Market Segmentation Research: Stop Guessing Who Your Customers Are

Market segmentation research is the process of dividing a broad customer base into distinct groups based on shared characteristics, behaviours, or needs, so that marketing and product decisions can be targeted rather than generic. Done well, it replaces assumption with evidence and gives commercial teams a defensible basis for where to focus, what to say, and who to prioritise.

Most businesses think they know their segments. Most are working from a combination of gut feel, legacy personas built years ago, and data that describes what customers have done rather than why. That gap between assumption and evidence is where budgets get wasted.

Why Most Segmentation Work Produces Shelved Decks

I have sat in enough agency and client-side strategy sessions to know how segmentation projects typically end. A research agency runs a survey, clusters the data into five or six segments, gives each one a name and a stock photo, and presents a deck that everyone agrees is interesting. Six months later, the deck is on a shared drive and nobody can remember which segment the business was supposed to be targeting.

The problem is not the research. The problem is that segmentation is treated as an output rather than an input. It gets commissioned as a standalone project, disconnected from the commercial question it was supposed to answer. Good segmentation research starts with the decision, not the data. Before you design a single survey question, you need to know what you are going to do differently depending on what you find.

If you cannot name the commercial decision that this research will inform, you are not ready to commission it. That sounds obvious. It is not widely practised.

What Are the Main Types of Market Segmentation?

There are four segmentation approaches that come up consistently, and they are not mutually exclusive. Most strong segmentation work draws on more than one.

Demographic segmentation groups customers by age, gender, income, education, occupation, or household structure. It is the most commonly used and the least interesting on its own. Demographics tell you who someone is in census terms. They rarely tell you why they buy, what they value, or how to reach them in a way that resonates. Two people with identical demographic profiles can have completely different purchasing behaviour and brand relationships.

Geographic segmentation divides markets by location, from country level down to postcode. For businesses where location drives meaningful differences in need, regulation, or behaviour, this is genuinely useful. For most digital businesses, it is a starting point rather than a destination.

Psychographic segmentation groups customers by values, attitudes, interests, and lifestyle. This is where segmentation starts to get commercially interesting, because it gets closer to motivation. If you understand what a customer believes and what they are trying to achieve, you can build messaging and product positioning that actually connects. The challenge is that psychographic data is harder to collect and harder to activate than demographic data.

Behavioural segmentation uses purchase history, usage patterns, brand interactions, and loyalty signals to group customers. This is often the most actionable because it is built from what people actually do rather than what they say they do. The limitation is that it is retrospective. It tells you how customers have behaved, not necessarily how they will behave or what would change their behaviour.

The segmentation frameworks that hold up in practice tend to combine behavioural data with attitudinal research. One tells you what is happening, the other tells you why. Together, they give you something you can actually build strategy on.

If you want a broader view of how segmentation fits into the research toolkit, the Market Research and Competitive Intel hub covers the full landscape, from customer insight to competitive positioning.

How Do You Design Segmentation Research That Actually Works?

The design phase is where most segmentation projects go wrong, and it usually comes down to one of three mistakes: the wrong sample, the wrong questions, or the wrong analytical approach.

Start with qualitative research. Before you run a quantitative survey, spend time with real customers. Not focus groups where people perform for each other, but individual depth interviews where you can probe properly. You are not trying to validate hypotheses at this stage. You are trying to understand the language customers use, the problems they are trying to solve, and the tensions that exist in their relationship with your category. This feeds directly into the quantitative instrument and makes it far more likely that your survey questions are capturing something real.

Design for discrimination, not description. A segmentation survey that asks people to rate their satisfaction with your brand is a brand tracker, not a segmentation study. The questions that produce useful segments are the ones that reveal meaningful differences in attitude, need, or behaviour. Ask about trade-offs, priorities, and frustrations. Ask what good looks like. Ask what would make them switch. These questions create the variation in the data that cluster analysis needs to find genuine groups.

Sample from the right universe. This is more contested than it sounds. Are you segmenting your existing customers, your category users, or the total addressable market? Each gives you a different answer. If you only survey your existing customers, you will build segments that reflect who you have already attracted, not necessarily who you could attract. If you survey the total market, your segments will include people who are genuinely unreachable for your brand at this point in time. Neither is wrong, but they answer different questions, and you need to know which question you are answering.

Be honest about sample size. Segmentation analysis needs enough respondents in each segment to be statistically meaningful. If you end up with a segment that represents 8% of your sample and your total sample was 400, you are drawing commercial conclusions from 32 people. That is not a foundation for strategy.

What Does Good Cluster Analysis Actually Look Like?

Cluster analysis is the statistical technique most commonly used to identify segments from survey data. It groups respondents based on similarity across a set of variables, and it can produce genuinely useful results when the input data is well-designed and the analyst knows what they are doing.

It can also produce results that look mathematically clean but are commercially useless. Segments that are statistically distinct but behaviourally indistinguishable. Segments that make sense in a presentation but cannot be identified in your CRM or reached through your media channels.

There are a few things worth checking when you are evaluating segmentation output. First, are the segments meaningfully different from each other on the dimensions that matter commercially? Not just statistically different, but different in ways that would lead you to make different decisions. Second, can you identify which segment a given customer belongs to without running a full survey? If the answer is no, the segmentation may be analytically interesting but operationally useless. Third, is the size of each segment large enough to justify a distinct strategy? A segment that represents 3% of your market is probably not worth a dedicated channel plan.

I have seen segmentation work where the agency presented seven segments and the client, quite reasonably, asked which three they should focus on. The agency had no good answer because they had not been briefed to think about prioritisation. The research answered the wrong question. Good segmentation research is designed with prioritisation in mind from the start.

How Do You Prioritise Segments Once You Have Them?

Segmentation gives you a map. Prioritisation tells you which roads to take. The two most useful dimensions for prioritisation are attractiveness and fit.

Attractiveness is a function of segment size, growth trajectory, purchase frequency, average transaction value, and competitive intensity. A segment can be large and still be unattractive if it is shrinking, commoditised, or dominated by a competitor with structural advantages you cannot match.

Fit is about whether your brand, product, and capabilities give you a genuine right to win in that segment. This is where honest self-assessment matters. When I was running an agency and we were evaluating which client sectors to pursue, the temptation was always to chase the biggest opportunity. The smarter question was always where we had a real advantage over the alternatives. The answer was usually a smaller set of sectors than the initial list suggested, and focusing on those produced better outcomes than spreading across everything that looked attractive.

A simple two-by-two of attractiveness against fit gives you a starting framework. Segments that score high on both are your primary focus. Segments that score high on attractiveness but low on fit are worth understanding, but probably not worth investing in unless you can credibly close the capability gap. Segments that score low on both are a distraction.

Tools like behavioural analysis platforms can help you cross-reference your segmentation hypotheses against actual on-site behaviour, particularly useful when you are trying to validate whether a segment you have identified in research maps onto a group you can actually observe and reach.

How Do You Activate Segmentation Across the Business?

Activation is where segmentation projects most commonly fail. The research is done, the segments are defined, and then nothing changes. The media plan looks the same. The messaging looks the same. The product roadmap looks the same. The segmentation sits in a deck.

For segmentation to drive commercial outcomes, it needs to change decisions. That means connecting it to specific workstreams: the brief for a creative campaign, the targeting logic for a paid media plan, the product prioritisation process, the pricing architecture, the content strategy. Each of these should look different once you have a clear segmentation framework in place.

One of the practical challenges is that most CRM and media platforms organise data by demographic or behavioural signals, not by the attitudinal segments that research produces. Bridging that gap requires profiling work: identifying the observable signals, whether demographic, behavioural, or contextual, that predict segment membership. This is not always perfect, but it does not need to be perfect. It needs to be better than the alternative, which is treating everyone the same.

Forrester has written about the challenge of translating insight into organisational action, and the pattern they describe, research that informs rather than transforms, is consistent with what I have seen across multiple businesses and sectors. The organisations that get value from segmentation are the ones that treat it as an operating framework, not a research project.

There is also a content dimension worth thinking about. Once you know which segments you are targeting and what matters to each, your content strategy should reflect that. Lead generation content that speaks to a specific segment’s problem will consistently outperform generic content that tries to be relevant to everyone.

When Should You Revisit Your Segmentation?

Segmentation is not a one-time exercise. Markets change, customer needs evolve, new competitors reshape category dynamics, and the assumptions that underpinned your original segmentation become stale. The question is when to revisit it and how to know when it is no longer serving you.

There are a few trigger events that should prompt a review. A significant shift in market conditions, whether economic, regulatory, or competitive, is one. A meaningful change in your product or service offer is another. If you have expanded into new channels or geographies, your original segmentation may not travel. And if your commercial performance is diverging from what the segmentation predicted, that is a signal worth investigating.

The mistake I see most often is the opposite: businesses that treat segmentation as a fixed truth and resist revisiting it even when the evidence suggests it is no longer accurate. I understand the reluctance. Segmentation projects are expensive and time-consuming, and there is an organisational inertia around frameworks that have been embedded in strategy documents and presentations. But working from an outdated segmentation is not neutral. It actively distorts decisions.

A lighter-touch review, using behavioural data and qualitative customer conversations, can often tell you whether your core segmentation is still holding without commissioning a full research programme. Reserve the full programme for when the evidence suggests something has fundamentally shifted.

For more on how segmentation connects to the broader research and intelligence picture, including how to use competitive data alongside customer insight, the Market Research and Competitive Intel hub is worth working through systematically.

What Does Segmentation Research Look Like in Practice?

When I was at iProspect and we were growing the agency from around 20 people to closer to 100, one of the things that became clear was that not all clients were equal, and not all prospects were worth pursuing. We were effectively doing informal segmentation on our own market: which types of clients valued what we were good at, which sectors had the budget cycles that suited our model, which decision-makers were the right fit for how we worked.

We did not call it segmentation research at the time. But the process of getting specific about who we were for, and being willing to walk away from opportunities that did not fit, was what allowed us to build a proposition that was coherent rather than generic. The agencies that try to be everything to everyone tend to be nothing to anyone in particular.

The same logic applies to any business running segmentation research. The goal is not to describe your market in exhaustive detail. The goal is to identify where you have a genuine right to win, and then to concentrate your resources there with enough conviction to actually win.

Early in my career, when I was still learning what good marketing looked like, I noticed that the campaigns which performed best were rarely the ones with the broadest targeting. They were the ones where someone had made a clear decision about who they were talking to and had built everything, the message, the channel, the timing, around that specific person. That instinct has held up across 20 years and 30 industries. Specificity outperforms breadth more often than not.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions