012 min
Why segmentation matters
A market is rarely one group of buyers with one set of wants. When wants differ, one offer fits some buyers well and the rest badly. The company then pays to reach buyers who will not choose it. A marketing mix built around the needs of chosen segments is likely to give a higher return on investment, because less effort is wasted on buyers whose needs the company could never satisfy. The term marketing mix means the four things a company controls: the product, its price, how it is promoted and how it is distributed.
A founder meets this problem when choosing the first customer to build for. A product manager meets it when one roadmap has to serve a small team and a large company. A designer meets it when the onboarding flow assumes a user who knows the product category already. In each case the question is the same: which group of buyers is this for, and what does that group need that another group does not?
Segmentation also forces a company to say what it is not for. A company that serves every buyer a little cannot be the best choice for any one of them. A phone company that offers one phone to the entire market is unlikely to satisfy the needs of any of the three groups. The next section shows where those three groups come from.
022 min
How segmentation works
Segmentation starts from one fact: buyers want different things. Economists call this heterogeneous demand, meaning that no single product, price or message is the best choice for everyone. Smith defined segmentation as viewing a heterogeneous market, which means a market with divergent demand, as a number of smaller homogeneous markets.
Dolnicar, Grün and Leisch, in their open-access textbook on segmentation, use a simple illustration. Picture a market for mobile phones. Plot each buyer by the number of features they want and the price they will pay. Three groups appear:
- A small group wants many features and will pay a lot.
- A large group wants the opposite: a simple, cheap phone.
- Another large group wants a mid-range phone at a mid-range price.
These three groups are the segments. Inside each one the buyers want nearly the same thing. Between groups they want different things. The textbook uses this example to illustrate Smith's definition, and it is a teaching example, not data from a real company.
The segmentation criterion
To make groups, the company first picks the buyer characteristic that matters most for its decision. The consumer characteristics that management treats as critical for a segmentation are called segmentation criteria. A criterion can be one characteristic, such as age or country, or a set of them, such as the benefits a buyer looks for.
Two ways to find the groups
The textbook describes two approaches. In a priori segmentation, the decision about what defines each segment is made in advance, before any data analysis. This is also called commonsense segmentation. For example, a company decides to split its customers by company size before looking at any data.
In data-driven segmentation, the company does not know the exact profile of the segments until after the analysis. The analyst chooses the variables, runs a statistical grouping method such as clustering, and reads the groups that come out. In practice the two are often combined, one after the other.
The result is a set of groups. Whether a group is worth serving is a separate question, covered under measurement below.
032 min
Common ways to divide a market
Every segmentation rests on a basis, which is the type of buyer information used to form the groups. The most common bases for segmenting consumer markets are geography, demographics, psychographics and behaviour.
| Basis | What it groups buyers by | Example criterion |
|---|---|---|
| Geographic | Where the buyer is | Country, region, city size |
| Demographic | Measurable facts about the person | Age, income, household size |
| Psychographic | Attitudes, values and lifestyle | Preference for novelty or for safety |
| Behavioural | What the buyer does | How often they buy, how they use the product |
Behavioural bases include the benefit a buyer is looking for. Benefit segmentation divides buyers by the distinct needs, perceived value or benefits sought from a product. This links segmentation to the jobs to be done idea, where a group is defined by the progress its members want to make rather than by who they are.
Companies that sell to other companies use a different set. Business-to-business marketers may segment markets by company type, industry or geographic location. Company size is another common criterion, and the group of these company-level bases is called firmographics.
A basis that is easy to measure is not always a basis that predicts behaviour. Age is easy to record, but two people of the same age may want different products. That is the main reason behavioural and benefit bases are often preferred.
041 min
Where segmentation comes from
Selling different things to different groups is older than the theory. Retailers outside large cities, for example, could not afford to serve one type of customer only. The formal idea entered the marketing literature in 1956. Wendell R. Smith, writing in the Journal of Marketing in 1956, was the first to propose segmentation as a marketing strategy. The article is titled "Product Differentiation and Market Segmentation as Alternative Marketing Strategies." Smith presented segmentation as a different choice from product differentiation, which tries to move the whole market's preferences toward one company's product.
Later historians placed segmentation inside a longer story of mass markets. The business historian Richard Tedlow described four stages in the history of segmentation. Under this account, the first stage was local sellers, the second was standard branded goods sold nationally, the third was products pitched to different groups, and the fourth is ever narrower groups made possible by digital communication.
052 min
Ford and General Motors
The clearest historical case of the choice between serving one market and serving segments is the Ford Motor Company and General Motors in the 1910s and 1920s.
Ford sold one car. Ford wrote in his autobiography that in 1909 he told his managers that any customer could have a car painted any colour so long as it was black. The black-only policy was put into effect only in 1914. Early Model Ts, from 1908 to 1913, were not available in black, so the famous policy describes a later stage. One product made at very large volume kept costs low, and the strategy worked: By 1918, half of all cars in the United States were Model Ts.
General Motors took the opposite approach. Under Alfred P. Sloan, General Motors set a price ladder of Chevrolet, Pontiac, Oldsmobile, Buick and Cadillac, from lowest to highest priced, so that the brands did not compete with each other. A buyer could start with a Chevrolet and move up the ladder as their income grew, without leaving the company.
| Ford Model T | General Motors under Sloan | |
|---|---|---|
| Products | One basic model, one colour policy | Five brands at five price levels |
| Assumption about buyers | Most want the same low-priced car | Buyers differ in income and preference |
| Buyer's path | Buy the one car | Move up the ladder over time |
The variable that separates the two cases is how much the company lets the product vary by buyer. This article does not claim one policy was always better. Ford's single product fit a market where most buyers wanted a first car at the lowest price. General Motors' ladder fit a market where many buyers already owned a car and wanted something more.

061 min
Three ways a company can respond to segments
Once segments are known, a company still has to decide what to do about them. The textbook describes three choices.
| Strategy | What the company does | When it fits |
|---|---|---|
| Undifferentiated | Sells the same product with the same marketing mix to the entire market | Large companies, or a new product whose buyers cannot yet tell alternatives apart |
| Concentrated | Serves one segment very well | Small companies in a competitive market |
| Differentiated | Makes a separate offer for each of several segments | Mature markets where buyers can compare products |
A company that chooses not to use segmentation is choosing an undifferentiated strategy. A concentrated strategy suits a resource-poor company that faces fierce competition. The cost is risk, because the company depends on one segment entirely.
Take the phone market again. A company choosing a concentrated strategy would build only the high-feature, high-price phone and aim all its marketing at that group. It becomes known as the best phone for that group. A company choosing a differentiated strategy would build three phones and tailor price, promotion and distribution for each. That needs more resources, and the textbook says it is suited to mature markets.
072 min
How segmentation shows up in tech
Software companies use segmentation in decisions that rarely carry the word. Each role makes a different one.
- Founder. Choosing the first group of customers to serve. A new company with few resources fits the concentrated strategy above. A minimum viable offer is built for one segment, not for everyone.
- Product manager. Deciding which requests to build. A feature that one large-company segment asks for may mean nothing to a small-team segment. The manager reads feedback by segment, not as one list.
- Designer. Deciding who the interface is for. An onboarding flow for experts and one for newcomers are different designs, and a designer who knows the segments can choose which to build first.
- Engineer. Building plans and permissions. Pricing tiers are the visible result of a segmentation: each tier is an offer for a group with different needs and budgets.
For example, a project-management tool for software teams might find two groups in its customer data. Small teams sign up on their own and want to start in minutes. Large organisations need access controls and a contract. The groups differ on both criteria the textbook names, the benefit sought and the company size, so the tool needs two offers, with different onboarding, different pricing and different sales effort.
Segments also change how value is counted. Customers in different segments often stay for different lengths of time and spend different amounts, so the lifetime value of a customer is usually calculated per segment, not as a single average.
081 min
How to use segmentation
Segmentation is a sequence in which each step needs the one before it. It is often taught together with targeting and positioning as the STP process. The steps below follow the textbook's logic.
Choose the criterion.
Decide what difference between buyers matters for the decision at hand. Nothing later works if this is wrong, because every group is formed from it.
Form the groups.
Either decide the groups in advance or let analysis produce them. This step needs the criterion and data about buyers on it.
Test the groups.
Check each group against the usability tests in the measurement section. Drop groups that fail. This step needs the groups from step 2.
Choose the target.
Pick one segment or several, using the undifferentiated, concentrated or differentiated strategy. Only tested groups can be compared.
Fit the offer.
Adapt product, price, promotion and distribution for the chosen segments. Segmentation assumes that different segments need different marketing programmes, meaning different offers, prices, promotions, distribution or a mix of these.
Positioning, the last part of STP, is the decision about how the target segment should see the product compared with alternatives. It comes after the target is chosen because a position is always a position for someone.
092 min
How to tell whether a segment is worth serving
A group of buyers is a segment for practical purposes only if it passes some tests. A segment is usable when it is identifiable, substantial, responsive, actionable and stable. The Wikipedia overview of the topic lists these five tests. Each can be turned into a question that a team can answer from its own data.
| Test | Question to answer | Where to look |
|---|---|---|
| Identifiable | Can we tell which customers belong to the group? | A field in the customer database, or a survey question |
| Substantial | Is the group big enough to be worth serving? | Its share of customers and its share of revenue |
| Responsive | Does the group react differently to an offer? | Compare conversion by group in an A/B test |
| Actionable | Can we change something for this group? | A named change to product, price, message or channel |
| Stable | Will the group still exist next year? | Whether members stay in the same group over several periods |
A simple number for the substantial test is segment revenue divided by total revenue for the same period. A group with a small share of customers but a large share of revenue may deserve more attention than its size suggests.
The responsive test is the one most often skipped. If two groups behave the same way when shown the same offer, they are one segment for the purpose of that decision, whatever the data says about their ages or locations.
102 min
Where the evidence is contested
Segmentation is widely used, but its critics make a serious objection: the groups may not exist in the data.
Dolnicar, Grün and Leisch, who wrote the textbook that teaches the method, state it directly. In real consumer data, distinct and well-separated segments rarely exist naturally. Analysts often assume that groups are already in the data and only need to be found. If they are not, the analysis creates groups by drawing lines through a continuous spread of buyers. Dubes and Jain made the matching point about statistical grouping in 1979. Dubes and Jain (1979) wrote that it is foolish to impose a clustering structure on data known to be random.
A second objection concerns brands. Hammond and colleagues (1996) found that buyers of specific brands do not have distinct profiles on descriptor variables. The Wikipedia summary of the criticism adds a related point. In competitive markets, segments rarely differ much in how they use brands. The same summary records one more objection. Segments are categories that marketers create for consumers, and consumers do not identify with them.
The textbook does not drop segmentation because of this. It describes the groups as naturally existing or artificially created, and it treats artificial groups as acceptable when they are useful. The practical position is to treat a segmentation as a decision tool and to judge it by whether it changes what the company does. If the groups do not call for different offers, the segmentation has not helped, however clean the groups look in the data.
111 min
Segmentation vs. nearby concepts
Segmentation is the first step of a longer process, and it is often confused with the steps and tools around it. The axis that separates them is what each one decides.
| Term | What it decides | Relation to segmentation |
|---|---|---|
| Segmentation | Which groups exist in the market | The subject of this article |
| Targeting | Which of those groups to serve | Comes after segmentation and needs its output |
| Positioning | How the chosen group should see the product compared with alternatives | Comes after targeting |
| Persona | A fictional individual who stands for a segment | A way to describe a segment, not a way to find one |
| Product differentiation | How the product differs from competitors | Smith's alternative to segmentation: change the product, not the market served |
On personas, the Wikipedia overview says it directly. Personas are imaginary representations of consumers that are used to stand for a segment. A persona is built after the segment is found. A segment is a group of real buyers, and a persona is a story about one typical member of it.
?6 questions
Questions people ask
What is segmentation in marketing?
What is an example of segmentation?
What are the main types of segmentation?
How do you know if a segment is worth targeting?
What is the difference between segmentation and targeting?
Does segmentation always work?
§5 sources
Sources
Show all 5 sourcesHide sources
Dolnicar, S., Grün, B. and Leisch, F. (2018). Market Segmentation Analysis: Understanding It, Doing It, and Making It Useful. Springer, open access (CC BY 4.0). Full text
Smith, W. R. (1956). Product Differentiation and Market Segmentation as Alternative Marketing Strategies. Journal of Marketing 21(1), 3-8. The original article, paywalled
Market segmentation. Wikipedia overview, with the criticisms section and the sources it cites
Ford Model T. Wikipedia, section on colours
Alfred P. Sloan. Wikipedia, section on the General Motors pricing structure





