Customer segmentation is the practice of dividing your audience into distinct groups that share common traits so you can send more relevant messages to each. The four core types are demographic, behavioural, psychographic, and value-based segmentation. Done well, it lifts email revenue 3-10x versus unsegmented sends.
The message that converts one segment repels another. Segmentation is how modern marketers stop shouting into a crowded room and start speaking directly to the person they can actually help.
What is customer segmentation?
Customer segmentation is the process of splitting a customer or prospect base into groups that share meaningful attributes — behaviour, needs, demographics, revenue contribution — so marketing, product, and sales teams can address each group with something specifically useful. It is the operational foundation of every lifecycle program, personalisation engine, and paid-media targeting strategy.
The rule of thumb: a segment is only useful if you would send it a different message. If two "segments" get the same email, they are the same segment.
Klaviyo's 2024 benchmarks show segmented flows generate 3-10x more revenue per send than unsegmented broadcasts. Every mature CDP vendor benchmarks 20-45% lifts on lifecycle flows when segmentation is added. It is the single highest-ROI CRM upgrade most teams can make.
Segmentation is the input other tactics depend on. Email list segmentation applies it to sends, personalization applies it to on-site experience, and a buyer persona turns a segment into a description a writer can actually work from.
Why customer segmentation matters
- Revenue lift. Segmented email campaigns generate 3-10x more revenue per send. Paid campaigns targeted at behavioural segments outperform lookalike-only targeting by 40-60%.
- Smarter spend. Segmentation identifies the audiences most likely to convert, letting you concentrate budget where CAC is lowest.
- Product signal. Behaviour segments show which features resonate with which cohorts, feeding roadmap decisions with real evidence.
- Retention gains. Personalised messaging built on segmentation lowers churn — customers stay when messages feel targeted, not templated.
How to build a segmentation model
- Pick the outcome first. Segmentation for retention looks different from segmentation for acquisition or expansion.
- List available data. CRM, product, ecommerce, ad platforms, support. The data you already have shapes what's possible.
- Choose 3-5 dimensions. Start small. Lifecycle stage + product usage + plan tier is a common opening set.
- Define each segment as a query, not a list. Static lists rot; queries recompute as behaviour changes.
- Design different messaging per segment. If the copy doesn't change, the segment isn't real.
- Measure per segment. Compare open rate, CTR, revenue per recipient, and downstream retention.
Four types of customer segmentation
| Type | Sample attributes | Best used for | Data source |
|---|---|---|---|
| Behavioural | Purchase recency, feature use, session frequency | Lifecycle marketing, retention, expansion | Product + email engagement data |
| Demographic (B2C) / Firmographic (B2B) | Age, income, location · industry, company size, revenue | Ad targeting, ICP definition | CRM, ad platforms, enrichment tools |
| Psychographic | Values, interests, motivations, lifestyle | Brand positioning, creative strategy | Surveys, community signals, first-party research |
| Value-based | CLV tier, top 20% by spend, at-risk revenue | Retention, VIP programs, upsell prioritisation | Order data + CLV model |
Real customer segmentation examples
1. Ecommerce recency segmentation
An online retailer segmented its email list by purchase recency (0-30 days, 31-90 days, 91-180 days, 180+ days). Each segment received a different message. Revenue per email rose 45% versus the previous unsegmented broadcast.
2. B2B content track segmentation
A marketing agency split its content into two tracks — one for reseller agencies, one for in-house teams — and segmented email nurture by job title. Organic traffic grew 60% and demo booking rate on the reseller track outpaced the previous blended track by 2.4x.
3. SaaS value-based upsell segmentation
A SaaS product segmented users by feature-adoption depth. High-adoption free users received an upsell to premium tier; low-adoption users received an activation campaign instead. Upsell conversion rate lifted from 3.2% to 8.1% across a quarter.
Segmentation vs personalisation — how they differ
Segmentation is how you group. Personalisation is what you do inside each group.
Segmentation
- Grouping decision made at strategy layer
- Same message across everyone in a segment
- Foundational — required for personalisation
- Runs on rules or clustering models
- Measured by segment-level performance
Personalisation
- Individual-level content decisions
- Different message per person inside a segment
- Sits on top of segmentation infrastructure
- Often model-driven or generative
- Measured at individual send performance
6 best practices for customer segmentation
- Start with 3-5 actionable segments. More granular is only better if you can act on it.
- Favour behaviour over demographics. What people do is far more predictive than what they look like on paper.
- Make segments dynamic. Query-based memberships update as behaviour shifts — static lists decay in weeks.
- Own the definition centrally. One shared segmentation model across marketing, product, and sales — not siloed definitions per team.
- Test different content, not different subject lines. Real segmentation changes the offer, the proof, and the CTA — not just the salutation.
- Retire dead segments. If a segment hasn't gotten a unique message in a quarter, delete it.
15 segments feels sophisticated. In practice it produces a graveyard of empty flows and no clear owner. Fewer, sharper segments outperform more granular ones every time. Segment for action, not for the org chart.
Common segmentation mistakes to avoid
- Segments no one owns. If a segment isn't owned, it isn't managed.
- Segments defined once, never updated. Customer behaviour moves; definitions should too.
- Confusing personas with segments. Personas are qualitative sketches; segments are quantitative slices.
- Treating segmentation as an email problem. The same segments should power paid, product, sales, and CS.
- Ignoring value-based segmentation. The top 20% of customers by CLV usually deserves its own strategy.
How theStacc helps with customer segmentation
theStacc feeds segmentation with intent data — the searches, comparison pages, and reviews your customers actually engage with. That behavioural signal often reveals more useful segments than demographic data alone (like "customers researching switching" or "high-intent buyers of feature X"), and drops straight into your CDP or ESP as a segmentation input.
Frequently asked questions
Start with 3-5 actionable segments. More granular slicing is only useful once you have the operational capacity to send different messages to each segment. A segment you can't act on differently is a distraction.
The bare minimum is purchase history, basic engagement data (opens, clicks, sessions), and either firmographics (B2B) or demographics (B2C). Behavioural data is more predictive than demographic data for most marketing decisions.
Review segment definitions quarterly. Refresh membership continuously — most segments should be dynamic queries that recompute as customer behaviour changes, not static lists exported once.
Demographic (age, income, job title, company size, location), behavioural (purchase history, feature use, engagement), psychographic (values, attitudes, lifestyle), and value-based (revenue contribution, CLV tier, top 20% by spend).
Yes. Klaviyo and Mailchimp both report segmented email campaigns generating 3-10x more revenue per send than unsegmented blasts, and every mature CDP vendor benchmarks lifts of 20-45% on properly segmented flows.
Related glossary terms
Sources
