Segmentation
Also: Customer segmentation, Market segmentation, Segmenting, Segments, Segmentation (FR), Segmentation client
Dividing a market into distinct groups of customers who share similar needs, characteristics or behaviours, so each group can be served with a tailored approach.
What it is
Segmentation is the practice of splitting a broad market, customer base, or dataset into smaller groups whose members are similar to each other and different from members of other groups. Each group (a segment) is defined by shared attributes such as needs, behaviours, demographics, firmographics, value, or lifecycle stage.
Segmentation is not the same as a single sort or filter. A good segmentation is:
- Distinct: segments do not overlap and cover the whole population.
- Measurable: you can size each segment and count its members.
- Actionable: you can actually reach and treat each segment differently.
- Stable enough to plan against, yet refreshed as behaviour shifts.
Why it matters
Treating every customer the same wastes money and misses opportunity. Segmentation lets an organisation focus resources where returns are highest and personalise messaging, pricing, product, and service. It turns an undifferentiated "the market" into a set of concrete targets that teams can prioritise and measure.
Common segmentation bases include:
- Demographic / firmographic: age, income, industry, company size.
- Geographic: region, climate, urban versus rural.
- Behavioural: purchase frequency, usage, loyalty, channel preference.
- Psychographic: attitudes, values, motivations.
- Value based: revenue, margin, or lifetime value contribution.
How it is used in practice
1. Choose the objective (acquisition, retention, pricing, product roadmap).
2. Select variables relevant to that objective.
3. Build segments using rules (business logic) or unsupervised methods (for example k-means clustering, RFM scoring).
4. Profile each segment: size, value, needs, and how to reach it.
5. Activate: assign owners, tailor offers, and route through the right channels.
6. Measure and refresh as segments drift over time.
Worked example
A subscription software company has 50,000 accounts. It scores each on spend and product usage, producing four segments:
- Champions (high spend, high usage): protect and upsell.
- At risk (high spend, low usage): trigger onboarding and support.
- Growers (low spend, high usage): offer upgrade paths.
- Low touch (low spend, low usage): serve with automation only.
Marketing sends different campaigns to each, finance forecasts churn risk per segment, and the data team maintains the pipeline that recomputes membership monthly. The same four groups now anchor budgets, targets, and product decisions.
See also
Frequently asked questions
What is segmentation in marketing and data?
Segmentation is the practice of splitting a market, customer base or dataset into smaller groups whose members resemble each other and differ from other groups. Each segment is defined by shared attributes: needs, behaviours, demographics, firmographics, value or lifecycle stage. The point is to stop treating every customer the same and allocate budget where returns are highest.
What is the difference between a segmentation and a simple filter on a customer list?
A filter isolates one subset and ignores the rest; a segmentation covers the whole population with groups that do not overlap. It also has to be measurable (you can size each segment), actionable (you can reach and treat each group differently) and stable enough to plan against while being refreshed as behaviour shifts. A sort that fails those tests is a query, not a segmentation.
Which segmentation bases should I choose?
Start from the objective, then pick the variables that serve it. The usual bases are demographic or firmographic (age, income, industry, company size), geographic, behavioural (purchase frequency, usage, loyalty, channel preference), psychographic (attitudes, values, motivations) and value based (revenue, margin, lifetime value). A pricing question and a retention question rarely call for the same variables.
Where do I start if I have never segmented my customer base?
Fix the objective first: acquisition, retention, pricing or product roadmap. Then select the relevant variables, build the groups with business rules or unsupervised methods such as k-means clustering or RFM scoring, profile each segment (size, value, needs, how to reach it), activate by assigning owners and tailoring offers, and measure and refresh as segments drift. Skipping the objective is what produces segmentations nobody uses.
What does a value based segmentation look like in practice?
Take a subscription software company with 50,000 accounts that scores each one on spend and product usage, producing four segments: Champions (high spend, high usage) to protect and upsell, At risk (high spend, low usage) to trigger onboarding and support, Growers (low spend, high usage) to offer upgrade paths, and Low touch (low spend, low usage) served by automation only. Marketing sends different campaigns to each, finance forecasts churn risk per segment, and the data team recomputes membership monthly. The same four groups then anchor budgets, targets and product decisions.