Customer Segmentation: What It Is and How to Start

By Alva Chew · Updated 2026-11-06

Definition Customer segmentation means splitting your customers into groups that buy in similar ways, so you can treat each group differently. For a small shop the most practical split is by behaviour: how recently, how often and how much each person buys. That needs purchases tied to named customers, which a points program provides.

What is customer segmentation?

Customer segmentation is the practice of dividing customers into groups whose members buy in similar ways. You then decide what to do for each group, such as how to thank your best regulars or when to reach out to people who have gone quiet.

The idea has a long history in marketing. Wendell R. Smith’s article “Product Differentiation and Market Segmentation as Alternative Marketing Strategies” appeared in the Journal of Marketing, volume 21, issue 1, in July 1956. The term has been in use for close to 70 years, and the core step has not changed: group first, then act.

What are the ways to segment customers?

Basis	Example groups	Data you need
Behaviour	Weekly regulars, monthly visitors, lapsed	Visit dates and spend per customer
Product	Coffee-only buyers, food buyers	What each customer orders
Demographic	Students, office workers	Age or job, which you usually have to ask for
Location	Walk-in neighbours, commuters	Postcode or outlet visited

For an owner of one or a few outlets, behaviour is the strongest starting point. You already see it at the counter, and it needs no extra questions that members may skip. Demographic splits require data customers have to hand over, and many will not.

How do you segment customers by behaviour?

Use the three numbers behind RFM analysis: how recently a customer bought, how often, and how much they spend per order. Fader, Hardie and Lee describe grouping customers on these characteristics as “traditional RFM segmentation analyses” used for target marketing.

A simple version with three questions:

Pull one row per member with last visit date, visits in the past 12 months and total spend.
Mark who visited in the last 30 days, 31 to 90 days and over 90 days.
Within each, split by high and low visit count.

What does a segmented customer list look like?

These members are invented to show the idea. The period is the last 12 months and today is 6 November.

Segment	Rule	Members	Share of members	Share of spend
Core regulars	Visited in last 30 days, 20+ visits	60	12%	41%
Occasional	Visited in last 90 days, under 20 visits	190	38%	34%
Drifting	Last visit 91 to 180 days ago	110	22%	17%
Lapsed	Last visit over 180 days ago	140	28%	8%

The numbers are made up, not benchmarks. What matters is the shape of the exercise. A small group usually carries a large share of spend, and a large group at the bottom has stopped buying. Each row suggests a different action, which is the test of a useful segment.

What do you do with each segment?

Core regulars: keep redemption quick at the counter and make sure they can see their balance. Your attention here protects the most spend.
Occasional: a reachable reward threshold gives them a reason to visit once more. See purchase frequency.
Drifting: the best group to contact first. Our guide to how you win back lapsed customers covers it.
Lapsed: low priority. Spend little effort and track whether they return on their own.

What are the limits of segmentation?

Segments describe the past. Fader, Hardie and Lee studied 23,570 first-time buyers of the online retailer CDNOW across 78 weeks, split into two 39-week periods. In that data, a customer’s inactive status is not directly observed, so someone who stops buying may have left for good or may simply be between purchases.

Finding	Detail
Customers studied	23,570 first-time CDNOW buyers
Data window	78 weeks, two 39-week periods
Limit	Whether a quiet customer has become inactive is unobserved

Source: Fader, Hardie and Lee, “RFM and CLV: Using Iso-Value Curves for Customer Base Analysis”, Journal of Marketing Research 42(4), 2005. One online music retailer, not a café or shop.

Three practical limits follow. Small customer lists give thin segments, so treat group lines as rough. Segments built only on identified members leave out cash buyers who never sign up. And a segment is only useful if you act on it, so for forward-looking value see customer lifetime value and track your repeat purchase rate over time.

What data do you need to segment?

Every purchase must be linked to one person. A till records sales, not people. A points-based loyalty program creates that link, because each member has an account and earns points in proportion to what they spend. Report segment results as covering identified members only.

Segmenting with Gimify

Gimify members join by scanning a QR code and entering an email, and each member has one points balance across every outlet. The free plan covers one outlet with unlimited members, so you can build a named member list now and draw your own segment lines from it.

Questions owners ask

What is customer segmentation in simple terms?

It is sorting customers into groups that behave alike, such as weekday regulars, weekend big spenders and people who have stopped coming. Each group gets its own approach instead of one offer for everyone.

What are the main ways to segment customers?

Common bases are demographics (age, location), behaviour (visits, spend, last purchase) and the product bought. A small shop with a points program already holds behaviour data, so that is the easiest place to begin.

How many segments should a small business use?

Two to four. Each segment needs enough customers to be worth a separate action, and you need to be able to name what you will do for it. If you cannot, merge it into a neighbour.

Do I need software to segment customers?

No. A spreadsheet with one row per member, their last visit date, number of visits and total spend is enough. Sort by each column and draw your group lines by hand.

Is customer segmentation the same as RFM analysis?

RFM is one way to segment, using recency, frequency and monetary value. Segmentation is the wider idea, and the groups can also be built on other facts such as location or what people order.