What is customer segmentation?
Make the lessons yours
Tell us where you sit and we'll run every example through a world like yours. One tap, and you can change it whenever you like.
- Define customer segmentation and distinguish it from market segmentation
- Name the five model families and what each is based on
- Tell a segment from a persona, a cohort and a campaign audience
- Write your first three segment definitions from data you already hold
Customer segmentationCustomer segmentationDividing your existing customers into groups that share something meaningful, such as behaviour, value or circumstances, so each group can be treated differently. Distinct from market segmentation, which divides a whole market to shape strategy; customer segmentation shapes execution: who gets which message, offer and experience.View in glossary is the practice of dividing your existing customers into groups that share something meaningful (behaviour, value, needs or circumstances) so you can treat each group differently. The test of a segment is not how clever the grouping is. It is whether the group changes what you do.
It is not the same as market segmentation. Market segmentation divides a whole market, including people who have never heard of you, and mostly shapes strategy: which products to build, which audiences to enter. Customer segmentationCustomer segmentationDividing your existing customers into groups that share something meaningful, such as behaviour, value or circumstances, so each group can be treated differently. Distinct from market segmentation, which divides a whole market to shape strategy; customer segmentation shapes execution: who gets which message, offer and experience.View in glossary works on people you already have data about, and mostly shapes execution: who gets which message, offer, or experience. This pathway is about the second, though the muscles transfer.
The five model families
Every segmentation model you'll meet is a variation on five families. Later lessons take each in depth; here is the map, with a neutral example of each.
| Family | Groups customers by | Example |
|---|---|---|
| Demographic | Who they are | Customers in a given age band |
| Geographic | Where they are | Within 30 minutes of a store, for click-and-collect offers |
| Behavioural | What they do | Bought twice in 90 days; opens every email but never clicks |
| Value-based | What they're worth | Top 10% by yearly spend |
| Predictive | What they're likely to do | Likely seasonal buyers, based on last year's behaviour |
Real segments usually combine families. "High-spending customers near the shop who haven't visited since the café opened" touches three at once, and is more useful than any single ingredient.
What a segment is not
Three lookalikes cause most of the confusion.
A personaPersonaA fictional portrait of a typical customer, used to guide tone, creative and empathy. A persona is not a segment: you can't email a portrait. If a persona matters commercially, define the measurable segment that approximates it and check the two actually match.View in glossary is a portrait; a segment is a list. "Weekend gardener Wendy, 52, shops after payday" is a personaPersonaA fictional portrait of a typical customer, used to guide tone, creative and empathy. A persona is not a segment: you can't email a portrait. If a persona matters commercially, define the measurable segment that approximates it and check the two actually match.View in glossary: a fictional character that helps you write copy (this path uses the same trick with its own cast). But you cannot email Wendy. A segment is a defined group of real customer records you can count, reach and measure.
A cohortCohortA group fixed by a shared start event, most often first purchase in the same period. Membership never changes, which makes cohorts ideal for measuring how customer quality shifts over time, and the wrong tool for ongoing behaviour, which needs a segment.View in glossary is fixed by time; a segment is defined by traits. Everyone who first bought in January 2026 is a cohortCohortA group fixed by a shared start event, most often first purchase in the same period. Membership never changes, which makes cohorts ideal for measuring how customer quality shifts over time, and the wrong tool for ongoing behaviour, which needs a segment.View in glossary. Membership never changes, which is what makes cohorts good for measuring how customer quality shifts over time. A segment's membership changes as customers' behaviour does.
A campaign audience is a one-off selection; a segment persists. The list pulled for last month's sale was an audience. If it had a name, a definition and a reason to exist next month too, it would be a segment.
Matt's first three segments
After the May sale post-mortem, Matt writes three segment definitions. Every clause maps to a field his shop platform already holds:
- New customers: first purchase in the last 30 days. They need a welcome and a reason to come back, not a discount.
- Repeat customers: two or more purchases in the last 12 months. His future best customers; he'll stop bribing them and start serving them.
- Lapsing customers: bought before, nothing in six months. The cheapest revenue to win back, and the group his storewide blasts were slowly burning.
- New customersfirst purchase in the last 30 days~900
- Repeat customers2+ purchases in the last 12 months~4,200
- Lapsing customersno purchase in 6+ months~3,800
Three groups, three obviously different conversations. That instinct, formalised, is the whole discipline.
Which of these is a customer segment, as opposed to a persona or a cohort?
Key takeaways
- Customer segmentation groups your existing customers by shared traits or behaviour so each group can be treated differently.
- Customer segmentation shapes execution with people you already know; market segmentation shapes strategy for a whole market.
- All models belong to five families: demographic, geographic, behavioural, value-based and predictive. Strong segments combine them.
- A segment is a countable, reachable list of real customers defined by a rule. Personas, cohorts and one-off campaign audiences are different tools.
Common questions
A gym wants to win back "lapsed members". Which definition is ready to build as a segment?