Guide
Customer segmentation, methods beyond demographics
How to segment customers for research and product decisions, beyond demographics, into behavior, JTBD, and lifecycle. With examples and pitfalls.
Customer segmentation means grouping customers by something that meaningfully changes how they behave, buy, or get value from your product. If a segment does not change a decision you make, it is not a segment, it is a label.
Most teams over-rely on demographics and firmographics because they sit in the CRM already. The richer lenses (behavioral, lifecycle, jobs-to-be-done) take more work but pay back in roadmap calls, pricing tests, and lifecycle email that actually performs. The five lenses below are the ones worth knowing.
The five segmentation lenses#
Demographic
Age, gender, income, education, geography. Useful for B2C reach and ad targeting. Rarely sufficient for product decisions on its own.
Firmographic (B2B)
Company size, industry, revenue, tech stack, region. The B2B equivalent of demographics. Necessary for ICP work, not enough by itself.
Behavioral
What people actually do in the product: feature usage, frequency, depth, recency. The strongest predictor of retention and expansion.
Lifecycle
Where the customer sits in their journey with you: prospect, trial, activated, retained, churned, won-back. The backbone of CRM and lifecycle marketing.
Jobs-to-be-Done
The progress the customer is trying to make and the situation that triggered them. Strategically the most useful, operationally the most often skipped.
Pick two or three lenses and combine them. A behavioral segment crossed with a JTBD segment usually beats any single dimension on its own.
Why JTBD-based segmentation often beats demographics#
Demographic and firmographic segments describe what a customer looks like. JTBD segments describe what they are trying to get done. The second one predicts behavior, the first one mostly predicts the marketing channel.
If a SaaS team segments only by company size, they miss the fact that the people actually using the product are similar across SMB and enterprise. They build for procurement instead of users. A 28-year-old solo founder hiring your tool to "stop guessing what users want before the next planning cycle" is in the same job-segment as a 52-year-old PM at a 4,000-person company hiring it for the same reason. Same onboarding, same in-product copy, similar pricing pitch. Different ad creative, sure, but the product strategy is identical.
Jobs-to-be-Done segmentation tends to expose two things at once: surprising overlaps between groups marketing treats as different, and surprising splits inside groups marketing treats as the same. Both are useful. Both come from interviews, not from CRM filters.
Behavioral segmentation in practice#
Behavioral segments are the ones product and growth teams actually use day-to-day, because each one maps to an action you can take inside the product or in lifecycle email.
- Power users:
High frequency, high depth, multiple features in regular use. Source of advocacy, case studies, and the next-stage feature requests. Interview them to understand what makes the product irreplaceable, then protect those workflows.
- Dormant users:
Activated once, then stopped. Sometimes life happened, sometimes the product failed them. A short re-engagement script will recover a slice and tell you which onboarding promises did not land.
- Expansion candidates:
Hitting plan limits, inviting teammates, using premium-adjacent features. The cleanest signal for sales-assist outreach or self-serve upgrade prompts. Talk to them about workflow gaps before they ask for a discount.
- Churn risk:
Falling usage, support tickets with negative sentiment, missed renewal milestones. Skip the save offer first. Interview them while they still care enough to answer, then design the save.
These four cover most of the value of behavioral segmentation for a B2B SaaS team. A consumer product might add "casual" and "loyalist" tiers, or split power users by primary use case. Resist the urge to add a sixth and seventh segment until the first four are wired into actual workflows.
Lifecycle segmentation#
Lifecycle segmentation overlays "where in the journey" on top of everything else. Each stage has its own questions worth asking and its own mistakes worth avoiding.
Prospect
Knows you exist, has not signed up. Best researched through landing-page intercepts and outbound problem interviews. Talk about the problem, not the product.
Trial
Signed up, evaluating. Intent is at its peak and the memory of the alternative they almost picked is freshest. Ask what they tried before and what made them click.
Activated
Hit a meaningful first value moment. The right time to ask what almost stopped them. Onboarding insights are most accurate inside the first week.
Retained
Repeat usage, paying. The segment to mine for voice-of-customer themes, advocacy, and the next set of jobs they want done.
Churned
Cancelled or lapsed. Exit interviews here are gold and almost nobody runs them well. Ask about the trigger event, not just the dissatisfaction.
Won-back
Returned after churning. Tiny segment, huge signal. They will tell you exactly what changed in their world or in your product.
Lifecycle stages are the segments your CRM and lifecycle email tools already know how to address. The work is making sure each stage has at least one owner, one message, and one in-product surface tuned to it.
Common mistakes#
Too many segments. If a team has 14 personas, it has zero. Three to five active segments is the realistic working range. Past that, the team stops remembering which one applies to which decision and defaults back to "users".
Segments without research-backed differences. "Enterprise vs. SMB" is only a real segmentation if those two groups actually buy, use, and value the product differently. Validate with interviews and behavioral data before splitting the roadmap. Otherwise a sales-team intuition becomes a permanent product split.
Segments built but never used. A segmentation deck that lives in a research folder is a failed project. Each segment needs at least one owner, one channel, and one decision it influences this quarter. If none exists, the segment is theatre.
A useful sanity check: pick any segment and ask which roadmap item, pricing change, onboarding flow, or campaign would be different if that segment did not exist. If the answer is nothing, kill the segment.
Researching segments separately#
A single research method across every segment wastes time on some and under-serves others. Match the method to the segment: depth interviews and advisory panels for power users and retained customers, short contextual interviews for trial and activated, async low-friction conversations for dormant and churned, problem interviews for prospects. The right user research method is the one a given segment will actually complete.
If you want to talk to each segment separately to validate the segmentation, AI-moderated interview tools like Diaform can run targeted scripts per segment via one shared link. A churn-risk script probes for trigger events. A power-user script probes for missing capabilities and adjacent jobs. A trial script probes for the alternative they almost picked. Same product, different conversations, all running in parallel.
The test of a good segment#
A segment is worth keeping if naming it changes a decision. New onboarding step, different price, different ad creative, different feature on the roadmap, different save offer. If two segments produce the same decisions across product, pricing, and marketing, they are one segment with two names. Collapse them, and spend the freed attention on a split that actually moves something.