Market segmentation: what it is, the four classic types, and why behavioral segmentation is gaining ground

Market segmentation is the practice of dividing a total market into smaller subgroups whose members share characteristics or behaviors relevant to a business decision. The idea sounds obvious today, but as a formal concept within the marketing discipline it has a specific date: 1956, when Wendell R. Smith published Product Differentiation and Market Segmentation as Alternative Marketing Strategies in the Journal of Marketing. Before that paper, companies thought in terms of mass markets and differentiated their products; afterward, the modern discipline we now call segmentation began to take shape.

Philip Kotler popularized and formalized it in Marketing Management (1967, now in its 16th edition and still a staple of any business school) by integrating it into what is known as the STP model: Segmentation, Targeting, Positioning. That sequence—segment the market, choose one or more priority segments, position the brand for those segments—is the foundation on which nearly all serious marketing strategies have been built for half a century.

Clarifying that context matters because the first persistent confusion is treating segmentation as if it were the same as targeting or positioning. It isn't. They are three distinct steps in the same chain.

Segmentation, targeting, and positioning: the operational difference

Segmentation describes the market: how it divides into meaningfully distinct groups. It's analytical. It gives you a map. It doesn't yet involve choosing.

Targeting is the decision: of the segments you've identified, which ones you're going to actively serve. It involves resources, priority, focus. More on this in targeting.

Positioning is how you build the perception of your brand in the mind of the chosen segment—which associations, which differentiators, which promise. More on this in digital positioning map.

A company with well-done segmentation but no targeting ends up talking to everyone at once (which is talking to no one). A company with targeting but no positioning makes clear offers to a specific audience but gives no reason to choose it. The three layers work in a chain, and the real value appears when all three are written down, defended, and applied.

The four classic bases of segmentation

Any marketing course teaches the same four bases on which a market can be segmented. The taxonomy isn't arbitrary—it reflects four genuinely distinct ways of grouping audiences, with different operational implications.

Demographic

Age, gender, income, occupation, education level, marital status, family size. In B2B, their firmographic equivalents: company size, industry, revenue, number of employees, headquarters location.

It's the most historically used base because it was the easiest to obtain: censuses, surveys, forms. Its weakness has been known for decades—two demographically identical people can behave completely differently. A 35-year-old woman with a €50,000 income could be buying her first house or her third trip of the year; the demographic data doesn't tell you which.

In the era before the privacy shift, demographic segmentation worked well operationally because advertising platforms allowed you to target it precisely. That has changed.

Geographic

Country, region, city, postal code, climate, population density. Useful when the product, distribution, or local culture strongly shape the offering. A winter footwear brand has little to say to Seville and a lot to say to Asturias; a fast-delivery concept thrives in metro areas and dies in small towns.

It remains a solid base when the business depends on physical presence or cultural adaptation. For many global digital services, its weight has fallen.

Psychographic

Values, interests, lifestyle, personality, opinions. It goes beyond who you are categorically to reach how you think. A user interested in sustainability will react differently to a fast-fashion brand than to a brand with a transparent supply chain, even if they're demographically identical.

Psychographics has historically been hard to measure rigorously—it relies on long surveys, inference, ethnography. Frameworks like VALS (Values, Attitudes and Lifestyles), created by SRI International in the 1970s, tried to systematize it. They have worked better as mental maps than as pure targeting tools.

Behavioral

How the user uses the product, how often, in what context, what stage of the relationship they're in, what they do when they enter, what they avoid. Real behavior, not declared attributes.

This is the base that has gained the most ground since 2022, and it's worth explaining why.

Why behavioral segmentation is gaining weight

Three converging forces have shifted the balance:

The first is the progressive death of third-party tracking. Safari and Firefox blocked third-party cookies years ago; Chrome has entered its gradual phase-out. iOS App Tracking Transparency, since 2021, has reduced the viability of cross-app demographic targeting. The operational consequence: companies that relied on demographic targeting purchased from the platforms have seen their precision drop significantly.

The second is that first-party behavioral data—the kind you observe in your product, website, or owned channels—remains fully accessible. What the user does when they interact with you is something you know directly. That makes behavior the richest segmentation base among those that have survived the regulatory wave.

The third is that AI models applied to clustering have improved dramatically. Identifying groups of users with similar behavior patterns—without having defined the groups beforehand—has become operationally accessible to mid-sized companies, not just to giants with data science teams. The cluster emerges from the data; it isn't imposed by a demographic hypothesis.

The practical result in 2026 is that the companies that segment best usually start from real behavior (what they did in my product, what they read, what they searched for, what stage they're in), enrich it with psychographic data when they can honestly infer it, and leave demographic data as additional context rather than as the main base.

The real test of a useful segment

The distinction between segmentation that works and decorative segmentation is operational, not theoretical. One simple question resolves most cases:

Do I change the message, channel, product, price, or value proposition for this segment?

If the answer is no to all five, that "segment" doesn't exist operationally. It's an audience description, which is different.

In consulting engagements, I've seen teams show up with detailed profiles of six personas—"María, 32, lives in Barcelona, enjoys yoga"—and then produce generic content for everyone alike. That segmentation doesn't affect any real decision. It functions as a strategic ritual, not as a tool. The segmentation that matters is the one reflected in at least one of five places: the landing page copy, the channel where it's promoted, a product variant, a pricing tier, or a differentiated value proposition.

The New Coke case (1985): an example of incomplete segmentation

In 1985, Coca-Cola launched New Coke after one of the most expensive market research processes in its history: 200,000 taste tests showed that consumers preferred the new formula to Pepsi and to the original Coca-Cola. On paper, the segmentation seemed clear: there was a taste-sensitive segment that preferred sweeter notes, and the new formula won them over.

What segmentation by taste preference failed to capture was the emotional and identity dimension of the brand. Coca-Cola wasn't just a drink; it was a cultural symbol with deep emotional loyalty. The backlash was immediate—organized protests, thousands of letters, jammed phone lines. Seventy-seven days after launch, the company brought back the original formula as "Coca-Cola Classic."

The lesson taught in business schools ever since: a segmentation that captures one real dimension (flavor preference) but ignores another equally real dimension (brand identity) can lead to technically justified and commercially disastrous decisions. Useful segmentation usually requires several crossed dimensions, not just one.

How to do actionable segmentation in practice

For a mid-sized company without dedicated data science teams, there's an order that works:

Start with your own data. Before hypotheses, look at what your CRM, your analytics, your support tickets, and your sales calls say. There are almost always segments visible in the data that the team had never named.

Identify behavior patterns that predict value. What do customers who renew do in their first 14 days versus those who don't? What pages do leads who convert visit versus those who don't? Those patterns are more useful segmentation material than any fictional persona.

Cross-reference with accessible contextual data. Industry, size, geography, lifecycle stage, acquisition source. It adds nuance to the behavioral pattern without depending on tracking that's no longer possible.

Test with specific campaigns and content before committing product architecture. A landing page that varies by segment, an email sequence by pattern, a BOFU piece by situation—that's the cheap test. If the segments respond differently, they're real. If they respond the same, they weren't segments.

And then, the most mundane part: review every quarter whether the segments still predict what they used to predict. Segmentation ages. Markets change, products change, audiences turn over. A two-year-old segmentation may be describing a market that no longer exists.

Segmentation and creative operations

The reason so many segmentations stay in PowerPoint and never reach production is operational, not strategic. A living segmentation demands producing different pieces for different segments: copy variants, cases by industry, demos by size, cadences by stage. If that production is done in silos without coordination, the cost becomes unsustainable and the segments revert to theory.

That's why segmentation connects directly to creative operations: the editorial calendar decides which pieces are produced for which segment, content production coordinates the variants so each segment has consistent material, and creative KPIs let you measure whether the segments actually respond differently to the pieces designed for them.

At Polimake, that coordination has its own place: Studio to schedule campaigns and pieces by segment, Studio to produce variants with a consistent brand system, Media as a repository where each segment has its specific material accessible—so the second campaign aimed at that segment doesn't start from the same point as the first.

---

If you work in marketing, strategy, or product and you've landed here looking for a clear answer about what market segmentation is, the most useful thing you can take away from this article is probably the operational test: segmentation that works is the one that changes at least one concrete decision. The rest is audience description dressed up as method.

To complement this, the empathy map covers how to build deep knowledge of a specific segment, market research techniques covers how to gather the evidence that supports segmentation, and market segmentation criteria goes deeper into specific variables beyond the four classic bases.

Quick references

• Targeting — the step after segmenting.
• Digital positioning map — the third step of the STP model.
• Empathy map — to go deeper into a specific segment.
• Market research — the foundation on which honest segmentations are built.
• Market research techniques — practical methods for obtaining the data.
• Market segmentation criteria — complementary variables and approaches.