Why FLYDE
Platform How it works Integrations Security ★ FLYDE Brain Core Customer 360 Market Trends Analytics
Solutions FLYDE Brain Omnichannel attributionIncrementality Omnichannel customer journey B2B + B2C identity Predictive models See all solutions Retail Automotive NGOs Banking Restaurants EnergyFitness
Resources Blog FLYDE Talks Whitepapers & ebooks Self-assessment ROI Calculator Glossary
Login
Home / Blog / Customer segmentation

Customer segmentation: types, examples and how to automate it

By Paco Herranz · November 17, 2026 · 9 min read

Customer segmentation means dividing the people who already buy from you into groups that behave alike, so you can give each group the offer, message and channel that work best for it. Done well, it is not a spreadsheet updated once a quarter: it is a system that moves each customer between segments as soon as their behaviour changes and triggers the right action without anyone exporting anything.

In this guide you will see the types of customer segmentation actually used in retail and ecommerce, concrete examples and how to automate it step by step. If you are looking for the theory, with the four classic types and consumer types, see our guide to market segmentation.

What is customer segmentation?

Segmenting customers means grouping your base by what you know about each person: what they buy, how often, how much they spend, through which channel, what they are worth and what they are likely to do. Unlike market segmentation, which works with studies and external data, here each segment is a real list of named customers you can send a campaign to today.

A useful customer segment meets three conditions: it can be calculated from your data, it has an action attached and it keeps itself up to date. If any of the three is missing, it stays on a slide.

Static and dynamic segments

The difference that changes results most is not which variables you use, but whether the segment updates itself.

Static segment
Dynamic segment
How it is created
A query or an export at a given moment
A rule that recalculates itself when the data changes
Who enters and leaves
Nobody: the list stays as it was
Each customer enters or leaves as soon as they meet or stop meeting the rule
What it is for
Control groups, prize draws, one-off campaigns
Always-on campaigns: welcome, reactivation, retention
The risk
It goes stale within weeks and ends up targeting people who no longer fit
None, if the rule is well defined and reviewed

Automating segmentation is mostly about moving from static to dynamic segments, and keeping static ones only for what really needs them, such as control groups.

Types of customer segmentation

With your customer data you can segment by six kinds of criteria. The most useful segments rarely use just one: a good segment combines two or three.

Basic data

Customer type (registered, buyer, subscriber), source, language, contact permissions. Mainly used to filter who can be contacted and through which channel.

Purchase behaviour

First and last purchase date, number of orders, average order value, products and categories, returns. The best predictor of the next purchase.

Value

Total revenue, margin and customer lifetime value. Tells you how much it is worth investing in each group.

Scores

RFM and your own scores that combine several signals into one number. They let you segment by trend, for example customers whose RFM has dropped since the previous period.

Predictive

Likelihood to buy, likelihood to churn, future value or next product. Segments by what will happen, not by what already happened.

Relationship and engagement

Email opens, browsing, support tickets, satisfaction, loyalty points. Adds the context purchases do not tell you.

Ten customer segment examples

These segments work in retail and ecommerce. Each one comes with its rule and the action it triggers:

New, no second purchase

First purchase 15 to 45 days ago and a single order

→ recommendation based on their first purchase.

Best customers

High RFM and value in the top 10%

→ early access and priority service, no discount.

Best customers cooling off

High RFM last period and low this period

→ personal contact before they leave.

At risk of churn

High churn probability in the next 60 days

→ retention campaign prioritised by value.

Recoverable dormant

No purchases for over six months but a good history

→ reactivation with new arrivals in their category.

Promotion-only

Every purchase made with a discount

→ deal alerts at the right moment and out of other campaigns.

Single-category buyers

Always buy from the same product family

→ cross-sell the complementary category.

Omnichannel vs single channel

Buy in store and online, or only in one

→ take store customers to their first online purchase.

Open issue

Unresolved return or complaint

→ exclude them from sales campaigns until it is resolved.

Points about to expire

Loyalty members with points expiring this month

→ reminder with a redemption suggestion.

How to automate customer segmentation step by step

01

Start with a goal, not with the data

One measurable sentence: win back best customers who are cooling off, get new customers to a second purchase or cut spending on unnecessary discounts.

02

State a hypothesis

Which group of customers, receiving which action, will move that goal. Without a hypothesis, the segment has no success criterion.

03

Build the rule with filters

Each condition narrows the group; watch how much each one cuts so you do not end up with a segment that is too small or too broad.

04

Keep a control group

Set aside 5% to 20% of the segment at random and freeze it. It is the only thing that will tell you whether the action worked.

05

Activate in your channels

Connect the segment to email, ad platforms or your sales team so that additions and removals arrive automatically, without exporting files.

06

Measure and review

Compare the segment with its control group and review each segment at least once a quarter: if it no longer behaves differently, merge it or drop it.

This is what a rule built as a funnel looks like, with the control group set aside at the end:

Illustrative example; not taken from any customer.

How many segments to create

Fewer than you think. A practical rule: as many segments as different actions you can sustain. If your team can run five always-on campaigns, five well-defined segments perform better than twenty nobody works on. Start with those that have the most direct impact on revenue, such as new customers without a second purchase and best customers cooling off, and add the next one once the previous is working.

Common customer segmentation mistakes

  • Segmenting with data that is not unified. If the same customer appears three times (store, web and email), they fall into different segments and get contradictory messages.
  • Overlapping segments with no priority. A customer can be both a "best customer" and "at risk". Decide which segment wins and exclude them from the others.
  • Segments with no action. If nobody knows what to do with a segment, it is not needed.
  • Not measuring against a control. Without a control group you do not know whether the campaign worked or those customers would have bought anyway.
  • Forgetting exclusions. Sometimes the most profitable segment is the one that should receive nothing this week.

Frequently asked questions about customer segmentation

What is customer segmentation? It is dividing your current customers into groups that behave alike, by purchases, value, scores or predictions, so each group gets a different offer, message and channel.

What types of customer segmentation are there? With your own data you can segment by basic data, purchase behaviour, value, scores such as RFM, predictions and relationship or engagement. The best segments combine two or three of these criteria.

What is the difference between customer and market segmentation? Market segmentation divides the whole market, including people who do not buy from you, using studies and external data. Customer segmentation divides those who already buy from you, using your own data, and each segment is a real list you can activate.

How do you automate customer segmentation? By defining each segment as a rule that recalculates itself when data changes, connecting it to your channels so additions and removals arrive without exports, and measuring each action against a control group.

How often should segments be updated? Dynamic segments update themselves with every purchase or interaction. What is worth reviewing every quarter is whether each segment still behaves differently and whether its action still works.

How FLYDE does it

In FLYDE, segments are built in Market as a funnel of filters on each customer's unified profile, with all their online and in-store purchases. You can filter by basic data, purchase behaviour, value, RFM and predictions such as churn probability or future value, and see at each step how many customers remain.

Audiences are dynamic by default: they recalculate when the data changes. For a control group, you freeze them and set aside the percentage you need. Each audience is sent to your channels, such as Meta Ads, Google Ads or your email tool, and syncs automatically. And if you prefer to start with a question, FLYDE Brain returns the segment and turns it into an audience.

If you want the segments with the most impact in retail, download the whitepaper "The best customer segments for retail".

Free whitepaper · Mexico edition 2026

The best customer segments for retail

The 8 highest-impact segmentation models, from RFM to predictive, with the signal, the KPI of success and the common mistake of each. Hot Sale and El Buen Fin as the margin laboratory.

Download the guide
Keep reading

More from the FLYDE blog

Marketing strategies, first-party data and AI for marketing teams. No fluff.

Shall we start?

Give us 15 minutes and see how much FLYDE can untangle your day-to-day.

Let's talk