Market segmentation is the process of dividing a broad market into smaller groups of buyers who share similar needs, behaviors, or characteristics — so a company can design products, pricing, and messaging for each group instead of one generic version for everyone. It's the analysis step; deciding which segment(s) to pursue (targeting) and how to position your product to them come after. Most of the confusion around this term comes from skipping straight to targeting without doing the segmentation groundwork first.
Every segmentation model boils down to some combination of four variable types. A running-shoe brand illustrates all four at once:
Geographic — where the buyer is located. Trail-running shoes sell differently in mountain-adjacent regions than in flat urban markets; humid-climate customers care more about breathability than insulation.
Demographic — age, income, gender, occupation. A $180 carbon-plate racing shoe and a $60 everyday trainer are aimed at different income brackets and different levels of running commitment, even though both are "running shoes."
Psychographic — values, lifestyle, personality. One buyer runs to compete and tracks splits obsessively; another runs to decompress and never looks at pace data. Same product category, entirely different motivation to sell to.
Behavioral — purchase habits, usage rate, brand loyalty, occasion. First-time marathoners buying one pair of shoes behave nothing like a run-club regular who replaces shoes every 300 miles and buys three colorways of the same model.
A segment can be real and still be useless to a business. Before building a plan around one, run it through four checks:
A segment that fails even one of these tests usually isn't worth a separate strategy — fold it into an adjacent segment instead. A common failure case: a startup identifies "environmentally conscious consumers" as a segment, which is measurable in surveys and clearly substantial in size, but turns out not to be accessible through any channel the team can afford, and not actionable because the product itself doesn't differ for that buyer versus any other. That's a value, not a segment — useful for messaging tone, not for building a separate go-to-market plan.
Segmentation is the first of three linked steps that marketers usually run together:
| Step | Question It Answers |
|---|---|
| Segmentation | What distinct groups exist in this market? |
| Targeting | Which of those groups can we serve better than competitors? |
| Positioning | How do we want the target segment to perceive us relative to alternatives? |
Skipping segmentation and jumping straight to "who's our target market" is how companies end up targeting a group defined by gut feel rather than by evidence that the group behaves differently from everyone else.
Take a mid-size skincare brand deciding how to split its market before writing a single ad. A reasonable first pass, combining demographic and behavioral data, might produce four segments:
Each of those segments passes the MASA test — they're measurable via purchase data, reachable through distinct channels (TikTok for teens, search + email for the anti-aging segment), large enough to matter, and each genuinely wants a different message. That's the difference between a real segmentation model and a demographic slide that never changes how the business actually operates.
Formal market research is useful but not required to start. CRM purchase history, on-site search queries, post-purchase surveys (even a two-question one at checkout), and Google Analytics audience reports all provide real behavioral and demographic signal. The goal at the start isn't statistical certainty — it's a hypothesis specific enough to test with one campaign aimed at one segment.
Everything above leans B2C, but B2B segmentation swaps the four bases for firmographic and technographic equivalents: company size (employee count or revenue band), industry vertical, buying-committee structure, and existing tech stack. A project-management SaaS tool might segment prospects by company headcount (a 15-person startup and a 2,000-person enterprise buy on completely different criteria — speed and price versus security review and procurement process) rather than by anything about the individual buyer. The MASA test still applies: an enterprise segment that's measurable and substantial but not accessible without a six-month sales cycle you can't staff for isn't actionable yet, no matter how attractive the contract value looks on paper.
Segmentation is the analysis — dividing the whole market into distinct groups. A target market is the decision that follows — which of those groups you've chosen to actually pursue. You segment first to see what groups exist, then target the ones worth serving.
There's no fixed number, but if a team can't name a genuinely different message, price point, or channel for each segment, there are too many. Most small and mid-size businesses operate well with three to five active segments; more than that usually means some are being defined but never acted on.
Yes. CRM data, checkout surveys, on-site search terms, and analytics audience breakdowns give real signal without a formal study. Start with the behavioral data you already collect — purchase frequency and order value split customers more reliably than guessed-at demographics.