types. You can combine conditions of different types when building a group. This means setting general conditions first and then adding more narrow and specific parameters to them. You can include and exclude any number of parameters and conditions when building audience segments in Yespo. Let's take a closer look at the different types of segmentation.

1. Basic Segmentation

With basic segmentation, a marketer can split the target audience by key characteristics if they are pre-saved in the CDP. 

Demographic segmentation categorizes individuals based on average human characteristics common to large groups of people:

• pet ownership;
• job, and more.

While this type serves as an initial step in the segmentation strategy, it provides valuable insights into consumer behavior and preferences. It allows businesses to tailor their marketing efforts and offerings to specific demographic groups. For example, before International Women's Day, a marketer segments the audience into women and men. The brand offers women products relevant to them, while men are encouraged to look at women's products for gifts. Demographic segmentation helps businesses make informed decisions regarding product development, pricing strategies, and audience touchpoints.

Geographic segmentation refers to factors related to the physical location of clients. These variables provide valuable insights into customers' surroundings and enable businesses to adapt their marketing strategies accordingly. Some of the most common geographic variables applied in the customer segmentation model:

• Country – allows to address unique cultural aspects, legal requirements, and economic conditions that vary across different regions.
• City – marketers can customize their messaging to reflect the local culture, preferences, or events. This approach is particularly effective for brick-and-mortar businesses.
• Time zone – is essential to ensure that communications reach customers at appropriate times, avoiding the annoyance of untimely or intrusive messages.

Multiplex, the largest cinema chain in Ukraine, segments its customers by city of residence and favorite cinema. For example, residents of a city receive news about upcoming movie premieres or other cultural events, while fans of a particular cinema are informed about updated seats, new film formats, etc. The company's marketer conducted a test between generic and segmented emails. The test showed that for the same open rate, segmented messages had a 4% higher CTOR. It was also found that personalized email content resulted in 51.3% more sales than generic emails.

2. Advanced Segmentation

All CDPs allow you to segment contacts by rules, while more complex platforms, such as Yespo, have sophisticated architecture and extensive data processing capabilities.

Yespo omnichannel CDP provides advanced segmentation

– the platform can consistently associate events with a specific contact without the need for further processing by other systems, such as CRM. Additionally, it features built-in AI that makes predictions and identifies groups by the likelihood of potential actions.

Our clients use basic segmentation for free. Advanced segmentation is available in paid plans . Advanced segmentation allows you to drill down into more individual characteristics. The groups identified in this way are smaller, but they can be targeted with more appropriate offers that take into account the unique wishes and needs of the customers.

1. Behavioral segmentation

Behavioral segmentation allows you to divide the audience by analyzing customer interactions with a business. There are several common types of behavioral segmentation based on:

• frequency of use (active customers, mid-clients, weak consumers, non-users);
• advantages (customers sharing the same key benefits they are looking for in a product or service);
• brand loyalty (newcomers, repeat buyers, loyal customers, brand ambassadors, raving fans).

Personalized experience based on behavior is one of the main requirements driving conversion in retail. Analyzing customer actions through behavioral segmentation allows businesses to uncover meaningful signs that indicate customers' intentions. Marketers can send timely triggers and personalized messages to stimulate conversion actions, such as offering a limited-time discount or providing additional product recommendations.

This type of segmentation is crucial for the Ukrainian marketplace of promotional offers Pokupon . The content of promo emails is customized to the interests of each subscriber based on their order history. Subscribers are segmented using multiple conditions. Their activity is taken into account: the time period since the last purchase, as well as customer actions on the website, in the app, and in marketing campaigns. In addition, all contacts are divided by favorite categories: Beauty, Food and Restaurants, Health, and Entertainment. Users are included in the respective segments if they have made at least one purchase in a category. The content of each recipient's email is made up of promotions selected based on their interests. Dynamic content is added using SRT blocks . Thanks to segmentation and personalization, the company has achieved results such as:

• 49.1% growth of the website traffic from the email channel;
• 17.7% growth of the number of active customers;
• 15.6% revenue increase.

Marketers were able to get the following performance indicators in just three months.

3. RFM segmentation

RFM segmentation is based on analyzing the date, number of purchases, and total value of orders to categorize customers into distinct segments. Here are the main segments that can be identified using the RFM method and the options for triggers and campaigns for them:

• newcomers – a welcome chain of messages;
• regular buyers – promotional campaigns, emails with useful content;
• VIP – exclusive offers (referral or loyalty programs, early access to sales, and unique rewards);
• customers at risk – use additional motivators for purchase (targeted comeback offers such as limited-time discounts or free shipping);
• lost clients – those who have not bought for more than a year. It is better to leave them alone.

By utilizing this data-driven approach, businesses can effectively identify customer behavior patterns and preferences. It forecasts their likelihood of transitioning between segments. Leveraging this segmentation analysis, businesses can establish triggers to automatically send relevant campaigns to users when they move to specific segments.

Purchases and Email Engagement

Traditionally, RFM is based on purchases. However, with Yespo CDP, we have taken it a step further by offering the same analysis for email engagement. Instead of the number of sales, the matrix is built on the basis of opens and clicks, enabling companies to identify the most engaged subscribers and automate communications based on the contacts' movements through such segments.

For example, marketers at Ukrainian retailer Stylus used RFM analysis to identify inactive customers and try to reactivate them. They sent emails with personalized discounts and a survey to subscribers in the dormant segment. As a result, the company regained full communication with 5% of previously inactive subscribers.

4. Cohort Analysis Segmentation

With cohort analysis, a marketer segments the contact base into groups or cohorts. These include customers who are related to each other by a certain event that occurred at a specific moment: 

• adding a contact to the base;
• registering on the website; 
• making the first purchase;
• installing a mobile app, etc.

A marketer can identify and analyze behavioral patterns throughout the customer lifecycle. In the Yespo omnichannel CDP, segmentation by cohort helps track customer retention. This allows you to identify periods when customer engagement drops across all cohorts. If you see this pattern in consumer behavior during certain times of the year, you can plan marketing campaigns. For example, if you see that contacts' activity in channels decreases in July every year, you may want to hold off releasing new products during that month. You can also compare the performance of different cohorts. For example, compare the activity of contacts who subscribed to the newsletter during the BFCM with other cohorts. This will help you understand whether big sales are attracting long-term customers or just people who were interested in a one-time purchase with a big discount.

5. Value-based customer segmentation

Value-based customer segmentation uses key value indicators to measure buying behavior. Let's explore some examples of them:

• number of purchases – this allows to identify loyal and repeat buyers, as well as those who may require incentives to increase their purchase frequency;
• average purchase value – it enables to target customers who have a higher ability for spending and tailor marketing efforts to maximize revenue generation;
• order contents (items) – by identifying patterns and preferences, businesses can deliver personalized recommendations to drive additional sales.

Value-based segmentation helps businesses to create highly targeted marketing campaigns tailored to specific customer groups. By delivering personalized messages and offers, companies can significantly enhance customer engagement and drive conversion rates.

This segmentation approach also enables to identify valuable customers who may display signs of declining engagement or satisfaction. By addressing their concerns, improving their experience, and providing tailored solutions, companies can effectively reduce customer churn and increase customer loyalty.

Moreover, value-based segmentation allows businesses to pinpoint clients with a high potential lifetime value. These individuals demonstrate significant purchasing abilities and are more likely to establish long-term relationships with the company. By focusing resources on these customers and nurturing those relationships, businesses can optimize long-term profitability.

Make data truly profitable with advanced segmentation solutions in Yespo CDP!

6. Predictive segmentation

Predictive segmentation helps businesses to proactively understand clients' behavior and tailor their marketing strategies accordingly. Yespo provides a set of AI algorithms forecasting the likelihood of customer acquisition or churn and building corresponding contact segments. Artificial intelligence can also predict the likelihood that a customer is ready to make a purchase. In this case, our AI algorithms consider the following markers of consumer behavior:

• increased website visits;
• browsing activity;
• reviews and feedback exploration;
• chat interactions;
• discount clicks;
• average decision-making time for clients.

Unlike the above-mentioned ‘algorithmic’ segmentation strategies mentioned above, this strategy takes into account not only previous purchases or engagement in some direct channel but also a host of other factors. All the data points related to a customer, such as web pages viewed, emails opened, banners clicked, and recent events, when considered together, can provide additional insight into their current state – whether they are ready to buy or likely to churn.

Once you realize that the customer is almost ready to finalize the deal, you can encourage them to complete it with some perks, such as free shipping or a promotional code for a discount on their next order, etc. Using this AI-powered customer segmentation tool, you’ll enhance CLV rate, showing consumers you care about their needs.

For example, by implementing customized predictive segmentation, the RetouchMe app was able to increase the number of VIP users and increase revenue. Yespo's Data Science team has developed algorithms that can identify potential VIP users among newcomers with 99% accuracy . They do this quite quickly – within a week from the first customer order. The algorithms analyze various parameters: the frequency of orders, customer preferences for photo editing, whether or not discounts are applied, the frequency of requests for re-editing, etc.

Artificial intelligence not only identifies the segment of the most valuable users but also forms the group of candidates to join the customer retention program. The algorithms predict when a particular user will place their next order. Thanks to Yespo's predictive user segmentation, the number of VIP customers of the app increased by 35% in 3 months, and revenue grew by 17%.

7. Parametric segmentation

Parametric segmentation offers a time-saving solution when creating campaigns for similar groups that differ only in a specific parameter. Traditionally, announcing promotions in different categories required separate segments and workflows for each product category. However, with parametric segmentation, only one segment is needed. For example, today, an event with one condition, such as a product section, is entered into the system, and tomorrow with another. Different contacts from the group are automatically selected for the campaign according to the event parameters. At the same time, the group itself and the workflow do not need to be changed.

Parametric groups are used only in workflows. You can set conditions directly in the segment or import them from events. This tool works with group blocks of sending messages across all media channels.

This streamlined approach eliminates the need for duplicating efforts and allows for efficient and automated campaign launches, saving valuable time and resources.

To learn more

about Yespo's advanced functionality and segmentation pricing, please contact us at [email protected]

Measuring and Evaluating Results of Customer Segmentation

To understand whether your segmentation strategy is working, regularly measure the key KPIs important to your business. Watch their performance in the groups you've selected. These can be:

• Conversion rate (analyzing the number of contacts who took a conversion action, such as placing an order, using a promo code, etc.);
• Customer retention;
• Customer lifetime value;
• Engagement metrics (click-through rates, open rates);
• Customer churn rate;
• Customer satisfaction score, etc;
• Segment profitability (determining whether the profit is higher than budgeted).

By monitoring key performance indicators, businesses can gain insights into the effectiveness of their segmentation efforts. For example, tracking the conversion rate within each segment allows to know how effectively the targeted messaging and offers are driving customer actions, such as purchases, registrations, or other desired outcomes. Comparing conversion rates across different segments helps identify segments that respond most favorably to the tailored marketing efforts.

Customer segmentation is not a one-time task but an ongoing process that requires constant improvement. Based on the information gained from tracking campaign performance metrics, companies can make data-driven decisions to refine and improve their segmentation strategies.

Analyzing the performance of each segment allows to identify segments that are underperforming or have untapped potential. This analysis helps to find areas where segmentation strategies can be improved or changed.

Conducting A/B tests in specific segments allows companies to experiment with different approaches and messages to find out what resonates most deeply with each group. These experiments provide valuable insights to refine segmentation strategies and optimize customer engagement.

Continuous monitoring of campaign performance metrics and customer feedback is essential to ensure that segmentation strategies are always in line with changing customer preferences.

Conclusions

Customer segmentation is a powerful tool offering a solution to overcome non-personalized marketing, wastage of resources, customer churn, and challenges in acquiring new clients. Using the full potential of CDPs and various segmentation methods, companies can tailor their marketing efforts with accuracy and relevance.

For example, the Multiplex case study showed that emails with personalized content for a geographically segmented audience generated more sales than mass mailings. By implementing behavioral segmentation and personalization, Pokupon increased traffic to its website from the email channel and grew the number of active customers. Using event-based segmentation, an online women's apparel store doubled customer engagement and increased active customers by 20%.

Yespo's advanced segmentation based on AI and machine learning helps companies use sophisticated algorithms to gain richer insights and discover hidden patterns in customer data. This is confirmed by the case of the RetouchMe app, which showed that Yespo's AI algorithms are able to identify potential VIP customers among new users with extreme accuracy.

It is important to realize that the work doesn't end with the implementation. Measuring and evaluating the results of customer segmentation is crucial for continuous improvement. By analyzing the effectiveness of segmentation strategies, marketers can refine their approaches, adapt to changing customer needs, and increase the overall effectiveness of marketing efforts.

For more information on how our CDP can help your business, contact us at [email protected], and we will be happy to answer any questions you may have. Or book a demo and see how easy it is to work with the platform.

5.0 from 5 based on 1 reviews

Technology and Operations Management

Mba student perspectives.

• Assignments
• Assignment: RC TOM Challenge 2018

MetLife: A Case Study in Customer Segmentation

In 2015, MetLife began a year-long brand discovery process that centered around using data and machine learning to develop a more refined view of their customer segments and enable a more nuanced go to market strategy. By better understanding their customers' needs, attitudes, and behaviors, MetLife hoped to gain a competitive advantage in targeting and better serving an increasingly demanding set of customers.

In 2015, MetLife began a year-long brand discovery process that resulted in what they would later call “the most significant change to their brand in over 30 years”. [i]  At the core of this strategic refresh, was a fundamentally data driven approach, enabled by advances in machine learning, that revealed to MetLife that the insurance landscape around them was changing: Technological innovations such as the proliferation of internet connections and increased penetration of mobile devices changed the way business was done. [ii] Disruptive newcomers, such as Lemonade, were redefining the market place with their simplified approaches to underwriting. And despite that, customer and shareholder expectations were higher than ever. [iii] In the months that followed, MetLife interviewed and surveyed more than 50,000 customers and with the help of big data clustering techniques used the information to better understand and segment their customers and subsequently redesign their go to market approach. [iii] As an employee of Bain and Company, working with the MetLife team, I had the privilege to see the beginnings of the transformation firsthand.

Rethinking customer segmentation

Traditionally, insurance organizations tried to glean directional insights about their customers’ needs, attitudes, and behaviors through demographics. [iv] In the case of retail customers, age tended to be an important demographic that proxied attainment of certain life stages and thus the sophistication of the individual customer. In the case of corporate customers, the number of employees tended to be an important demographic that proxied sophistication of the organization. Armed with these types of rudimentary insights, insurers would use their best judgement in deciding the bundle of products to offer customers. However, using only demographics, insurers had at best only a rough outline of who their customers were let alone what they wanted or how to target them.

To better understand their customers, MetLife strove to “move from basic demographics and life stages to a view based on mindsets and attitudes.” [v] They collected data on their customers through in-depth surveys designed to extract a combination of demographic, firmographic, attitudinal, and need-type information. Using advanced segmentation tools, survey respondents were clustered into distinct groups based on their individual survey responses resulting in, for the first time in the company’s history, a refined picture of who their customers were. These groups (or segments) provided a new way to think about allocating resources against the pursuit of the “right” customers. Publicly available results of one such clustering (dates back to 2013 corresponding to some earlier work with segmentation), and the strategic targeting implications, are shown in the images below. [vi]

Pathways to Just Digital Future

The path forward

As part of their brand refresh, MetLife committed to a data-driven approach “focused on identifying the right customers and creating truly differentiated customer value propositions.” [iii] They committed to an $800 million net annual savings target which they expect to be at full run rate by 2020. [iii] MetLife management stated that realizing the savings would require an estimated $1 billion in investments, a significant portion of which was in technology aimed at getting better data to fuel their increasingly robust data analytics capabilities. [iii]

Further, a core aspect of the customer segmentation work that MetLife engaged in was predicated on the idea that ideal customer segments needed to be “strategic and tactical in nature.” [vii] As part of the of the customer segmentation work, members of the sales force were made aware of the customer segments and given tools to help them effectively engage with target customers.

MetLife took its segmentation practices one step further and began educating its corporate customers, encouraging them to think about their employees through a combination of demographic and psychographic data. [v] MetLife’s business offerings now include “helping HR leaders select their benefits and adjust current programs to suit their diverse employees.” [v]

In many ways, MetLife’s data-driven strategic refresh was significant moment for the company and the broader insurance industry. It applied machine learning towards sales generation when most traditional insurance companies were focused on applying machine learning solely from a risk and improved underwriting perspective.

Going forward, MetLife should continue to embed machine learning deeper within their organization. A 2017 McKinsey article outlined four broad areas where machine learning could create value for an organization: projecting (forecasting), producing (operations), promoting (sales and marketing) and providing (enhanced user experiences). [viii]

MetLife’s efforts in this strategic refresh focused on promoting. Going forward, management should be cognizant not to neglect other areas in which machine learning can add value to the organization.

Ultimately, are sequential improvements in the way MetLife uses machine learning enough to give them a competitive advantage over disruptive newcomers, or is some form of transformational improvement necessary for them to remain relevant?

(768 words)

[i] Stout, Craig. 2016. “The Power Of A Customer Centered Approach – The Metlife Rebrand”.  Brand And Marketing Consultancy | Prophet . https://www.prophet.com/2016/10/power-customer-centered-approach-metlife-rebrand/ .

[ii] OECD (2017), Technology and innovation in the insurance sector, accessed November 2018

[iii] Metlife inc corporate investor day – final. (2016, Nov 10). Fair Disclosure Wire Retrieved from http://search.proquest.com.ezp-prod1.hul.harvard.edu/docview/1842918111?accountid=11311

[iv] Carr, Mark, and Amy Modini. 2012. “A New Approach To Segmentation For The Changing Insurance Industry”.  Cmbinfo.Com . https://www.cmbinfo.com/cmb-cms/wp-content/uploads/2012/03/HealthDoc_FINAL.pdf .

[v] “Building Stronger Engagement Through Employee Segmentation | Workforce”. 2018.  Metlife.Com . https://www.metlife.com/workforce/stronger-engagement-segmentation/ .

[vi] Mehra, Sanjay, and Leah van Zelm. 2013. “Segmentation. Customer Strategy Done Right – PDF”.  Docplayer.Net . https://docplayer.net/13983641-Segmentation-customer-strategy-done-right.html .

[vii] Barlyn, Suzanne. 2017. “Metlife To Invest $1 Billion In Tech To Reach Cost-Savings Goals”.  U.S. . https://www.reuters.com/article/us-metlife-investment-technology-idUSKBN17T2R6 .

[viii] Bughin, Jacques, Eric Hazan, James Manyika, and Jonathan Woetzel. 2017. “Artificial Intelligence The Next Digital Frontier”.  Mckinsey Global Institute , 5.

Leave a comment Cancel reply

You must be logged in to post a comment.

8 Companies Mastering Customer Segmentation [+ Examples]

Updated: October 02, 2023

Published: November 22, 2022

In a product-abundant world, creating a personalized experience can help your company stand out. In fact, 80% of buyers say they prefer to purchase from brands that offer tailored experiences.

However, before you start personalizing your products or services, you need to first segment your customers.

In this piece, you'll learn about customer segmentation and see examples that can inspire you.

Table of Contents

Creating Customer Segments

Companies mastering customer segmentation, getting customer segmentation right.

Customer segmentation is the process of splitting your entire customer base into smaller segments based on shared characteristics. These can include demographics, behavioral, geographical, or psychographic data. When done correctly, customer segmentation helps you better understand users' needs and adjust your offering to satisfy them.

Type of segmentation: Demographic (date of birth)

One of the easiest and most common ways to segment your customers is by their date of birth. It gives you a great opportunity — and an excuse — to send them a personalized email without appearing pushy.

H&M is one of the brands that use this segmentation criterion. They offer birthday discounts, which are valid for a fixed time period. The 25% off is quite generous, and it would be a shame to let it go to waste.

Image source

Pro tip: You can go above and beyond by adding the customer's name to your birthday promotions. This makes your marketing seem more friendly and personal.

Type of segmentation : Demographic (income)

Understanding how much a customer is willing to spend is crucial for any business. This information is particularly useful if you offer both affordable and expensive options. If you assess your clients' income correctly, you're able to present the right offers to the right group.

A great example of this is the retail chain Argos. They offer a wide range of products, from appliances and furniture to clothing and jewelry. By looking at their customers' past purchases, they're able to estimate their budget and age.

Below is an example of a coupon Argos emailed its middle-class clients around payday.

Notice how the message acknowledges their clients' hard work and encourages them to "treat themselves." This can mean getting their kids a new toy, buying a new pair of jeans, or some new piece of home decor.

Pro tip: Creating an income-focused segment helps you show empathy. It also allows you to assess what each group will consider an attractive price. By doing so, you can increase your chances of making a sale.

Type of segmentation: Behavioral

If you're a traveler, there's a high chance you're a member of a frequent flyer program, like Flying Blue by KLM. This customer loyalty program allows you to earn miles and then spend them on tickets, extra checked luggage, shopping, or charity donations.

You can earn miles in numerous ways, including:

Flying with KLM or one of their partners.

Booking a hotel.

Renting a car.

Shopping with one of KLM's partners.

The number of miles customers get is based on the ticket price and their elite status level. Flying Blue has four status levels. The higher the level, the more points customers get.

What we love: Introducing a customer loyalty program is a great tactic to keep customers engaged and prevent them from switching to another airline. Segmenting customers by points allows the airline to reward those who fly more frequently.

Type of segmentation: Psychographic/demographic

Did you know that college students in the US have $574 billion in spending power? It's an attractive demographic. For this reason, Comcast created a special offer targeted at students specifically.

The company partnered with Amazon Music and HBO to give students access to thousands of free shows, movies, and live sports. The deal was gated, and students had to verify their eligibility before subscribing, which made the offer even more exclusive.

The deal was a huge success, and within a week Comcast saw a massive increase in web traffic and a boost in conversions.

What we love: This program segmented customers by demographic (age and education level) as well as behaviorally. College students are less likely to pay for cable but still want ways to watch their favorite shows and movies.

5. Coca-Cola

Type of segmentation: Geographical

Another customer segmentation example is splitting buyers into groups based on their location. That's the approach Coca-Cola follows. They sell their products globally. While many drinks like Coca-Cola are available in over 200 countries, some like Ciel bottled water are only sold in Mexico, Morocco, and Angola.

The brand tailors its offering to local tastes. For example, they acknowledge that consumers in Asia prefer sweeter flavors than those in the US or Europe. They modify their drink formula to satisfy different preferences.

What we love: Segmenting buyers based on geography allows the brand to better recognize cultural and climate differences, which drive customers' buying preferences. Tailoring their offer accordingly helps companies reach a wider audience and maintain a high market share.

Type of segmentation: Psychographic (state of urgency)

Nalu is a boutique women's clothing brand. Each collection features a limited number of items, which are added to the online store throughout the season in "product drops." Like many small businesses, Nalu has a relatively modest, but loyal client base. There's one customer segment that's particularly enthusiastic about new products, i.e., newsletter subscribers.

Nalu knows how to make this group feel appreciated, all the while creating a sense of urgency. Newsletter subscribers receive an exclusive link, which unlocks new items 24 hours ahead of their launch in the store. This way, clients can buy clothing in their size before it runs out of stock. They use the same approach for their archive collection sales.

What we love: Nalu is a great example of how you can use customer segmentation to strengthen the bond with your high-value clients. Not to mention, this tactic offers a great way to boost revenue from clearing older items.

7. Duolingo

Customer segmentation : Behavioral

For the language app Duolingo, retaining users comes down to keeping them motivated as they memorize new words and phrases. They realize that most language learners eventually lose their initial drive and need a pick-me-up to continue studying. That's why Duolingo has decided to create user segments based on in-app behavior and milestone achievements.

Those who complete lessons receive badges and rewards. These can be exchanged for extra assignments. Meanwhile, those who've skipped a lesson (or two, or ten) start receiving motivational reminders.

Pro tip: Tailoring your in-app notifications and email sequences for each customer segment is a great way of preventing app abandonment. Active users feel appreciated, while those who lose motivation are brought back on track.

Customer segmentation type : Demographic/psychographic

L'Oreal is one of the most recognizable beauty brands in the world. Many of their clients have been loyal to the brand for decades. Perhaps unsurprisingly, they are very popular among older clients.

The average L'Oreal customer is around 50 years old in the Netherlands. That being said, the company noticed that they can't resonate with younger clients well. They've failed to do so by simply breaking down their customer base into age segments.

To tackle this, L'Oreal joined forces with Google and McCann. They wanted to learn how they can create more accurate segmentation criteria. The brand broke down Gen-Z's and Millennials into not just age, but also behavior-focused batches.

By using Google's audience segments, they created twelve ad variations. Each is for a different group of potential buyers. One of the video ads they've created targeted young music lovers. It featured a product for acne-prone skin, and a tagline "99 problems, and your skin is one?" The ad references one of the most famous pieces in hip-hop history that will be familiar to Millennials.

What we love: Digging deep into behavioral data helps L'Oreal restore (and maintain) relevance among all customer groups. It's also a great example of how you can supplement demographics with more sophisticated user contexts.

Customer segmentation is a powerful tool. One that puts personalization and unique client needs front and center. As you've seen in the customer segmentation examples above, there are plenty of criteria you can use to strengthen your bond with clients.

Be it age, location, budget, personal beliefs, or on-site behavior — the choice is yours. Good luck!

Don't forget to share this post!

Related articles.

Can AI Segment Your Customers? I Ran This Experiment to Find Out

Customer Segmentation: How to Segment Users & Clients Effectively

Share of Wallet: What It Is and How to Calculate It

Geographic Segmentation: The Complete Guide

What Is a Customer Profitability Analysis? How to Make Your Business More Efficient

What Is Customer Communication Management? (Plus 5 Tools to Adopt)

The Complete Guide to Customer Micro-Segmentation

8 Customer Characteristics Every Service Rep Should Know

The Expert's Guide to Contextual Inquiry Interviews

Outline your company's customer journey and experience with these 7 free customer journey map templates.

Service Hub provides everything you need to delight and retain customers while supporting the success of your whole front office

6 customer segmentation case studies show big results

Customer Segmentation

Customer segmentation is the practice of putting customers into groups that are similar in specific ways. The goal is to tailor marketing to best meet individual needs. As a result, it improves customer satisfaction with customer satisfaction surveys , revenue, and profitability.

There are four types of customer segmentation: Demographic, Psychographic Geographic, and Behavioral. With the rise of machine learning , artificial in t elligence , personalization , and split testing , changes in customer segmentation can occur quicker with a significant impact.

Here are 6 customer segmentation case studies that show big results.

Airbnb uses  machine learning  to gain insights from user reviews. Then, they use this behavioral data and preference to pair hosts and guests. With A/B testing, they discover how website changes affect consumer behavior. They are able to adjust and personalize the content that users see when browsing the website. 

BabyCenter (Johnson & Johnson)

For BabyCenter , Johnson & Johnson uses a Facebook Messenger App to suggest personalized advice. Through a series of questions and answers, they make targeted recommendations based on the input it receives from the user,

They look at the data to see what drove the highest levels of traffic to the website – a chatbot, email marketing, or the app. What they find is the messenger app has a read rate of 84% and a click-through rate (CTR) of 53%. The app’s overall engagement rate is 1,428% higher than email. Because it offers the greatest personalization.

DavidsTea customer segmentation

DavidsTea uses email marketing to recognize customer loyalty. When a customer reaches a specific anniversary with the company, they receive a “look back” email. It contains data on their first purchase, their most purchased teas, and how much they bought.

Therefore, by receiving this email, the customer feels unique and valued throughout their customer journey and is inclined to continue purchasing.

The Lego Group faces the challenge of marketing Lego Bricks on social media. The company identifies six distinct personas based on purchase and usage:

• Lead Users—people LEGO actively engages with on product design
• 1:1 Community—people whose names and addresses they know
• Connected Community—people who have bought LEGO and have also been to either a LEGO shop or a LEGO park
• Active Households—people who have bought LEGO in the last 12 months
• Covered Households—people who have bought LEGO once
• All Households—those who have never bought LEGO

The first three personas represent the most fertile ground. Because they share a deeper involvement with the brand. From there, Lego builds online communities on the social networks these segments use most often.

Lego takes advantage of their most valuable asset, their fans, who post pictures, videos, and provide new product ideas. This effort helps them to increase to the world’s fourth-largest toy manufacturer.

Netflix uses personalization that begins as soon as a user creates an account with Netflix and streams even just one TV show or movie. They use an  algorithm  that allows them to consistently and accurately A/B test and experiment with viewer preferences. Netflix’s algorithm dictates everything – the homepage layout, the recommended content, and even the visuals, or landing cards, for each piece of cinema. What’s more, Netflix personalizes the image  you see based on the actors, actresses, or genres that it thinks you like.  Netflix’s recommendation system saves them a massive  $1Billion per year .

Olay customer segmentation

Olay creates Skin Advisor. The  artificial intelligence beauty tool  collects data from customers by asking them five to seven quick questions about their skin. The advisor then reveals the true age of the customer’s skin, and recommend products.

The data shows many customers are seeking Retinol based products. However, the subsequent lack of Retinol products in its range is contributing to the brand losing customers. Therefore, Olay releases Retinol 24 which has gone on to be one of the brand’s best selling products and has helped transform their sales. 

Do these case studies help you see the impact of effective customer segmentation? And the tools you can apply, today ?

Rob Petersen

Related posts.

8 distinguished data science case studies

9 reasons why we love a good case study

15 inspiring case studies of pivoting

10 lead nurturing case studies show the ROI

Leave a reply cancel reply.

Your email address will not be published. Required fields are marked *

Save my name, email, and website in this browser for the next time I comment.

Back to blog

Case study: how segmentation improved a company's customer retention.

November 17, 2023 | Jimit Mehta

Have you ever noticed how some companies seem to have an almost magical ability to keep their customers coming back for more? It's as if they have an uncanny understanding of their customers' needs and preferences, and are able to provide them with exactly what they want, time and time again. Well, it's not magic - it's good old-fashioned customer segmentation. In this case study, we'll take a closer look at how one company was able to significantly improve their customer retention rates by using targeted segmentation strategies. By tailoring their products, messaging, and customer service to the specific needs and preferences of different customer groups, this company was able to create a loyal following that kept coming back for more. So, if you're looking for ways to improve your own customer retention rates, read on to learn from their success.

Understanding Customer Segmentation

Understanding customer segmentation is a crucial first step in any successful marketing strategy. It involves breaking down your customer base into distinct groups based on shared characteristics such as demographics, behavior, or needs. By doing this, you can gain a deeper understanding of your customers and what motivates them to buy your products or services.

For example, you might segment your customers based on age, income level, geographic location, or previous purchase history. This information can then be used to tailor your marketing efforts to each specific segment, such as creating targeted advertisements or customized email campaigns.

By understanding customer segmentation, you can better meet the unique needs and preferences of different customer groups. This can lead to increased customer satisfaction, brand loyalty, and ultimately, improved customer retention. Without a clear understanding of customer segmentation, it can be difficult to create effective marketing strategies that resonate with your target audience.

Identifying Different Customer Groups

Identifying different customer groups is a critical step in any customer segmentation strategy. This involves identifying the different types of customers who interact with your business and grouping them together based on similar traits or behaviors. By doing so, you can gain a deeper understanding of each group's unique needs and preferences, allowing you to tailor your marketing efforts to each group.

To identify customer groups, you can start by analyzing your customer data to look for patterns and trends. For example, you might find that one group of customers tends to purchase higher-end products, while another group prefers to buy more affordable options. Or, you might discover that one group of customers is more interested in certain features or benefits of your products, while another group is more concerned with price.

Once you've identified your customer groups, you can start to create targeted marketing messages and product offerings for each group. By doing so, you can better meet the unique needs of each group, ultimately leading to increased customer satisfaction and retention. In short, identifying different customer groups is a crucial first step in creating effective customer segmentation strategies that can help your business succeed.

Tailoring Products to Specific Customer Needs

Tailoring products to specific customer needs is a key component of any successful customer segmentation strategy. This involves analyzing the unique needs and preferences of each customer group and adjusting your product offerings to meet those needs.

For example, if you have identified a customer group that is particularly interested in environmentally-friendly products, you might create a line of eco-friendly products that specifically target that group. Alternatively, if you have a customer group that prefers more affordable options, you might create a lower-priced line of products that meet their needs.

By tailoring your products to specific customer needs, you can create a deeper connection with your customers and build brand loyalty. When customers feel that your products are specifically designed for them, they are more likely to choose your brand over competitors.

Of course, tailoring products to specific customer needs is not always easy, and it may require significant investment in research and development. However, the benefits of meeting the unique needs of each customer group can be significant, including increased customer satisfaction, higher retention rates, and ultimately, improved business success.

Crafting Messaging for Different Customer Segments

Crafting messaging for different customer segments is a key component of effective customer segmentation. It involves developing customized marketing messages that speak directly to the unique needs and preferences of each customer group.

For example, if you have identified a customer group that is particularly interested in high-quality, luxury products, you might craft messaging that emphasizes the exclusivity and sophistication of your brand. On the other hand, if you have a customer group that is more price-sensitive, you might emphasize the affordability and value of your products.

When crafting messaging for different customer segments, it's important to keep in mind that what works for one group may not work for another. The messaging that resonates with a younger demographic might not work for an older one, and vice versa. That's why it's crucial to have a deep understanding of each customer group's unique needs and preferences, so that you can create messaging that speaks directly to them.

By crafting messaging that is tailored to each customer group, you can create a more personalized and engaging experience for your customers. This can help to build brand loyalty and increase customer retention, ultimately leading to improved business success.

Personalizing Customer Service for Different Segments

Personalizing customer service for different segments is an important aspect of customer segmentation. It involves tailoring your customer service experience to meet the unique needs of each customer group.

For example, if you have a customer group that prefers to communicate via email , you might ensure that you have a dedicated customer service email address and that your representatives are trained to respond quickly and efficiently to email inquiries. On the other hand, if you have a customer group that prefers to communicate over the phone, you might offer a dedicated customer service phone line with representatives who are trained to provide personalized assistance over the phone.

When personalizing customer service for different segments, it's important to keep in mind the specific needs and preferences of each group. For example, a customer group that is primarily composed of elderly customers might require a different type of customer service experience than one that is primarily composed of younger customers.

By personalizing your customer service experience for each customer segment, you can create a more positive and engaging experience for your customers. This can help to build brand loyalty and increase customer retention, ultimately leading to improved business success.

Analyzing the Impact of Segmentation on Retention

Analyzing the impact of segmentation on retention is a critical step in any customer segmentation strategy. It involves looking at how customer segmentation has affected customer retention rates and identifying which strategies have been the most effective in improving retention.

One way to analyze the impact of segmentation on retention is to look at retention rates before and after implementing a segmentation strategy. If you have seen an improvement in retention rates after implementing a segmentation strategy, you can look at which specific strategies were the most effective in achieving that improvement.

For example, you might find that tailoring products to specific customer needs had a significant impact on retention rates for a particular customer group. Alternatively, you might find that personalizing customer service for different segments led to increased customer satisfaction and retention.

Analyzing the impact of segmentation on retention is crucial because it allows you to refine and improve your segmentation strategy over time. By identifying which strategies are the most effective in improving retention, you can focus your resources on those strategies and continue to refine them to further improve retention rates.

In short, analyzing the impact of segmentation on retention is an important part of any customer segmentation strategy, as it allows you to identify which strategies are the most effective in improving retention rates and ultimately driving business success.

Measuring the Success of Segmentation Strategies

Measuring the success of segmentation strategies is an important part of any customer segmentation plan. It involves tracking and analyzing the results of your segmentation efforts to determine how well they are achieving your business goals.

There are a few key metrics that you can use to measure the success of your segmentation strategies. One is customer retention rate, which can help you to determine if your segmentation efforts are helping to keep customers engaged and loyal to your brand.

Another important metric is CLV, which looks at how much value each customer brings to your business over the course of their lifetime. By analyzing CLV for different customer segments, you can determine which segments are the most valuable to your business and adjust your segmentation strategies accordingly.

In addition to retention rate and CLV, other metrics that can be used to measure the success of segmentation strategies include customer acquisition cost (CAC), customer satisfaction, and revenue growth.

By measuring the success of your segmentation strategies, you can determine which strategies are the most effective in achieving your business goals and adjust your segmentation efforts accordingly. This can help to improve customer retention, increase revenue, and drive overall business success.

In summary, measuring the success of segmentation strategies is a crucial component of any customer segmentation plan, as it allows you to track the impact of your efforts and adjust your strategies to achieve your business goals.

Identifying Opportunities for Further Segmentation

Identifying opportunities for further segmentation is an important part of any customer segmentation strategy. It involves looking for new ways to divide your customer base into smaller, more targeted segments that can be marketed to more effectively.

One way to identify opportunities for further segmentation is to conduct customer research to understand their needs, preferences, and behaviors. This can help you to identify commonalities among customers that you may not have noticed before, which can be used to create new customer segments.

Another way to identify opportunities for further segmentation is to analyze customer data to identify patterns and trends. This can include looking at purchasing history, demographics, and other factors that can help you to create more targeted segments.

When identifying opportunities for further segmentation, it's important to keep in mind the potential benefits and costs of creating new segments. While creating more targeted segments can help to improve marketing effectiveness and customer retention, it can also require additional resources and time.

Ultimately, identifying opportunities for further segmentation is a continuous process that should be revisited regularly. By continuously seeking out new ways to divide your customer base into smaller, more targeted segments, you can help to improve the effectiveness of your marketing efforts, enhance customer satisfaction, and drive overall business success.

Avoiding Common Segmentation Pitfalls

Customer segmentation can be an effective strategy for improving customer retention and driving business growth, but it's not without its challenges. There are several common segmentation pitfalls that businesses should be aware of and take steps to avoid.

One common pitfall is over-segmentation, where a company creates too many segments that are too small to be effectively marketed to. This can result in increased costs and decreased marketing effectiveness. To avoid over-segmentation, it's important to focus on creating segments that are large enough to be targeted effectively.

Another common pitfall is under-segmentation, where a company creates broad segments that don't effectively capture the unique needs and preferences of its customers. To avoid under-segmentation, it's important to conduct customer research and analyze data to create segments that are targeted and specific.

A third common pitfall is failing to update segmentation strategies over time. As customer needs and behaviors change, it's important to adjust segmentation strategies to keep pace. Failure to do so can result in decreased effectiveness and missed opportunities.

Finally, a common pitfall is treating segments as static and homogeneous, when in reality they are dynamic and diverse. To avoid this pitfall, it's important to constantly monitor and analyze segments to identify changes and adjust strategies accordingly.

By being aware of these common segmentation pitfalls and taking steps to avoid them, businesses can create effective segmentation strategies that improve customer retention, drive growth, and ultimately lead to greater business success.

Conclusion: Key Takeaways from the Case Study

In conclusion, the case study highlights the importance of customer segmentation in improving customer retention and driving business growth. The company in the case study was able to successfully implement a segmentation strategy by identifying different customer groups, tailoring products and messaging to specific segments, personalizing customer service, and measuring the impact of segmentation on retention.

The key takeaways from this case study include the importance of conducting customer research and analyzing data to create effective segmentation strategies, and the need to continuously update and adjust segmentation strategies as customer needs and behaviors change.

The case study also highlights the benefits of segmentation, including improved marketing effectiveness, increased customer satisfaction, and greater customer retention. By creating targeted segments that address the unique needs and preferences of different customer groups, businesses can drive growth and build a more loyal customer base.

Overall, the case study underscores the importance of customer segmentation in today's competitive business environment. By using segmentation to create more personalized and effective marketing strategies, businesses can gain a competitive edge and achieve greater success.

Wrapping up

In today's highly competitive business environment, customer retention is critical for business growth and success. A case study has highlighted the importance of customer segmentation in improving customer retention and driving growth. The company in the case study was able to implement an effective segmentation strategy by identifying different customer groups, tailoring products and messaging to specific segments, personalizing customer service, and measuring the impact of segmentation on retention.

The case study also underlines the need to continuously update and adjust segmentation strategies as customer needs and behaviors change. By creating targeted segments that address the unique needs and preferences of different customer groups, businesses can improve marketing effectiveness, increase customer satisfaction, and drive growth.

The article explains the importance of understanding customer segmentation, identifying different customer groups, tailoring products to specific customer needs, crafting messaging for different customer segments, personalizing customer service, analyzing the impact of segmentation on retention, and measuring the success of segmentation strategies. It also highlights common segmentation pitfalls that businesses should be aware of and take steps to avoid.

Overall, the case study demonstrates that customer segmentation is a powerful tool for improving customer retention and driving growth. By creating more personalized and effective marketing strategies, businesses can gain a competitive edge and achieve greater success.

Want show different customer segments more relevant content using personalization?  Try Markettailor for free.

Related posts

Case study: how customer segmentation helped a company improve customer satisfaction.

Imagine walking into a store and being greeted with a personalized shopping experience tailored just for you. The sales representatives know your name, your preferences, and your buying habits. They suggest products that match your interests, and you leave feeling satisfied with your purchases....

Case study: how customer segmentation helped a company increase sales

Have you ever wondered how companies manage to tailor their products and services to meet the specific needs of their customers? The answer lies in customer segmentation, a marketing strategy that involves dividing a large customer base into smaller groups based on common characteristics.

How can algorithms help in segmenting users and customers? A systematic review and research agenda for algorithmic customer segmentation

• Original Article
• Open access
• Published: 06 July 2023
• Volume 11 , pages 677–692, ( 2023 )

Cite this article

You have full access to this open access article

• Joni Salminen   ORCID: orcid.org/0000-0003-3230-0561 1 ,
• Mekhail Mustak 2 ,
• Muhammad Sufyan 3 &
• Bernard J. Jansen 4  

4179 Accesses

Explore all metrics

What algorithm to choose for customer segmentation? Should you use one algorithm or many? How many customer segments should you create? How to evaluate the results? In this research, we carry out a systematic literature review to address such central questions in customer segmentation research and practice. The results from extracting information from 172 relevant articles show that algorithmic customer segmentation is the predominant approach for customer segmentation. We found researchers employing 46 different algorithms and 14 different evaluation metrics. For the algorithms, K-means clustering is the most employed. For the metrics, separation-focused metrics are slightly more prevalent than statistics-focused metrics. However, extant studies rarely use domain experts in evaluating the outcomes. Out of the 169 studies that provided details about hyperparameters, more than four out of five used segment size as their only hyperparameter. Typically, studies generate four segments, although the maximum number rarely exceeds twenty, and in most cases, is less than ten. Based on these findings, we propose seven key goals and three practical implications to enhance customer segmentation research and application.

Similar content being viewed by others

Knowing Your Customers Using Customer Segmentation

A review on customer segmentation methods for personalized customer targeting in e-commerce use cases

Customer Segmentation Analysis Using Clustering Algorithms

Avoid common mistakes on your manuscript.

Introduction

Business success depends on understanding customers and their needs. A key method to achieve this is customer segmentation , i.e., dividing individual customers into groups based on their similarities and differences (Cooil et al. 2008 ). As postulated by Punj and Stewart ( 1983 : 135), “All segmentation research, regardless of the method used, is designed to identify groups of entities (people, markets, organizations) that share certain common characteristics (attitudes, purchase propensities, media habits, etc.)”. Customer segmentation, in particular, allows businesses to create targeted marketing strategies, improve customer experience, and ultimately, increase revenue (Hosseini and Shabani 2015 ; Simões and Nogueira 2022 ; Spoor 2022 ). As the potential benefits are plentiful, most firms perform customer segmentation across industries, irrespective of their size (Zhou et al. 2021 ). With the rise of artificial intelligence (AI) and machine learning (ML) technologies (Mustak et al. 2021 ), firms are increasingly turning to more advanced AI and ML algorithms (we refer to these as AI/ML algorithms henceforth; for the reader interested in the conceptual distinction of these two terms, we recommend reading Kühl et al. 2022 ) to perform customer segmentation. To this end, customer segmentation undoubtedly represents a cornerstone application of AI for business purposes, as it represents an unsupervised learning problem where AI/ML algorithms are known to be applicable (Joung and Kim 2023 ; Ranjan and Srivastava 2022 ). This relationship is illustrated in Fig.  1 .

The hierarchy of key concepts in this study. AI is a general concept referring to (pseudo-)intelligent algorithms performing tasks that require intelligence. Machine learning is an application of AI. Unsupervised learning is a type of machine learning, of which clustering is the most common approach. Clustering, when applied to a customer dataset, then becomes the customer segmentation task

However, at the same time, firms struggle to understand these novel AI/ML methods and their implementation in everyday customer segmentation—for example, what algorithm to choose for their given data? Should they use one algorithm or many? How many customer segments should be created? How to evaluate the results? These motivational questions reveal a practical research gap for customer segmentation research. Simultaneously, extant academic research lacks a synthesis of how customer segmentation is carried out in research studies, in terms of methods, parameters, evaluation, and so on—in other words, there is a theoretical research gap. Taken together, these two gaps hinder the development of the body of knowledge around the practice and theory of customer segmentation, especially in the light of novel AI technologies (the most cited article assessing the use of clustering in marketing, by Punj and Stewart, is from 1983, so there is a need for an updated study). It becomes harder for firms and researchers to develop better approaches, evaluate them, and implement them if we do not know adequately well how customer segmentation has been done in the past.

We address this knowledge gap by exploring the various customer segmentation methods used in business practices and then delve into the various AI/ML algorithms used for customer segmentation. Therefore, this research emphasizes algorithmic approaches to study what we term algorithmic customer segmentation (ACS). The central premise of our treatise is that AI/ML methods have become increasingly commonplace in marketing, specifically in customer segmentation. However, the extant academic literature lacks a holistic view of how customer ACS is done . Offering such a view is valuable for scholars (in terms of identifying patterns and gaps) and practitioners, as lessons from previous work are likely to offer a useful starting point for robust customer segmentation while helping to identify novel angles for future work in this domain.

Overall, this study aims to increase knowledge of customer segmentation research, and, more explicitly, the role of algorithms and AI in customer segmentation, toward understanding the theory and practice of ACS. In addition, based on our analysis of the extant literature, we offer a detailed and comprehensive agenda for future academic inquiry. To this end, we formulate specific research questions (RQs) that are addressed by a systematic literature review (SLR):

RQ1: What algorithms are typically used for customer segmentation? To understand ACS, we must familiarize ourselves with the different algorithms applied for the customer segmentation task in scholarly literature.

RQ2: Does customer segmentation typically use one or many algorithms? When many, what algorithms emerge in combination? The interaction among different algorithms poses an impactful question, as the field of customer segmentation migrates toward interactive systems that enable stakeholders to interact with the customer segments directly—thus, one algorithm might not be sufficient for more advanced systems (Jung et al. 2018 ).

RQ3: How many customer segments are typically created? There could be two, three, or ten segments—but it is unclear what the optimal number of customer segments should be. Furthermore, this number will likely vary by the dataset in question (Salminen et al. 2022 ). The current body of knowledge lacks systematic insights into how many customer segments are created in research, so we investigate this RQ.

RQ4: How is customer segmentation typically evaluated? ‘Evaluation’ means determining the quality of an algorithmic process; so, how well the customer segmentation worked (Thirumuruganathan et al. 2023 ). This typically measures the segmentation algorithm’s ability to clearly distinguish one group of customers from another group. As evaluation is an essential part of the customer segmentation process, we investigate this RQ.

RQ5: What hyperparameters are used in algorithmic customer segmentation? A ‘hyperparameter’ refers to a parameter external to the model (Jansen et al. 2021 ). The human (or an automated script) selects the “optimal” hyperparameters based on manual viewing or some technical measure. The most central hyperparameter for customer segmentation is the number of segments created (i.e., the segment size). However, there could be other relevant hyperparameters, which is why we are investigating this RQ.

RQ6: How frequently are subject matter experts used for evaluating the customer segments? The quality of customer segmentation can be ascertained using both automatic and manual means. ‘Automatic’ implies that the decision-making relies on technical metrics. However, at least of equal importance is exposing the customer segments to human decision-makers, e.g., managers, software developers, designers, and others that rely on customer segmentation as inputs in their decision-making. To this end, it is worthwhile to investigate how these stakeholders are involved in the customer segmentation process.

Methodology

We adopt the SLR methodology to discover and analyze pertinent existing studies. The SLR process provides a precise and reliable appraisal of the topic under examination, acknowledges existing flaws, and is less biased than typical judgment-based evaluations done by professionals in a specific field. In our approach, we adhered to the review procedure described by Kitchenham et al. ( 2009 ) and conducted the review in three sequential stages: (a) literature search, (b) assessment of the evidence base, and (c) analysis and synthesis of the findings.

Literature search

In this study, we have included literature from various business domains, for instance, marketing and management. Further, as a high degree of relevant knowledge is available in the field of computer science, it is worthwhile to pursue knowledge developed there while concurrently examining its marketing implications (Mustak et al. 2021 ; Salminen et al. 2019 ). Predominantly, we focus on customer segmentation, not market segmentation. Although these two concepts appear similar, they are not: Market segmentation is the process of dividing a market into different subgroups of consumers with similar needs, wants, behavior, or other characteristics. In turn, customer segmentation is the process of dividing customers into groups based on their qualities, attributes, and behaviors. So, the former deals with the overall market and the latter with the specific, current customer base.

To identify relevant literature, we used four prominent academic databases: Web of Science (WoS), Emerald Insight, ACM Digital Library, and ABI/INFORM Collection (ProQuest). WoS is the most comprehensive generic database, encompassing more than 12,000 high-impact journals and research articles from more than 3,300 publishers. The Emerald Insight and ABI/INFORM Collection (ProQuest) are also similar—as generic databases, they offer a collection of many relevant journals and scientific articles. The ACM Digital Library is a specialized database focusing on technical disciplines and thus helps uncover the articles focused on technical aspects of customer segmentation. Combined, these four databases offer balanced coverage of the existing literature on customer segmentation from multiple scientific disciplines.

We conducted detailed searches in each of the four databases. We did not want to pre-limit the searches with highly specific keywords and narrow terms, which may exclude crucial articles, as those articles may address the very same topic but use different terms. Rather, to identify a wide range of publications that may shed light on customer segmentation, we used only the following keywords: “customer segmentation*”, “user segmentation*”, and “audience segmentation*” (‘*’ denotes plural forms) and identified all the associated articles. However, in the case of WoS, after the initial search, which resulted in 574 entries, we kept articles from the fields of business and computer science (as categorized by WoS) and excluded articles from other fields that we deemed outside the scope of this study. These fields included, for example, materials science, physics, chemistry, or integrative complementary medicine. The specific search details, along with the assessment of the evidence base, are presented in Table 1 .

Assessment of the evidence base

We carried out the article screening in two stages. In the first stage, we focused on “hygiene factors” (see Table 2 ), such as removing duplicates, articles with missing information, those written in a language other than English, and articles published before the year 2000, as we considered the past 20 years or so to contain most relevant work for our research purpose. Figure  2 illustrates the general interest in segmentation among studies in or related to marketing.

Approximate search results when searching Google Scholar for [+ segmentation + marketing]. The results indicate a general increase in interest in segmentation among studies in or related to marketing. The vertical line (the year 2000) indicates our sample’s cut-off year

After the first round of screening, we checked the remaining 204 articles based on their contents, i.e., assessing their relevance to our research purpose. We found that 32 (15.7%) articles were irrelevant to our research purpose (e.g., they were literature reviews or did not use empirical data to conduct customer segmentation). This reduction left us with 172 (84.3%) articles, 134 (77.9%) of which contained algorithm-based approaches to customer segmentation (i.e., representing ACS), while 38 (22.1%) articles applied non-algorithm-based customer segmentation (see Fig.  3 ). In our analysis, for RQ1-2 and RQ4-RQ5, we focus on the 134 articles that were algorithm-based. For RQ3 and RQ6, we focus on the full 172 articles that passed the pre-screening as these RQs do not require the articles to be algorithm-based.

Research process leading to article coding. Essentially, this took place in four stages: Screening, Relevance Assessment, Algorithm Assessment, and Data Extraction

Article coding

After the screening, we extracted information from the articles to address our RQs using a data extraction sheet. The information fields were designed to correspond with the RQs (see Table 3 ). The coding was carried out by one researcher, with another researcher verifying the quality of the coding outcomes by randomly investigating a sample of 20 articles. This inspection was carried out successfully and revealed that the data was coded following the guidelines in Table 3 . The results (see Section " RQ2: does customer segmentation typically use one or many algorithms? ") were obtained by carefully reviewing the full-text articles.

RQ1: What algorithms are typically used for customer segmentation?

After a thorough overview of the ACS literature, we identified 46 different algorithms used for customer segmentation , whose usage frequencies are illustrated in Fig.  4 .

Frequency chart for the algorithms used. The y -axis represents the name of the algorithms used for customer segmentation, and the x-axis shows the frequency, i.e., the number of times each algorithm has been used in the literature for customer segmentation. The ‘Other’ class contains all the algorithms that were used only once

As can be seen in Fig.  4 , K-means clustering is the most frequently used algorithm, as it is used 27 times (20.1%) in our sample of reviewed literature. Other prominent algorithms include variants of K-means clustering with a frequency of 10 (7.5%), fuzzy algorithms, and latent class analysis models are all used 8 (6.0%) times, respectively. The recency-frequency-monetary gain (RFM) and its variants have been used 6 (4.5%) times, while Self-Organizing Maps (SOM) and Genetic Algorithms (GA) have been used three times (2.25%) each. Other algorithms, for example, the Louvain algorithm, Ward’s algorithm, and hierarchical clustering algorithms have been used two to three times varyingly. Furthermore, some algorithms have only been used once, indicating that these approaches may not have been thoroughly explored in the context of customer segmentation. Examples include direct grouping iterative merge and consistency-based clustering algorithms, suggesting that there is scope for further nuanced research in these areas.

Overall, the results highlight the wide range of algorithms available for customer segmentation and the need for further exploration and comparison of these methods to determine the most effective approach for different business scenarios.

RQ2: Does customer segmentation typically use one or many algorithms?

After a thorough overview of the ACS literature, it appears that in most cases, researchers have utilized one algorithm for customer segmentation (i.e., in roughly 80% of them). However, there are also instances where multiple algorithms have been combined for more effective results. For example, K-means clustering, SOM, and RFM approaches have been applied in combination with other approaches (see Table 4 ).

In general, the employment of multiple algorithms in combination may aid in addressing the shortcomings specific to each algorithm and may also help in the creation of more robust and distinct customer segments. Furthermore, technical reasons hailing from the domain of applied ML can explain the use of multiple algorithms. It is expected that AI/ML research studies compare and evaluate multiple algorithms for one task; this involves experimenting with combinations of different algorithms and calculating the accuracy/performance among them. This is often done by conducting an ablation study (Symeonidis et al. 2018 ). An ablation study is a form of experimental design used to study the effect of removing a specific part or feature of a model on its overall performance, typically used in the field of ML and AI. It can involve removing individual components of a larger model or system, a subset of data features, or hyperparameters to see how they affect the accuracy, cost, and other metrics of the ML model. Similarly, the researchers developing new customer segmentation algorithms are required to demonstrate the value of their approach; for this reason, algorithms are often compared against one another.

RQ3: How many customer segments are typically created?

Our analyses show that various numbers of customer segments have been suggested/created by researchers based on the application area and/or target market and objectives of the research (see Fig.  5 ). Most commonly, researchers have suggested/created four customer segments ( n  = 32, 21.2%). On average, researchers created 5.7 segments (SD = 3.9). Interestingly, no study produced more than 20 customer segments (apart from the outlier study mentioned in Fig.  1 caption), while the lowest number of segments produced was one (in this particular study, the researchers applied a decision-rule algorithm to discover the most ideal customer type (Lee, J. H. & Park, 2005)). Interestingly, more than half ( n  = 103, 68.2%) of the studies generated between two and five segments (see Fig.  5 ). Furthermore, 92.1% of the studies generated ten or fewer segments.

Number of customer segments created in research articles. The numbers are based on 151 articles (87.7% of the total 172) that expressed the number of segments created. In the case the researchers presented multiple segmentations with different numbers (e.g., 7, 7, 8), as was sometimes the case when experimenting with multiple algorithms or datasets, we have taken the average of the reported numbers and rounded it either up or down based on standard rounding rules (e.g., 7, 7, 8 would yield the average of 7.3 which rounds to the segment size of 7). Twenty-one articles (12.2%) did not report the segment size. Also, we omitted one outlier article from the analysis, as the researchers created 1209 and 8984 segments from two datasets, respectively (Böttcher et al. 2009 )

Typically, the number of segments is not determined beforehand but in a data-driven way (Hiziroglu 2013 ; Hong and Kim 2012 ), which means determining the segment size based on quantitative evaluation metrics (i.e., the number of segments is such that best fits the data based on an evaluation metric when experimentally varying the segment size). For example, the researchers may attempt multiple numbers for k where k indicates the number of segments, and then visualize at what number of k the obtained information noticeably decreases—the so-called elbow method where elbow indicates this decrease (Syakur et al. 2018 ). So, the number of clusters is neither random nor determined a priori, but all the reviewed articles used some quantitative metrics/criteria to validate or choose the number of segments.

RQ4: How is customer segmentation typically evaluated?

In ACS, the evaluation of the customer segments is crucial for the effectiveness of the segmentation process. As ACS tends to be an unsupervised ML task (where there is no single “correct” value for a segment, but instead, the algorithm aims to organize the data into groups, i.e., segments), accuracy is used more rarely in ACS than in other types of ML tasks, including supervised learning tasks such as prediction. Hence, there are several measures and criteria that researchers have used to evaluate the quality of customer segments. Overall, we identified 14 different metrics for evaluating customer segmentation outputs, of which six (42.9%) focused on statistical indicators and eight (57.1%) focused on distances and/or similarity calculation (see Table 5 ). We discuss the metrics in alphabetical order.

Accuracy (ACC) is used to evaluate the quality of segments, i.e., how well “unseen” or new members (customers) the segmentation algorithm can allocate to the correct segment (Wu, S. et al., 2021). ACC is typically calculated by dividing the number of correct values by the total number of predicted values. The use of ACC requires that there are ‘ground truth’ values or labels against which the predictive ability of an algorithm is compared.

The Adjusted Rand Index (ARI) measures the similarity between two segmentations (Xu, X. et al., 2007). The ARI considers the random chance that objects between the segmentations might be similar. It compares object pairs in two sets of segments and computes the difference between the observed agreement and the expected agreement under random labeling. The value of -1 indicates perfect dissimilarity, and + 1 indicates perfect similarity. So, the higher the ARI, the more similar the two customer segmentations are.

The Analysis of Variance (ANOVA) test is a statistical method used to test the difference in means between two or more groups. In customer segmentation, ANOVA can test for significant numerical differences in customer characteristics between the segments (Ballestar et al. 2018 ; Kashwan and Velu 2013 ).

The Average Clustering Error (ACE) evaluates the average distance between data points within each segment (Manjunath and Kashef 2021 ). The ACE is calculated by taking the average of the sum of the distances from each point in the dataset to its closest segment center. It indicates how well the segmentation algorithm has been able to group points with similar characteristics. If the ACE is low, this indicates an effective segmentation algorithm.

The Bayesian Information Criterion (BIC) is a measure used for model selection and comparison, including in segmentation (Bhade et al. 2018 ). To calculate the BIC, the log-likelihood of the segmentation model is adjusted by penalizing complex models. The BIC score is obtained by subtracting the penalty term from the maximum log-likelihood, where a lower BIC value indicates a better-fitting and more parsimonious model.

The Calinski-Harabasz Index (CHI) evaluates the separation between the segments and the compactness within each segment (Kandeil et al. 2014 ). The CHI is calculated by comparing the within-segment dispersion to the between-segment dispersion. A higher value indicates that the segments are well-defined and that the dataset has been well-split into distinct segments.

The Davies Bouldin Index (DBI) assesses the similarity between the segments based on the distance between their centroids (i.e., midpoints). The lower the DBI score, the better the segmentation result, indicating that the segments are more compact and less scattered (Aryuni et al. 2018 ). The DBI penalizes algorithms that produce segments with a wide variety of sizes and shapes, with larger diameters relative to the separation between the segments.

The Dunn Index (DI) evaluates the separation between the segments and the compactness within each segment (Khajvand and Tarokh 2011 ). The calculation of the DI is done by taking the ratio of the minimum intra-segment distance to the maximum inter-segment distance. A higher value indicates that the segments are better separated.

The Fukuyama and Sugeno method (FS) is an approach to evaluate segmentation results based on fuzzy sets and fuzzy logic. The FS involves assigning membership values to data points indicating their degree of belongingness to each segment. The process incorporates expert knowledge through the formulation of fuzzy rules, which guide the decision-making process. These membership values are then used to calculate a validity index, which measures the quality of the segmentation by considering their compactness and separation (Nemati et al. 2018 ).

The Mann–Whitney rank test (MW) is a non-parametric statistical test used to compare the difference in the median between two groups. In customer segmentation, the MW can be used to test whether there are significant differences in customer characteristics between the different segments (Jiang and Tuzhilin 2009 ).

The Silhouette Index (SI) is a metric used to measure how well-defined a segment is and how strongly a data point is assigned to its associated segment (Dzulhaq et al. 2019 ). The SI ranges from − 1 to 1, with a score of 1 indicating that the data point is perfectly matched to its own segment, a score of − 1 indicating that the data point is more closely associated with another segment, and a score close to 0 indicating that the data point does not have a clear segment assignment. The SI is calculated by taking the average of the difference between the data point’s own segment similarity and the lowest segment similarity with the other segments.

The Total Clustering Effectiveness (TCE) is a metric used by Lu and Wu ( 2009 ). The TCE combines an inter-cluster correlation indicator and an inner density indicator. The numerator represents the sum of densities for two segments, where a higher value indicates better performance. The denominator represents the correlation coefficient between the two segments, with a smaller value indicating better results. Incorporating both values, a higher TCE indicates better results.

The Validity Index (VI) is a measure that evaluates the quality of the segmentation result based on specific criteria (Pramono et al. 2019 ). These criteria can include factors such as intra-segment cohesion (compactness), inter-segment separation, or the overall structure of the segments. VIs provide a numerical score or value that indicates the goodness of fit of the segmentation solution, with higher values suggesting better quality or validity. Different VIs employ distinct formulas or methodologies to capture different aspects of segmentation performance.

Finally, the Xie-Beni Index (XBI) is a validity index used to evaluate the quality of segmentation results (Munusamy and Murugesan 2020 ). The XBI quantifies the trade-off between segment compactness and separation by calculating the ratio of the sum of squared distances between data points and segment centroids to the product of the segment compactness and the number of data points. A lower value indicates better segmentation with tighter and well-separated segments.

A few conclusions can be made. First, the metrics can be divided into statistics- and separation-focused metrics, with the latter being slightly more typical. Second, statistics-focused metrics emphasize segment-to-segment differences, while separation-focused metrics emphasize low intra-segment distance (i.e., compactness) and high inter-segment distance (i.e., separation). Third, customer segmentation evaluation is centered on using metrics derived from clustering practices, rather than using metrics especially tailored to customer segmentation, business outcomes, or ecological validity.

RQ5: What hyperparameters are used for algorithmic customer segmentation?

In ML, a hyperparameter refers to a configuration setting (i.e., a numeric value) that is external to the model itself and is typically set before the learning process begins (Jansen et al. 2021 ). Unlike model parameters, which are learned from the training data, hyperparameters are predefined choices (range of values when experimenting with multiple hyperparameter values) that affect the model’s performance and behavior. These parameters can include things like the learning rate, the number of hidden layers in a neural network, or the regularization parameter. Selecting appropriate hyperparameter values is crucial for achieving optimal model performance, and it often involves experimentation, trial and error, or using techniques like grid search or Bayesian optimization (Jansen et al. 2021 ).

In our review, out of the 169 studies that offered information about hyperparameters, more than four out of every five articles ( n  = 138, 81.7%) applied only segment size as a hyperparameter, while less than one out of five ( n  = 31, 18.3%) applied additional hyperparameters. In ACS, the algorithms may combine technical and business hyperparameters, with the technical parameters stemming from the inputs required by the algorithm (most commonly, the number of segments to create, i.e., the segment size and the distance measure—how the distance between the segments is calculated) to perform its computation and the latter arising from the particular business scenario the segmentation aims to address. For example, Munusamy and Murugesan ( 2020 ) performed customer segmentation based on the Fuzzy C-mean clustering algorithm and defined U matrix parameters to make their data compatible with the Fuzzy C-mean algorithm.

In contrast, Peker et al. ( 2017 ) formulated a hybrid approach for customer behavior prediction and used many AI/ML algorithms, including Neural Networks, SVM, Decision Tree, and Radial Basis Functions that make use of different hyperparameters, for instance, cost, gamma, number of hidden layers, weights, number of leaf nodes and number of trees, etc. In another contribution, the authors formulated a hybrid big data model for analyzing customer patterns in an integrated supply chain network (Wang et al. 2020 ). They applied Linked Based Bloom Filters (LBF) that served as parameter functions directly linked with customer segmentation. Liu et al. ( 2009 ) formulated a hybrid approach for a product recommendation that directly relates to customer segmentation. Their proposed approach used learning rate, grid structure, and distance normalization as hyperparameters.

Although these parameters are not directly linked with customer segmentation, they contribute to the overall segmentation process in terms of providing a method for ascertaining a technically optimal number (and structure) of the segments. In turn, the business parameters aim to provide more information about the domain-specific business context. These may include the likes of Length of customer involvement (L) and Periodicity (P) that were applied by Nemati et al. ( 2018 ). Zhu et al. ( 2015 ) applied Profitability ( prof ), Accuracy ( acc ), and lead time as hyperparameters that contribute to customer segmentation in demand fulfillment of customers in case of supply shortage. Wu and Liu ( 2020 ) incorporated group preferences and linguistics parameters into their Type 2 fuzzy customer segmentation models and concluded that these parameters greatly affect the customer segmentation task.

Overall, the hyperparameters applied by most articles stem from the standard/default hyperparameters used by these algorithms in any dimensionality reduction context, of which customer segmentation is a special case.

RQ6: How frequently are subject matter experts used for evaluating the customer segments?

In this study, we encountered only seven cases (4.1%) where subject matter experts were used to evaluate the quality of customer segmentation and provide expert opinions. In other words, it is a rare, perhaps too rare, practice to invite stakeholders and subject matter experts to validate the results of the customer segmentation process in academic research articles. Out of the rare examples that do exist, Nemati et al. ( 2018 ) formulated a customer lifetime value (CLV) approach for prioritizing marketing strategies in the telecom industry. The experts were first asked through a questionnaire to provide different parameters for the said tasks. Once the segmentation was done, the experts were again consulted to evaluate and validate the results. Safari et al. ( 2016 ) formulated an RFM-based CLV determination approach that performs customer segmentation based on the RFM values. To do so, subject matter experts were asked through a questionnaire, and once the segmentation was carried out, a total number of 16 experts expressed their opinions about the accuracy of the segments.

Similarly, Sun et al. ( 2021 ) introduced a heuristic approach to customer segmentation. The experimental results show that the customer segmentation output by their proposed method was consistent with the customer segmentation result given by experts. Manidatta et al. ( 2021 ) introduced an integrated approach for customer segmentation and evaluated their approach through experiments. They collected responses from nine subject matter experts from the Indian retail industry regarding their perception of the relative importance of four CLV criteria and evaluated the weights of each criterion using fuzzy AHP. Transaction data for 18 months was analyzed to segment 1,600 customers into eight segments using the fuzzy c-means clustering analysis technique. The segmentation results of their proposed integrated method were further validated by the nine experts from the Indian retail industry.

In another study, Li et al. ( 2011 ) formulated an agglomerative clustering-based approach for customer segmentation, and as a result of their proposed approach, the customer was segmented into four distinct groups/segments. Subject experts validated and evaluated the experimental results for customer segmentation by their approach (Li et al. 2011 ). Lee and Cho ( 2021 ) formulated a customer segmentation approach based on the Leuven algorithm. To verify and validate the segmentation results of their proposed approach, they consulted a subject matter expert, and the algorithm determined the modularity for ten segments, which was the same number of segments identified by the domain expert. Warner ( 2019 ) conducted a study on audience segmentation using a survey approach. Before conducting the study, a seven-member expert panel was asked to review the instrument and provide their expert opinion on the number of segments created. The domain expert team validated the audience segmentation results.

Thus, our analysis indicates that most often, the results of customer segmentation (including both algorithmic and non-algorithmic customer segmentation research) are not validated using external feedback, but the authors tend to rely on technical evaluation metrics to justify the quality of their work. This practice likely stems from the ML research tradition, in which metrics such as accuracy, precision, recall, etc., are used to evaluate the performance of an algorithm (Bell 2014 ; Kühl et al. 2022 ), rather than “subjective” human feedback. However, when they are used, most typically, multiple subject matter experts participate in evaluating the created segments.

Study highlights

Here, we discuss the highlights of our findings.

We identified 46 different algorithms applied by researchers for customer segmentation . This finding highlights not only the methodological (algorithmic) plurality within customer segmentation studies but also the influence of the current AI and ML technologies in this domain, as these algorithms overwhelmingly stem from the ML research tradition.

Most of the reviewed studies used one segmentation algorithm, making multi-algorithm customer segmentation a minority endeavor . The promise in multi-algorithms customer segmentation is that, in theory, it is better able to handle the plurality of segmentation criteria and be more responsive to organizational requirements for changing the segmentation parameters with updating business requirements.

In ACS, the number of segments is not pre-assigned, but it is inductively determined based on quantitative evaluation metrics. On average, researchers create 5.7 customer segments per study (SD  =  3.9, Mode  =  4, Median  =  5) . So, even though creating four segments is the most common, the number of segments created varies substantially across the reviewed studies. No analyzed study created more than 20 segments (Min = 1, Max = 20).

Few studies explicitly define the concept of customer segmentation . Instead, the concept is often treated implicitly, as “everyone knows what it is”. This conceptual vagueness can hinder the development of scientific advances in customer segmentation, as ‘customer segmentation’ might not be a similar task to other clustering tasks [the (dis)similarity of customer segmentation to other clustering tasks remains unaddressed in the literature].

Research outlines numerous ‘theoretical’ use cases and benefits for customer segmentation . These benefits, ranging from pricing to targeting and personalization of offerings and messaging, emphasize the central role of customer segmentation as a key business application of AI and ML technologies. Concurrently, few studies show empirical evidence of these benefits in organizational use cases or systems (i.e., ecological validity).

There is no one set of customer segmentation criteria, but the studies vastly vary in terms of the segmentation criteria applied . In fact, no two studies may have a single criterion (i.e., customer attribute) in common. This plurality of criteria partially explains the algorithmic or methodological plurality, as different criteria represent different data types that require distinct preprocessing and analysis approaches to apply the algorithms. Therefore, it is unlikely that we would end up in a situation where only one (or even a few) algorithms would cover all use cases for customer segmentation.

Researchers developing novel customer segmentation algorithms tend to see customer segmentation as a computational task . This viewpoint is visible in how algorithms are used, studies are structured, and outputs are evaluated. Machine learning studies follow a particular paradigm of benchmark comparison, which may explain why a large portion of literature puts less emphasis on conceptual and theoretical aspects of customer segmentation and instead focuses on it as a technical problem or ‘task’.

We identified 14 unique evaluation metrics for the quality of customer segmentation, all technical . The many metrics are the consequence of multiple algorithms: because the statistical and mathematical properties of different algorithms vary, one metric cannot be applied to evaluate the (internal) success of the modeling task. However, it would be valuable to have more centralized evaluation metrics for customer segmentation success; as the internal evaluation is affected by computational specificities, perhaps researchers could shift their focus on external (ecological) evaluation metrics, focusing on the business outcomes and the customer dynamics of applying customer segmentation rather than the segments’ creation process.

Business orientation separates customer segmentation from other clustering tasks . In addition to ML-dependent technical hyperparameters (e.g., number of segments, distance function), researchers utilize business-specific hyperparameters (e.g., length of customer involvement, profitability) for customer segmentation inputs. However, most studies reviewed (82%) only applied segment size as the hyperparameter.

Considering the historical development of the use of clustering in marketing, the post-2000 sample we analyzed shows some progress compared to the previous analysis made by Punj and Stewart in 1983. Specifically, the researchers indicated then that there is a “failure of numerous authors in the marketing literature to specify what clustering method is being used.” (p. 134). In our more recent sample, this condition does not take place as the authors are more explicit on the precise method being used—so, there has been progress in the reporting of clustering details in marketing work.

However, our results confirm the fundamental challenges of clustering (segmentation) as stated by Punj and Stewart ( 1983 ): “choice of an appropriate metric, selection of variables, cross- validation, and external validation” (p. 134). As then, these challenges remain topical and fundamentally unresolved. There are many metrics to choose from. The selection of segmentation variables is arbitrary. Cross-validation and external validation (ecological validity) are difficult to execute and thus often omitted. So, the fundamental nature of segmentation has not changed with the novel AI technologies, at least yet.

Interestingly, there can be seen as a continuation of knowledge. That is, the analysis by Punj and Stewart from 1983 (see their Table 4 on pp. 141–142) indicates K-Means as one of the most popular algorithms. Forty years later, this algorithm still maintains its position as the leading segmentation algorithm. We can interpret this finding as either proof of its superiority in this problem, or as traditionalism. However, either way, the conclusion remains that the novel AI-based approaches have not been able to replace the “old AI” approaches, at least when it comes to K-Means.

In addition to the above highlights, in the following subsection, we provide a taxonomy of algorithms for customer segmentation.

Central goals and directions for future research

From our review, there are multiple avenues for future research to advance customer segmentation research. In the following, we outline seven key goals (KG) for future work:

KG01: Providing taxonomies of algorithms and metrics. There is a need for conceptual frameworks, classifications, and taxonomies that help address the undeniable plurality of algorithms in the domain of customer segmentation research, which includes at least (a) algorithm selection plurality, (b) segmentation criteria plurality, (c) hyperparameter plurality, and (d) evaluation metric plurality. Our taxonomy of algorithms for customer segmentation provides a starting point and an example of outputs that can help address this gap. We invite other researchers to provide conceptual work (not only empirical!) that systemically categorizes the extant work on customer segmentation. Additionally, consensus on some fundamental concepts is much needed—for instance, how can ‘high-quality’ customer segmentation be distinguished from ‘low-quality’ customer segmentation? Should we focus on the quality of the process, the quality of the evaluation metrics, or the quality of the actual customer segments? Propositions (or even discussions) concerning these matters are direly needed.

KG02: Providing empirical evidence on customer segmentation outcomes. Based on our literature review, we observed that few studies provide an empirical analysis of the actual application of ACS in organizations. Algorithmic studies tend to stop at the stage of creating the customer segments; their application in companies is not explored. On the one hand, this casts shadows on whether the potential benefits of customer segmentation mentioned in the studies are rather hypothetical, or whether they can be backed up with empirical evidence. Therefore, we encourage researchers to shift their focus from creating segments to applying them in firms and other organizations. This not only represents an exciting research gap but efforts in this regard can help further enhance and incentivize research projects on customer segmentation, as they would be more strongly linked to key performance metrics that firms and other organizations value. To this end, case studies, field studies, A/B tests, experiments, and longitudinal studies would be welcome to address this vast knowledge gap.

KG03: Integrating algorithms into customer segmentation systems. One central direction for segmentation research is developing more comprehensive pipelines that can handle multiple different data types (i.e., customer segmentation criteria of different types) and changing business requirements. A key direction in this regard would be merging customer segmentation research with intelligent systems research (i.e., research that focuses on developing systems that can think, reason, and make decisions independently, without human intervention, interacting with the environment and making decisions to optimize quantitative outcomes (Bauer and Dey 2016 )) to generate and empirically investigate more comprehensive customer segmentation systems that stakeholders can interact with, not merely isolated attempts of testing how ‘Algorithm X’ fares with the customer segmentation task. Again, this corresponds to our postulation that, based on our review, customer segmentation research would benefit from a higher degree of ambition and scope, as the current body of work focuses on developing and testing algorithms instead of systems.

KG04: Proposing a standardized framework for evaluation. Due to methodological plurality, there is a lack of consensus on what constitutes ‘quality’ in customer segmentation. However, the consensus from prior research indicates that quality is perceived as models’ internal consistency and evaluated using technical performance metrics focused on this internal consistency. ACS has inherited its evaluation metrics from the ML research tradition, essentially adopting the metrics used in other clustering and data dimensionality tasks to customer segmentation as well. While we do not deny the merits of these technical metrics—they provide useful information about the model’s fit with the customer dataset—we call for further extensions and contributions to broaden the hierarchy of evaluating customer segmentation outputs. In addition to technical metrics, we ought to consider other metrics as well, such as stakeholder perception (“Are these segments useful? How useful?”) and organizational outcomes (e.g., the “ROI of customer segmentation”, i.e., how much does the implementation of customer segmentation improve the profit of the company?). A more holistic and nuanced, hierarchical way of measuring the quality of customer segmentation would further develop the field in academic and practical circles.

KG05: Exploring organizational challenges of customer segmentation. Challenges in customer segmentation arise, on the one hand, from organizational realities such as culture, capabilities, and individual experience and, on the other hand, from technical rationale such as data availability, selection of algorithms, and validating the quality of customer segmentation. The organizational aspect of applying customer segmentation in decision-making is a clear and present research gap. Again, qualitative studies can help generate rich insights into how customer segmentation algorithms transit into organizational adoption. So, we need a more collaborative approach to customer segmentation research. This means not giving up on developing better algorithms, but in addition to that, engaging with social science researchers in a pursuit to discover the impact of applying these algorithms in the real world. Such extension presents an exciting new field of study.

KG06: Providing more critical analyses. Researchers typically do not discuss the fundamental limitations of using AI and ML in customer segmentation, such as the fact that clustering algorithms were not originally developed for customer segmentation. Questions such as “What are the downsides of using AI for customer segmentation?” are not asked. Yet, they should be asked, as critical analyses can reveal insights into transformative improvements in this field. As a starting point for such analyses, we offer some ideas: first, the use of algorithms may distance the stakeholders from the data, especially if the creation of the segments is not a participatory process but outsourced to a group of analysts/data scientists who will not be part of the eventual use of the segments. So, there can be a problem of detachment and silos. Second, algorithms may cause segmentation to become overly rigid and resistant to change, so the same criteria used to group customers may not be relevant after a certain amount of time. Third, customers may resent being “put into a box” and may not appreciate feeling like just another anonymous statistic; thus, customer perceptions regarding customer segmentation could be explored. These and other negative aspects of ACS should be studied much more.

KG07: Making the role of humans explicit. Currently, humans play a central role in ACS; they select the algorithms and evaluation metrics, program the experiments, interpret the results, and choose final hyperparameter combinations. They also evaluate the results and make the final judgment of whether an algorithm did a good job at the segmentation. Finally, humans apply the segments in practice and make decisions based on them. Yet, we speak of ‘algorithmic customer segmentation’ and seem to delegate a lot of responsibility for the segmentation process to algorithms, often obfuscating the role of humans. As the field of customer segmentation is increasingly reliant on AI to do this work, a vital question is, what can AI learn from non-AI-reliant customer segmentation? In other words, there is a need to better understand human factors in the segmentation process and how these can support or bias the process when co-existing with algorithms.

Practical implications

There are three main practical implications (PI) for researchers and practitioners:

PI01: Segmenting Beyond K-means : Although nearly fifty algorithms were identified, k-means or a derivative is by far the most popular (27.6%) algorithmic approach for customer segmentation. This situation presents an opportunity and a need for exploring and comparing these algorithmic approaches to determine the most effective approach for different business scenarios—earlier research has partially done this (Punj and Stewart 1983 ), but not in a completely systematic way and not considering business scenarios. To do this comparing, it would entail both algorithmic investigation and business context research to identify characteristics for different business scenarios and how these relate to the selection of the best scenarios to address such analysis.

PI02: Sensemaking of the Segments : In addition to investigating approach algorithms given the attributes of business scenarios, there is a general lack of evaluation of the resulting segment, with most of the evaluations done using technical measures related to the chosen algorithm. However, given the algorithmic plurality, this results in a plurality of evaluation techniques as well. This calls for standard evaluation criteria for customer segments extending across the various algorithms. Finally, there is a critical and unmet need for evaluating segmentation beyond the algorithmic nuances, using broader criteria of accuracy, fairness, diversity, or coverage. For these, investigations of segmentation hyperparameters would most likely be needed.

PI03: Finding the “X” in [Segmentation + X] : The richness of segmentation criteria implies a nearly infinite number of ways of dividing a customer population into segments and these segments into sub-segments. Also, the segmentation criteria can be modified by adding or removing specific criteria at any time. Each of these segmentations offers one view of the segmentation that most likely has many possible views. So, which one is correct? This question is probably impossible without an ‘X’, i.e., some criteria external to the data for which to evaluate the segmentation. For this, the segmentation results need to be placed within the given business scenario, as discussed above, and the segments validated using external feedback, organizational key performance indicators, or achievement of business scenario goals.

Overall, customer segmentation, especially when applying AI and ML algorithms, is a socio-technical problem. In other words, both ‘technical’ (algorithm choice, data availability) and ‘social’ aspects (culture, goals) affect the success of customer segmentation projects. Therefore, mere technical solutions or more sophisticated algorithms are not adequate for ensuring successful customer segmentation projects. Thus, it is vital to understand customer segmentation projects as long-term processes that require stakeholder buy-in and effective implementation plans (i.e., the segmentation process does not end with creating the segments but only begins).

The current body of ACS literature focuses on the technical application of algorithms but largely omits the role of humans in this process, whether the role deals with various aspects of using the AI/ML algorithms (i.e., using judgment for the hyperparameter selection), evaluating the results, and eventually applying the results of ACS. Given the predominant focus on technical metrics and algorithms, there is a need to go beyond these aspects into the realm of inspecting the technology’s impact on actual organizations. According to our understanding, this can best be achieved by cross-disciplinary collaboration with social scientists, marketers, and other stakeholders who understand the qualitative side of customer segmentation and have access to organizational performance data. So, while this step of expanding ACS research into the realm of application is likely to involve a certain exit from the ML paradigm’s “comfort zone”, it is a necessary step to establish true scientific progress in this domain. Fernández-Delgado et al. ( 2014 ) famously asked in the context of classification, “Do we need hundreds of classifiers to solve real world classification problems?” (cited 3594 times at the time of writing this); we can paraphrase this question: Do we really need 46 different customer segmentation algorithms? Given the discrepancy between the large number of algorithms and the scarce number of articles applying subject matter experts for the evaluation of successful customer segmentation, the answer might be negative.

Study limitations

We did not include articles published before the year 2000 in our sample. We did not include keywords dealing with market segmentation or consumer segmentation (e.g., Kamakura and Russell 1989 )—conceptually, these are different goals, as ‘market’ includes non-customers as well. However, it could be interesting to compare the methods and variables used in these different segmentation tasks. We leave this for future work.

Customer segmentation has been a major focus in academic literature for many years and continues to be one. It is also of high value to marketers in industry. However, due to a myriad of different approaches, the field suffers from the lack of clarity that we aimed to address with this study. We found that researchers have used 46 different algorithms for customer segmentation. Interestingly, around 80% of them utilized a single algorithm for this purpose. On average, they created about 5.7 customer segments, deciding the exact number inductively based on quantitative evaluation metrics. Surprisingly, few articles offer empirical evidence of the benefits of customer segmentation. Our results point the way for future research, as addressing the proposed key goals helps successfully develop customer segmentation algorithms, make sense of the customer segments, and evaluate the impact of the segmentation.

Ballestar, M.T., P. Grau-Carles, and J. Sainz. 2018. Customer Segmentation in E-Commerce: Applications to the Cashback Business Model. Journal of Business Research 88: 407–414.

Article   Google Scholar  

Bauer, C., and A.K. Dey. 2016. Considering Context in the Design of Intelligent Systems: Current Practices and Suggestions for Improvement. Journal of Systems and Software 112: 26–47.

Bell, J. 2014. Machine Learning: Hands-on for Developers and Technical Professionals , 1st ed. Hoboken: John Wiley & Sons.

Book   Google Scholar  

Böttcher, M., M. Spott, D. Nauck, and R. Kruse. 2009. Mining Changing Customer Segments in Dynamic Markets. Expert Systems with Applications 36 (1): 155–164.

Cooil, B., L. Aksoy, and T.L. Keiningham. 2008. Approaches to Customer Segmentation. Journal of Relationship Marketing 6 (3–4): 9–39.

Coussement, K., F.A.M. Van den Bossche, and K.W. De Bock. 2014. Data Accuracy’s Impact on Segmentation Performance: Benchmarking RFM Analysis, Logistic Regression, and Decision Trees. Journal of Business Research 67 (1): 2751–2758.

Deng, Y., and Q. Gao. 2020. RETRACTED ARTICLE: A Study on e-Commerce Customer Segmentation Management Based on Improved K-Means Algorithm. Information Systems and e-Business Management 18 (4): 497–510.

Fernández-Delgado, M., E. Cernadas, S. Barro, and D. Amorim. 2014. Do We Need Hundreds of Classifiers to Solve Real World Classification Problems? The Journal of Machine Learning Research 15 (1): 3133–3181.

Google Scholar  

Güçdemir, H., and H. Selim. 2015. Integrating Multi-Criteria Decision Making and Clustering for Business Customer Segmentation. Industrial Management & Data Systems 115 (6): 1022–1040.

Hiziroglu, A. 2013. A Neuro-Fuzzy Two-Stage Clustering Approach to Customer Segmentation. Journal of Marketing Analytics 1 (4): 202–221.

Hong, T., and E. Kim. 2012. Segmenting Customers in Online Stores Based on Factors That Affect the Customer’s Intention to Purchase. Expert Systems with Applications 39 (2): 2127–2131.

Hosseini, M., and M. Shabani. 2015. New Approach to Customer Segmentation Based on Changes in Customer Value. Journal of Marketing Analytics 3: 110–121.

Jansen, B., S. Jung, and J. Salminen. 2021. The Effect of Hyperparameter Selection on the Personification of Customer Population Data. International Journal of Electrical and Computer Engineering Research . https://doi.org/10.53375/ijecer.2021.31 .

Jiang, T., and A. Tuzhilin. 2009. Improving Personalization Solutions through Optimal Segmentation of Customer Bases. IEEE Transactions on Knowledge and Data Engineering 21 (3): 305–320.

Joung, J., and H. Kim. 2023. Interpretable Machine Learning-Based Approach for Customer Segmentation for New Product Development from Online Product Reviews. International Journal of Information Management 70: 102641.

Kamakura, W.A., and G.J. Russell. 1989. A Probabilistic Choice Model for Market Segmentation and Elasticity Structure. Journal of Marketing Research 26 (4): 379–390.

Kashwan, K.R., and C. Velu. 2013. Customer Segmentation Using Clustering and Data Mining Techniques. International Journal of Computer Theory and Engineering 5: 856–861.

Khajvand, M., and M.J. Tarokh. 2011. Estimating Customer Future Value of Different Customer Segments Based on Adapted RFM Model in Retail Banking Context. Procedia Computer Science 3: 1327–1332.

Kitchenham, B., O.P. Brereton, D. Budgen, M. Turner, J. Bailey, and S. Linkman. 2009. Systematic Literature Reviews in Software Engineering–a Systematic Literature Review. Information and Software Technology 51 (1): 7–15.

Kühl, N., M. Schemmer, M. Goutier, and G. Satzger. 2022. Artificial Intelligence and Machine Learning. Electronic Markets 32 (4): 2235–2244.

Kuo, R.J., C.H. Mei, F.E. Zulvia, and C.Y. Tsai. 2016. An Application of a Metaheuristic Algorithm-Based Clustering Ensemble Method to APP Customer Segmentation. Neurocomputing 205: 116–129.

Lee, Y., and S. Cho. 2021. User Segmentation via Interpretable User Representation and Relative Similarity-Based Segmentation Method. Multimedia Systems 27 (1): 61–72.

Lee, J.H., and S.C. Park. 2005. Intelligent Profitable Customers Segmentation System Based on Business Intelligence Tools. Expert Systems with Applications 29 (1): 145–152.

Li, K.J., D.K.H. Fong, and S.H. Xu. 2011. Managing Trade-in Programs Based on Product Characteristics and Customer Heterogeneity in Business-to-Business Markets. Manufacturing & Service Operations Management 13 (1): 108–123.

Liu, D.-R., C.-H. Lai, and W.-J. Lee. 2009. A Hybrid of Sequential Rules and Collaborative Filtering for Product Recommendation. Information Sciences 179 (20): 3505–3519.

Lu, T.-C., and K.-Y. Wu. 2009. A Transaction Pattern Analysis System Based on Neural Network. Expert Systems with Applications 36 (3, Part 2): 6091–6099.

Manidatta, R., M. Ray, K. Muduli, A. Banaitis, and A. Kumar. 2021. Integrated Approach of Fuzzy Multi-Attribute Decision Making and Data Mining for Customer Segmentation. E a M: Ekonomie a Management 24: 174–188.

Manjunath, Y.S.K., and R.F. Kashef. 2021. Distributed Clustering Using Multi-Tier Hierarchical Overlay Super-Peer Peer-to-Peer Network Architecture for Efficient Customer Segmentation. Electronic Commerce Research and Applications 47: 101040.

Munusamy, S., and P. Murugesan. 2020. Modified Dynamic Fuzzy C-Means Clustering Algorithm—Application in Dynamic Customer Segmentation. Applied Intelligence 50 (6): 1922–1942.

Mustak, M., J. Salminen, L. Plé, and J. Wirtz. 2021. Artificial Intelligence in Marketing: Topic Modeling, Scientometric Analysis, and Research Agenda. Journal of Business Research 124: 389–404.

Nemati, Y., A. Mohaghar, M.H. Alavidoost, and H. Babazadeh. 2018. A CLV-Based Framework to Prioritize Promotion Marketing Strategies—Case Study of Telecom Industry. Iranian Journal of Management Studies . https://doi.org/10.22059/ijms.2018.242492.672837 .

Nie, D., P. Cappellari, and M. Roantree. 2020. A Methodology for Classification and Validation of Customer Datasets. Journal of Business & Industrial Marketing 36 (5): 821–833.

Nisan, S., Ş Elif, and Ç.U. Banu. 2022. Discrete Event Simulation Model Performed with Data Analytics for a Call Center Optimization. Istanbul Business Research 51 (1): 189–208.

Peker, S., A. Kocyigit, and P.E. Eren. 2017. A Hybrid Approach for Predicting Customers’ Individual Purchase Behavior. Kybernetes 46 (10): 1614–1631.

Punj, G., and D.W. Stewart. 1983. Cluster Analysis in Marketing Research: Review and Suggestions for Application. Journal of Marketing Research 20 (2): 134–148.

Repschlaeger, J., K. Erek, and R. Zarnekow. 2013. Cloud Computing Adoption: An Empirical Study of Customer Preferences among Start-up Companies. Electronic Markets 23 (2): 115–148.

Safari, F., N. Safari, and G.A. Montazer. 2016. Customer Lifetime Value Determination Based on RFM Model. Marketing Intelligence & Planning 34: 446–461.

Salminen, J., V. Yoganathan, J. Corporan, B.J. Jansen, and S.-G. Jung. 2019. Machine Learning Approach to Auto-Tagging Online Content for Content Marketing Efficiency: A Comparative Analysis between Methods and Content Type. Journal of Business Research 101: 203–217.

Salminen, J., K. Chhirang, S.-G. Jung, S. Thirumuruganathan, K.W. Guan, and B.J. Jansen. 2022. Big Data, Small Personas: How Algorithms Shape the Demographic Representation of Data-Driven User Segments. Big Data . https://doi.org/10.1089/big.2021.0177 .

Seret, A., S. Maldonado, and B. Baesens. 2015. Identifying next Relevant Variables for Segmentation by Using Feature Selection Approaches. Expert Systems with Applications 42 (15): 6255–6266.

Simões, D., and J. Nogueira. 2022. Learning about the Customer for Improving Customer Retention Proposal of an Analytical Framework. Journal of Marketing Analytics . https://doi.org/10.1057/s41270-021-00126-7 .

Spoor, J.M. 2022. Improving Customer Segmentation via Classification of Key Accounts as Outliers. Journal of Marketing Analytics . https://doi.org/10.1057/s41270-022-00185-4 .

Sun, Z.-H., T.-Y. Zuo, D. Liang, X. Ming, Z. Chen, and S. Qiu. 2021. GPHC: A Heuristic Clustering Method to Customer Segmentation. Applied Soft Computing 111: 107677.

Symeonidis, S., D. Effrosynidis, and A. Arampatzis. 2018. A Comparative Evaluation of Pre-Processing Techniques and Their Interactions for Twitter Sentiment Analysis. Expert Systems with Applications 110: 298–310.

Thirumuruganathan, S., N.A. Emadi, S. Jung, J. Salminen, D.R. Robillos, and B.J. Jansen. 2023. Will They Take This Offer? A Machine Learning Price Elasticity Model for Predicting Upselling Acceptance of Premium Airline Seating. Information & Management 60: 103759.

Tsai, C.-F., Y.-H. Hu, and Y.-H. Lu. 2015. Customer Segmentation Issues and Strategies for an Automobile Dealership with Two Clustering Techniques. Expert Systems 32 (1): 65–76.

Wang, S.-C., Y.-T. Tsai, and Y.-S. Ciou. 2020. A Hybrid Big Data Analytical Approach for Analyzing Customer Patterns through an Integrated Supply Chain Network. Journal of Industrial Information Integration 20: 100177.

Warner, L.A. 2019. Using Homeowners’ Association Membership to Define Audience Segments for Targeted Local Social Marketing Interventions: Implications From a Statewide Study. Social Marketing Quarterly 25 (4): 291–307.

Wong, E., and Y. Wei. 2018. Customer Online Shopping Experience Data Analytics: Integrated Customer Segmentation and Customised Services Prediction Model. International Journal of Retail & Distribution Management 46 (4): 406–420.

Wu, T., and X. Liu. 2020. A Dynamic Interval Type-2 Fuzzy Customer Segmentation Model and Its Application in E-Commerce. Applied Soft Computing 94: 106366.

Wu, S., W.-C. Yau, T.-S. Ong, and S.-C. Chong. 2021. Integrated Churn Prediction and Customer Segmentation Framework for Telco Business. IEEE Access 9: 62118–62136.

Yao, Z., P. Sarlin, T. Eklund, and B. Back. 2014. Combining Visual Customer Segmentation and Response Modeling. Neural Computing and Applications 25 (1): 123–134.

Zhou, J., J. Wei, and B. Xu. 2021. Customer Segmentation by Web Content Mining. Journal of Retailing and Consumer Services 61: 102588.

Aryuni, M., E.D. Madyatmadja, and E. Miranda. 2018. Customer Segmentation in XYZ Bank Using K-Means and K-Medoids clustering. In: 2018 International Conference on Information Management and Technology (ICIMTech), 412–416, 2018. IEEE.

Bhade, K., V. Gulalkari, N. Harwani, and S.N. Dhage. 2018. A Systematic Approach to Customer Segmentation and Buyer Targeting for Profit Maximization. In: 2018 9th International Conference on Computing, Communication and Networking Technologies (ICCCNT), 1–6, 2018. IEEE.

Dzulhaq, M.I., K.W. Sari, S. Ramdhan, and R. Tullah. 2019. Customer Segmentation Based on RFM Value Using K-Means Algorithm. In: 2019 Fourth International Conference on Informatics and Computing (ICIC), 1–7, 2019. IEEE.

Jung, S.-G., J. Salminen, J. An, H. Kwak, and B.J. Jansen. 2018. Automatically Conceptualizing Social Media Analytics Data via Personas. In: Proceedings of the International AAAI Conference on Web and Social Media (ICWSM 2018), June 25, 2018. San Francisco, California, USA.

Kandeil, D.A., A.A. Saad, and S.M. Youssef. 2014. A Two-Phase Clustering Analysis for B2B Customer Segmentation. In: 2014 International Conference on Intelligent Networking and Collaborative Systems, 221–228, 2014. IEEE.

Pramono, P.P., I. Surjandari, and E. Laoh. 2019. Estimating Customer Segmentation Based on Customer Lifetime Value Using Two-Stage Clustering Method. In: 2019 16th International Conference on Service Systems and Service Management (ICSSSM), 1–5, 2019. IEEE.

Ranjan, A., and S. Srivastava. 2022. Customer Segmentation Using Machine Learning: A Literature Review. In: AIP Conference Proceedings, 2022. vol. 2481. AIP Publishing LLC, p. 020036.

Syakur, M.A., B.K. Khotimah, E.M.S. Rochman, and B.D. Satoto. 2018. Integration K-Means Clustering Method and Elbow Method for Identification of the Best Customer Profile Cluster. In: IOP Conference Series: Materials Science and Engineering, 2018. vol. 336, 012017. IOP Publishing.

Xu, X., N. Yuruk, Z. Feng, and T.A. Schweiger. 2007. Scan: A Structural Clustering Algorithm for Networks. In: Proceedings of the 13th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 824–833, 2007.

Zhu, B., C. He, and X. Jiang. 2015. A Consistency-Based Validation for Data Clustering—IOS Press [Online]. https://content.iospress.com/articles/intelligent-data-analysis/ida727 . Accessed 3 Mar 2023.

Download references

Open Access funding provided by University of Vaasa (UVA). Funding was provided by Liikesivistysrahasto.

Author information

Authors and affiliations.

University of Vaasa, Vaasa, Finland

Joni Salminen

Hanken School of Economics, Helsinki, Finland

Mekhail Mustak

University of Jyväskylä, Jyvaskyla, Finland

Muhammad Sufyan

Qatar Computing Research Institute, Hamad Bin Khalifa University, Doha, Qatar

Bernard J. Jansen

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Joni Salminen .

Ethics declarations

Conflict of interest.

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Salminen, J., Mustak, M., Sufyan, M. et al. How can algorithms help in segmenting users and customers? A systematic review and research agenda for algorithmic customer segmentation. J Market Anal 11 , 677–692 (2023). https://doi.org/10.1057/s41270-023-00235-5

Download citation

Revised : 11 April 2023

Accepted : 21 June 2023

Published : 06 July 2023

Issue Date : December 2023

DOI : https://doi.org/10.1057/s41270-023-00235-5

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Customer segmentation
• Machine learning
• Find a journal
• Publish with us
• Track your research

Play with a live Neptune project -> Take a tour 📈

Implementing Customer Segmentation Using Machine Learning [Beginners Guide]

These days, you can personalize everything. There’s no one-size-fits-all approach. But, for business, this is actually a great thing. It creates a lot of space for healthy competition and opportunities for companies to get creative about how they acquire and retain customers.

One of the fundamental steps towards better personalization is customer segmentation. This is where personalization starts, and proper segmentation will help you make decisions regarding new features, new products, pricing, marketing strategies, even things like in-app recommendations.

But, doing segmentation manually can be exhausting. Why not employ machine learning to do it for us? In this article, I’ll tell you how to do just that.

What is customer segmentation

Customer segmentation simply means grouping your customers according to various characteristics (for example grouping customers by age).

It’s a way for organizations to understand their customers. Knowing the differences between customer groups, it’s easier to make strategic decisions regarding product growth and marketing.

The opportunities to segment are endless and depend mainly on how much customer data you have at your use. Starting from the basic criteria, like gender, hobby, or age, it goes all the way to things like “time spent of website X” or “time since user opened our app”. 

There are different methodologies for customer segmentation, and they depend on four types of parameters: 

• geographic, 
• demographic, 
• behavioral,
• psychological.  

Geographic customer segmentation is very simple, it’s all about the user’s location. This can be implemented in various ways. You can group by country, state, city, or zip code.

Demographic segmentation is related to the structure, size, and movements of customers over space and time. Many companies use gender differences to create and market products. Parental status is another important feature. You can obtain data like this from customer surveys.

Behavioral customer segmentation is based on past observed behaviors of customers that can be used to predict future actions. For example, brands that customers purchase, or moments when they buy the most. The behavioral aspect of customer segmentation not only tries to understand reasons for purchase but also how those reasons change throughout the year.

Psychological segmentation of customers generally deals with things like personality traits, attitudes, or beliefs. This data is obtained using customer surveys, and it can be used to gauge customer sentiment.

Advantages of customer segmentation

Implementing customer segmentation leads to plenty of new business opportunities. You can do a lot of optimization in:

• budgeting, 
• product design, 
• promotion, 
• marketing, 
• customer satisfaction. 

Let’s discuss these benefits in more depth.

Nobody likes to invest in campaigns that don’t generate any new customers. Most companies don’t have huge marketing budgets, so that money has to be spent right. Segmentation enables you to target customers with the highest potential value first, so you get the most out of your marketing budget. 

• Product design

Customer segmentation helps you understand what your users need. You can identify the most active users/customers, and optimize your application/offer towards their needs. 

Properly implemented customer segmentation helps you plan special offers and deals. Frequent deals have become a staple of e-commerce and commercial software in the past few years. If you reach a customer with just the right offer, at the right time, there’s a huge chance they’re going to buy. Customer segmentation will help you tailor your special offers perfectly.

The marketing strategy can be directly improved with segmentation because you can plan personalized marketing campaigns for different customer segments, using the channels that they use the most.

• Customer satisfaction

By studying different customer groups, you learn what they value the most about your company. This information will help you create personalized products and services that perfectly fit your customers’ preferences.

In the next section, we’re going to discuss why machine learning for customer segmentation.

Machine Learning for customer segmentation

Machine learning methodologies are a great tool for analyzing customer data and finding insights and patterns. Artificially intelligent models are powerful tools for decision-makers. They can precisely identify customer segments, which is much harder to do manually or with conventional analytical methods.

There are many machine learning algorithms, each suitable for a specific type of problem. One very common machine learning algorithm that’s suitable for customer segmentation problems is the k-means clustering algorithm . There are other clustering algorithms as well such as DBSCAN, Agglomerative Clustering, and BIRCH, etc.

Why would you implement machine learning for customer segmentation?

Manual customer segmentation is time-consuming. It takes months, even years to analyze piles of data and find patterns manually.  Also if done heuristically, it may not have the accuracy to be useful as expected.

Customer segmentation used to be done manually and wasn’t too precise. You’d manually create and populating different data tables, and analyze the data like a detective with a looking glass. Now, it’s much better (and relatively easy thanks to rapid progress in ML) to just use machine learning, which can free up your time to focus on more demanding problems that require creativity to solve.

Ease of retraining

Customer Segmentation is not a “develop once and use forever” type of project. Data is ever-changing, trends oscillate, everything keeps changing after your model is deployed. Usually, more labeled data becomes available after development, and it’s a great resource for improving the overall performance of your model. 

There are many ways to update customer segmentation models, but here are the two main approaches:

• Use the old model as the starting point and retrain it.
• Keep the existing model and combine its output with a new model.

Better scaling 

Machine learning models deployed in production support scalability, thanks to cloud infrastructure. These models are quite flexible for future changes and feedback. For example, consider a company that has 10000 customers today, and they’ve implemented a customer segmentation model. After a year, if the company has 1 million customers, then ideally we don’t need to create a separate project to handle this increased data. Machine Learning models have the inherent capability to handle more data and scale in production.

Higher accuracy

The value of an optimal number of clusters for given customer data is easy to find using machine learning methods like the elbow method. Not only the optimal number of clusters but also the performance of the model is far better when we use machine learning.

F1 Score vs ROC AUC vs Accuracy vs PR AUC: Which Evaluation Metric Should You Choose?

Exploring customer dataset and its features

Let’s analyze a customer dataset . Our dataset has 24,000 data points and four features. The features are:

• Customer ID – This is the id of a customer for a particular business.
• Products Purchased – This feature represents the number of products purchased by a customer in a year.
• Complaints – This column value indicates the number of complaints made by the customer in the last year
• Money Spent – This column value indicates the amount of money paid by the customer over the last year.

In the upcoming section, we’ll pre-process this dataset.

Pre-processing the dataset

Before feeding the data to the k-means clustering algorithm, we need to pre-process the dataset. Let’s implement the necessary pre-processing for the customer dataset.

Moving on, we’ll implement our k-means clustering algorithm in Python.

Might be useful

A Comprehensive Guide to Data Preprocessing

Implementing K-means clustering in Python

K-Means clustering is an efficient machine learning algorithm to solve data clustering problems. It’s an unsupervised algorithm that’s quite suitable for solving customer segmentation problems. Before we move on, let’s quickly explore two key concepts

Unsupervised Learning

Unsupervised machine learning is quite different from supervised machine learning. It’s a special kind of machine learning algorithm that discovers patterns in the dataset from unlabelled data. 

Unsupervised machine learning algorithms can group data points based on similar attributes in the dataset. One of the main types of unsupervised models is clustering models.

Note that, supervised learning helps us produce an output from the previous experience.

Clustering algorithms

A clustering machine learning algorithm is an unsupervised machine learning algorithm. It’s used for discovering natural groupings or patterns in the dataset. It’s worth noting that clustering algorithms just interpret the input data and find natural clusters in it.

Some of the most popular clustering algorithms are:

• K-Means Clustering
• Agglomerative Hierarchical Clustering
• Expectation-Maximization (EM) Clustering
• Density-Based Spatial Clustering 
• Mean-Shift Clustering

In the following section, we’re going to analyze the customer segmentation problem using the k-means clustering algorithm and machine learning. However, before that, let’s quickly discuss why we’re using the k-means clustering algorithm.

Why use K-means clustering for customer segmentation?

Unlike supervised learning algorithms, K-means clustering is an unsupervised machine learning algorithm. This algorithm is used when we have unlabelled data. Unlabelled data means input data without categories or groups provided. Our customer segmentation data is like this for this problem. 

The algorithm discovers groups (cluster) in the data, where the number of clusters is represented by the K value. The algorithm acts iteratively to assign each input data to one of K clusters, as per the features provided. All of this makes k-means quite suitable for the customer segmentation problem.

Given a set of data points are grouped as per feature similarity. The output of the K-means clustering algorithm is:

• The centroids values for K clusters,
• Labels for each input data point. 

At the end of implementation, we’re going to get output such as a group of clusters along with which customer belongs to which cluster.

First, we need to implement the required Python libraries as shown in the table below. 

We’ve imported the pandas, NumPy sklearn, plotly and matplotlib libraries. Pandas and NumPy are used for data wrangling and manipulation, sklearn is used for modelling, and plotly along with matplotlib will be used to plot graphs and images.

After importing the library, our next step is to load the data in the pandas data frame. For this, we’re going to use the read_csv method of pandas.

After loading the data, we need to define the K- means model. This is done with the help of the KMeans class that we imported from sklearn, as shown in the code below.

After defining the model, we want to train is using a training dataset. This is implemented with the use of the fit method, as shown in the code below.

Note that we’re passing three features to the fit method, namely products_purchased, complains, and money_spent.

Though we have trained a K-means model up to these points, we haven’t found the optimal number of clusters required in this case of customer segmentation. Finding the optimal number of clusters, for the given dataset is important for producing a high-performant k-means clustering model. 

In the upcoming section, we’re going to find the optimal number of clusters of the given dataset and then re-train the k-means clustering model with these optimal values of k. This will produce our final model.

Finding the optimal number of clusters

Finding the optimal number of clusters is one of the key tasks when implementing a k-means clustering algorithm. It’s worth noting that a k-means clustering model might converge for any value of K, but at the same time, not all values of K will produce the best model.

For some datasets, data visualization can help understand the optimal number of clusters, but this doesn’t apply to all datasets. We have a few methods, such as the elbow method, gap statistic method, and average silhouette method, to assess the optimal number of clusters for a given dataset. We’ll discuss them one by one.

• The elbow method  finds the value of the optimal number of clusters using the total within-cluster sum of square values. This represents how spread-apart the generated clusters are from one another. In this case, the K-means algorithm is evaluated for several values of k, and the within-cluster sum of square values is calculated for each value of k. After this, we plot the K versus the sum of square values. After analyzing this graph, the number of clusters is selected, so that adding a new cluster doesn’t change the values of the sum of square values significantly.
• Average silhouette method is a measure of how well each data point fits its corresponding cluster. This method evaluates the quality of clustering. As a general rule, a high average silhouette width denotes better clustering output.
• Gap statistic method  is a measure of the value of gap statistics. Gap statistics is the difference between the total intracluster changes for various values of k compared to their expected values. This calculation is done using the null reference distribution of data points. The optimal number of clusters is the value that maximizes the value of gap statistics.

We’re going to use the elbow method. The K-means clustering algorithm clusters data by separating given data points in k groups of equal variances. This effectively minimizes a parameter named inertia. Inertia is nothing but within-cluster sum-of-squares distances in this case.

When we use the elbow method, we gradually increase the number of clusters from 2 until we reach the number of clusters where adding more clusters won’t cause a significant drop in the values of inertia. 

The stage at this number of clusters is called the elbow of the clustering model. We’ll see that in our case it’s K =5.

For implementing the elbow method, the below function named “try_different_clusters” is created first. It takes two values as input:

• K (number of clusters),
• data (input data).

The method try_different_clusters is called using the below code, where we pass the values of K from 1 to 12 and calculate the inertia for each value of k.

Using the below code, we plot the value of K (on the x-axis) against corresponding values of inertia on the Y-axis.

We can generate the below plot using the above code. The elbow of the code is at K=5. We have chosen 5 as if we increase the number of clusters to more than 5, there is very small change in the inertia or sum of the squared distance.

Optimal value of K = 5

The stage at which the number of clusters is optimal is called the elbow of the clustering model. For example, in the below image, the elbow is at five clusters (K =5). Adding more than 5 clusters will cause the creation of an inefficient or less performant clustering model.

As discussed before, we need to train the k-means clustering model again with the optimal number of clusters found.

Note that we’re using the fit_predict method to train the model.

In the next section, we’re going to discuss how to visualize customer segmentation clusters in three dimensions.

Visualizing customer segments

In this section, we’ll be implementing some code using plotly express. This way we’ll visualize the clusters in three dimensions, formed by our k-means algorithm. Plotly express is a library based on plotly that works on several types of datasets and generates highly-styled plots.

First, let’s add a new column named ‘clusters’ to the existing customer data dataset. This column will be able to tell which customer belongs to what cluster.

Note that we’re using NumPy expm1 methods here. NumPy expm1 function returns the exponential value of minus one for each element given inside a NumPy array as output. Therefore, the np.expm1 method accepts arr_name and out arguments and then returns the array as outputs.

After adding the new column, named clusters, the customer data dataset will look as below.

Finally, we’re going to use the below code to visualize the five clusters created. This is done using plotly with the express library.

Plotly is a Python library used for graphing, statistics, plotting, and analytics. This can be used along with Python, R, Julia, and other programming languages. Plotly is a free and open-source library. 

Want to organize your experimentation process? Check how you can have interactive Plotly charts stored in the same place as the rest of your model metadata (metrics, parameters, weights, and more).

Plotly Express is a high-level interface over Plotly, that works on several types of datasets and generates highly-styled plots.

The plotly.express class has functions that can produce entire figures in one go. Generally, it’s referred to as px. It’s worth noting plotly express is the built-in module of the plotly library. This is the starting point of creating the most common plots as recommended. Note that each plotly express function creates  graph objects  internally and returns plotly.graph_objects. 

A graph created by a single method call using plotly express can be also created using graph objects only. However, in that case, it takes around 5 to 100 times as much code.

As the  2D scatter plot , px.scatter plots individual data in a two-dimensional space, and the 3D method px.scatter_3d plots individual data in a three-dimensional space.

Visualization of clusters of data points is very important. Various edges of the graph provide a quick view of the complex input data set.

It’s not wise to serve all customers with the same product model, email, text message campaign, or ad. Customers have different needs. A one-size-for-all approach to business will generally result in less engagement, lower-click through rates, and ultimately fewer sales. Customer segmentation is the cure for this problem.

Finding an optimal number of unique customer groups will help you understand how your customers differ, and help you give them exactly what they want. Customer segmentation improves customer experience and boosts company revenue. That’s why segmentation is a must if you want to surpass your competitors and get more customers. Doing it with machine learning is definitely the right way to go. 

If you made it this far, thanks for reading!

Was the article useful?

More about implementing customer segmentation using machine learning [beginners guide], check out our product resources and related articles below:, how to optimize hyperparameter search using bayesian optimization and optuna, customizing llm output: post-processing techniques, deep learning optimization algorithms, track and visualize information from your pipelines: neptune.ai + zenml integration, explore more content topics:, manage your model metadata in a single place.

Join 50,000+ ML Engineers & Data Scientists using Neptune to easily log, compare, register, and share ML metadata.

Case study: Building a customer-centric B2B organization

Customer experience (CX) is an increasingly important strategic topic in the boardrooms of B2B companies in China and throughout the world. Despite the rapid development of the previous decades, the “growth first” principle of Chinese enterprises sometimes implies customer experience can be sacrificed. But CX leaders, globally and within China, drive higher growth, lower cost, and superior customer satisfaction. In times of crisis, they achieve three-times-higher shareholder returns 1 Total return to shareholders tracked for publicly traded companies in the top 10 or bottom 10 of Forrester’s Customer Experience Performance Index in 2007–09. than laggards.

Start with a vision

A successful transformation starts from the top. Cases within and outside China confirm that the CEO must be in charge to continuously push and unify the organization.

The Chinese steel industry has taken an upturn amid the country’s overcapacity-reduction program, and companies have been enjoying robust price and volume increases. In this article, we consider one Chinese steel manufacturer whose CEO set a clear vision to build a customer-centric organization in order to gain a competitive edge and to keep the organization healthy through future downturns. The company took a series of steps to systematically and holistically shift the entire organization toward customer-centricity.

Would you like to learn more about our Marketing & Sales Practice ?

Identify the challenges.

Comprehensive diagnostics revealed that the company faced a series of challenges. In fact, interviews with some customers were alarming: the customer voice, though central to the CEO’s vision, had no conduit within the organization and was never heard by decision makers. One key account was lost well before corporate management heard its complaints. Analysis of the research revealed several serious shortfalls in customer-centricity:

Limited understanding of customers.

The company had not systemically mapped the diverse stakeholders behind each customer, relying instead in most cases on buy-side procurement managers and their associates as the only source of customer feedback. Company representatives rarely knew or approached other customer-decision influencers or the users behind procurement, thereby losing many potential customer insights. The company also lacked access to end customers further down the value chain.

Few channels for customer feedback.

As is true at many B2B organizations, sales was the major channel through which the company gathered customer feedback. But manual relays of messages could take a long time to reach managers, assuming they were not forgotten along the way. To make matters worse, sales representatives sometimes neglected to report feedback, fearing they would be punished if headquarters learned that their customers were unhappy.

Limited analysis of feedback for insights.

What customer feedback and CX data existed within the organization was not centrally managed and synthesized into easy-to-access reports to give top management the full picture. Other stakeholders also found it challenging to access the aggregated customer feedback related to their own roles.

Customer problems not addressed.

Many customers complained that issues they had reported many times had not been dealt with, and the same problems continued to persist.

Transform to a CX-centric organization through a holistic ‘diagnose, design, deliver’ process

A holistic transformation was crafted to move the company toward the CEO’s vision, knowing that no single silver bullet could address all challenges at the same time. The transformation plan consisted of multiple modules based on a “diagnose, design, deliver” process, which takes two to three years to implement fully (Exhibit 1).

The company proceeded through the process in three phases:

Phase 1: Diagnose

The first step was to map the customers and identify stakeholders beyond buy-side procurement. To achieve this, customers were divided into segments based on similar stakeholder dynamics and customer journeys. Then the segments were prioritized based on their value and strategic importance.

Phase 2: Design

After the journey diagnostics, the company built a structured “question library” based on the journey breakdown, with customized questionnaires and feedback forms for different stakeholders. This enabled the company to collect feedback and experience data, and perform a consistent longitudinal analysis across feedback channels. Using these designs, the company was able to systematically analyze experience data, dig into root causes, and identify improvement areas.

Phase 3: Deliver

An IT backbone had to be built to implement all the designs discussed in the previous paragraphs. To achieve this, the company broke down the system design into several modules and assessed how each one should be tackled. Among the three possible development options, “customized third-party solution, locally deployed” was chosen as the best option based on five evaluation criteria: feasibility, customization, data security, timeline, and price.

Survey: Chinese B2B decision maker response to COVID-19 crisis

Key learnings: prioritize segments, and collect feedback on multiple channels.

The company eventually prioritized three segments: (1) section-steel and steel-sheet-piling dealers, (2) section-steel manufacturers, and (3) steel-sheet-piling leasing companies, with the biggest customer in each category selected for deeper analysis. In analyzing the different customers, the company discovered a pattern: three journeys—scheduling inquiry, transport and delivery, and quality discrepancy—were deemed crucial by all customers.

A new, multichannel system was designed to address the company’s various challenges in collecting customer feedback. While customers can still share feedback directly with sales reps, the system incorporates new channels, including periodic on-site interviews and feedback sessions conducted by marketing personnel or the CX team, surveys on mobile devices, and a WeChat portal where customers can submit feedback whenever they want.

This system also allows the company to reach out to previously inaccessible or remote customers, who can simply scan product QR codes to submit feedback on features and quality, or even solicit technical support. A dashboard was designed to create CX transparency across the organization, allowing different stakeholders to analyze the data and generate insights. The multichannel-backed (PC and mobile) dashboard can make customer feedback and experience data visible for stakeholders from different divisions, so they can easily analyze data and generate insights.

Manage the change to maintain success

McKinsey research indicates that 70 percent of change programs fail, mostly because of human factors. Design-phase initiatives don’t stick without procedures for proper change management. McKinsey has a useful framework for change management (Exhibit 2), from which the steel manufacturer adopted key elements.

Real impact to the bottom line

To date, the company has already generated an estimated 4 percent increase in gross profit, or an 8 percent increase in pre-interest and pretax profit—a number matching the CEO’s initial expectations of the project. Moreover, the company believes that its transformation will have a lasting impact, producing better products, more satisfied and loyal customers, and a healthier, more efficient organization overall.

All in all, customer experience is an effective tool that Chinese B2B players can utilize to create long-term competitive advantages. A company should first define its priorities, lay out an implementation path based on its current reality, and use it to work toward a superior customer experience and, ultimately, excellence.

Hai Ye and Will Enger are partners in McKinsey’s Hong Kong office.

Explore a career with us

Related articles.

Two ingredients for successful B2B sales: Agility and stability

A post–COVID-19 commercial-recovery strategy for B2B companies

The B2B digital inflection point: How sales have changed during COVID-19

A Case Study on Customer Segmentation by using Machine Learning Methods

Ieee account.

• Change Username/Password
• Update Address

Purchase Details

• Payment Options
• Order History
• View Purchased Documents

Profile Information

• Communications Preferences
• Profession and Education
• Technical Interests
• US & Canada: +1 800 678 4333
• Worldwide: +1 732 981 0060
• Contact & Support
• About IEEE Xplore
• Accessibility
• Terms of Use
• Nondiscrimination Policy
• Privacy & Opting Out of Cookies

A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity. © Copyright 2024 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.

Brought to you by:

Customer Segmentation in Business-to-Business Markets

By: Robert E. Spekman, Joshua Stein

The purpose of this note is to help students better understand the concept of customer segmentation in a business-to-business (B2B) context, focused on such topics as the role segmentation plays in…

• Length: 28 page(s)
• Publication Date: Mar 18, 2011
• Discipline: Marketing
• Product #: UV5749-PDF-ENG

What's included:

• Educator Copy

$4.95 per student

degree granting course

$8.95 per student

non-degree granting course

Get access to this material, plus much more with a free Educator Account:

• Access to world-famous HBS cases
• Up to 60% off materials for your students
• Resources for teaching online
• Tips and reviews from other Educators

Already registered? Sign in

• Student Registration
• Non-Academic Registration
• Included Materials

The purpose of this note is to help students better understand the concept of customer segmentation in a business-to-business (B2B) context, focused on such topics as the role segmentation plays in the larger marketing strategy of which it is a part, the process, primary approaches, and variables.

Mar 18, 2011

Discipline:

Darden School of Business

UV5749-PDF-ENG

We use cookies to understand how you use our site and to improve your experience, including personalizing content. Learn More . By continuing to use our site, you accept our use of cookies and revised Privacy Policy .

10 Customer Segments Examples and Their Benefits

• May 9, 2024

Now that companies can segment buyers, the days of mass marketing are behind us. Customer segmentation offers various benefits for marketing, content creation, sales, analytics teams and more. Without customer segmentation, your personalised marketing efforts may fall flat. 

According to the Twilio 2023 state of personalisation report , 69% of business leaders have increased their investment in personalisation. There’s a key reason for this — customer retention and loyalty directly benefit from personalisation. In fact, 62% of businesses have cited improved customer retention due to personalisation efforts. The numbers don’t lie. 

Keep reading to learn how customer segments can help you fine-tune your personalised marketing campaigns. This article will give you a better understanding of customer segmentation and real-world customer segment examples. You’ll leave with the knowledge to empower your marketing strategies with effective customer segmentation. 

What are customer segments?

Customer segments are distinct groups of people or organisations with similar characteristics, needs and behaviours. Like different species of plants in a garden, each customer segment has specific needs and care requirements. Customer segments are useful for tailoring personalised marketing campaigns for specific groups.

Personalised marketing has been shown to have significant benefits — with 56% of consumers saying that a personalised experience would make them become repeat buyers . 

Successful marketing teams typically focus on these types of customer segmentation:

• Geographic segmentation: groups buyers based on their physical location — country, city, region or climate — and language.
• Purchase history segmentation: categorises buyers based on their purchasing habits — how often they make purchases — and allows brands to distinguish between frequent, occasional and one-time buyers. 
• Product-based segmentation: groups buyers according to the products they prefer or end up purchasing. 
• Customer lifecycle segmentation: segments buyers based on where they are in the customer journey. Examples include new, repeat and lapsed buyers. This segmentation category is also useful for understanding the behaviour of loyal buyers and those at risk of churning. 
• Technographic segmentation: focuses on buyers’ technology preferences, including device type, browser type, and operating system. 
• Channel preference segmentation: helps us understand why buyers prefer to purchase via specific channels — whether online channels, physical stores or a combination of both. 
• Value-based segmentation: categorises buyers based on their average purchase value and sensitivity to pricing, for example. This type of segmentation can provide insights into the behaviours of price-conscious buyers and those willing to pay premium prices. 

Customer segmentation vs. market segmentation

Customer segmentation and market segmentation are related concepts, but they refer to different aspects of the segmentation process in marketing. 

Market segmentation is the broader process of dividing the overall market into homogeneous groups. Market segmentation helps marketers identify different groups based on their characteristics or needs. These market segments make it easier for businesses to connect with new buyers by offering relevant products or new features. 

On the other hand, customer segmentation is used to help you dig deep into the behaviour and preferences of your current customer base. Marketers use customer segmentation insights to create buyer personas. Buyer personas are essential for ensuring your personalised marketing efforts are relevant to the target audience. 

10 customer segments examples

Now that you better understand different customer segmentation categories, we’ll provide real-world examples of how customer segmentation can be applied. You’ll be able to draw a direct connection between the segmentation category or categories each example falls under.

One thing to note is that you’ll want to consider privacy and compliance when you are considering collecting and analysing types of data such as gender, age, income level, profession or personal interests. Instead, you can focus on these privacy-friendly, ethical customer segmentation types:

1. Geographic location (category: geographic segmentation)

The North Face is an outdoor apparel and equipment company that relies on geographic segmentation to tailor its products toward buyers in specific regions and climates. 

For instance, they’ll send targeted advertisements for insulated jackets and snow gear to buyers in colder climates. For folks in seasonal climates, The North Face may send personalised ads for snow gear in winter and ads for hiking or swimming gear in summer. 

The North Face could also use geographic segmentation to determine buyers’ needs based on location. They can use this information to send targeted ads to specific customer segments during peak ski months to maximise profits.

2. Preferred language (category: geographic segmentation)

Your marketing approach will likely differ based on where your customers are and the language they speak. So, with that in mind, language may be another crucial variable you can introduce when identifying your target customers. 

Language-based segmentation becomes even more important when one of your main business objectives is to expand into new markets and target international customers — especially now that global reach is made possible through digital channels. 

Coca-Cola’s “Share a Coke” is a multi-national campaign with personalised cans and bottles featuring popular names from countries around the globe. It’s just one example of targeting customers based on language.

3. Repeat users and loyal customers (category: customer lifecycle segmentation)

Sephora, a large beauty supply company, is well-known for its Beauty Insider loyalty program. 

It segments customers based on their purchase history and preferences and rewards their loyalty with gifts, discounts, exclusive offers and free samples. And since customers receive personalised product recommendations and other perks, it incentivises them to remain members of the Beauty Insider program — adding a boost to customer loyalty.

By creating a memorable customer experience for this segment of their customer base, staying on top of beauty trends and listening to feedback, Sephora is able to keep buyers coming back.

4. New customers (category: customer lifecycle segmentation)

Subscription services use customer lifecycle segmentation to offer special promotions and trials for new customers. 

HBO Max is a great example of a real company that excels at this strategy: 

They offer 40% savings on an annual ad-free plan, which targets new customers who may be apprehensive about the added monthly cost of a recurring subscription.

This marketing strategy prioritises fostering long-term customer relationships with new buyers to avoid high churn rates. 

5. Cart abandonment (category: purchase history segmentation)

With a rate of 85% among US-based mobile users, cart abandonment is a huge issue for ecommerce businesses. One way to deal with this is to segment inactive customers and cart abandoners — those who showed interest by adding products to their cart but haven’t converted yet — and send targeted emails to remind them about their abandoned carts.

E-commerce companies like Ipsy, for example, track users who have added items to their cart but haven’t followed through on the purchase. The company’s messaging often contains incentives — like free shipping or a limited-time discount — to encourage passive users to return to their carts. 

Research has found that cart abandonment emails with a coupon code have a high 44.37% average open rate. 

6. Website activity (category: technographic segmentation)

It’s also possible to segment customers based on website activity. Now, keep in mind that this is a relatively broad approach; it covers every interaction that may occur while the customer is browsing your website. As such, it leaves room for many different types of segmentation. 

For instance, you can segment your audience based on the pages they visited, the elements they interacted with — like CTAs and forms — how long they stayed on each page and whether they added products to their cart. 

Matomo’s Event Tracking can provide additional context to each website visit and tell you more about the specific interactions that occur, making it particularly useful for segmenting customers based on how they spend their time on your website. 

Try Matomo for Free

Get the web insights you need, while respecting user privacy.

No credit card required

Amazon segments its customers based on browsing behaviour — recently viewed products and categories, among other things — which, in turn, allows them to improve the customer’s experience and drive sales.

7. Traffic source (category: channel segmentation) 

You can also segment your audience based on traffic sources. For example, you can determine if your website visitors arrived through Google and other search engines, email newsletters, social media platforms or referrals. 

In other words, you’ll create specific audience segments based on the original source. Matomo’s Acquisition feature can provide insights into five different types of traffic sources — search engines, social media, external websites, direct traffic and campaigns —  to help you understand how users enter your website.

You may find that most visitors arrive at your website through social media ads or predominantly discover your brand through search engines. Either way, by learning where they’re coming from, you’ll be able to determine which conversion paths you should prioritise and optimise further. 

8. Device type (category: technographic segmentation)

Device type is customer segmentation based on the devices that potential customers may use to access your website and view your content. 

It’s worth noting that, on a global level, most people (96%) use mobile devices — primarily smartphones — for internet access. So, there’s a high chance that most of your website visitors are coming from mobile devices, too. 

However, it’s best not to assume anything. Matomo can detect the operating system and the type of device — desktop, mobile device, tablet, console or TV, for example. 

By introducing the device type variable into your customer segmentation efforts, you’ll be able to determine if there’s a preference for mobile or desktop devices. In return, you’ll have a better idea of how to optimise your website — and whether you should consider developing an app to meet the needs of mobile users.

9. Browser type (category: technographic segmentation)

Besides devices, another type of segmentation that belongs to the technographic category and can provide valuable insights is browser-related. In this case, you’re tracking the internet browser your customers use. 

Many browser types are available — including Google Chrome, Microsoft Edge, Safari, Firefox and Brave — and each may display your website and other content differently. 

So, keeping track of your customers’ preferred choices is important. Otherwise, you won’t be able to fully understand their online experience — or ensure that these browsers are displaying your content properly. 

10. Ecommerce activity (category: purchase history, value based, channel or product based segmentation) 

Similar to website activity, looking at ecommerce activity can tell your sales teams more about which pages the customer has seen and how they have interacted with them. 

With Matomo’s Ecommerce Tracking , you’ll be able to keep an eye on customers’ on-site behaviours, conversion rates, cart abandonment, purchased products and transaction data — including total revenue and average order value.

Considering that the focus is on sales channels — such as your online store — this approach to customer segmentation can help you improve the sales experience and increase profitability. 

Start implementing these customer segments examples

With ever-evolving demographics and rapid technological advancements, customer segmentation is increasingly complex. The tips and real-world examples in this article break down and simplify customer segmentation so that you can adapt to your customer base. 

Customer segmentation lays the groundwork for your personalised marketing campaigns to take off. By understanding your users better, you can effectively tailor each campaign to different segments. 

If you’re ready to see how Matomo can elevate your personalised marketing campaigns, try it for free for 21 days . No credit card required.

21 day free trial. No credit card required.

7 Best Marketing Attribution Software in 2024

11 of the Most Effective Conversion Rate Optimisation Best Practices

What Is Ethical SEO & Why Does It Matter?

Subscribe to our newsletter to receive regular information about Matomo. You can unsubscribe at any time from it. This service uses MadMimi. Learn more about it within our privacy Policy page.

A powerful web analytics platform that gives you and your business 100% data ownership and user privacy protection.

No credit card required.

Free forever.

Newsletter Subscription - English