Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Digital Marketing › Marketing Strategy

Market Segmentation

How to divide a market into segments that respond differently, judge which are worth serving, choose targets, and turn segments into digital audiences.
Guideintermediate8 min readUpdated

Why segment

Treating everyone the same produces a message that is slightly wrong for everyone, a channel plan that reaches the wrong people as often as the right ones, and a budget spent without knowing which customers it was buying. Segmentation fixes this by making the differences explicit, and it does a second job that is less obvious: it makes trade-offs visible. Choosing to serve one segment well is also choosing not to serve another, and a strategy that refuses that choice ends up with no position at all. The positioning guide depends on this step having been taken.

Bases for segmentation

Markets can be divided on many dimensions. The useful ones are those that predict different behaviour.

Need or job. What people are trying to accomplish and in what situation. A person buying a laptop for video editing and one buying for school have different needs, judge different attributes and use different words. This is usually the most powerful basis because it explains the purchase, and the hardest to get from data alone — it comes from audience research.

Behaviour. What people have done: visited, bought, bought repeatedly, abandoned, lapsed; which products, how often, how recently, at what value. Behavioural segments are directly observable in analytics and CRM, and they map straight onto digital audiences (Remarketing lists, customer lists, lookalikes).

Value. Current and expected customer value (LTV), and the cost of acquiring each kind of customer (CAC). Value segmentation decides where to spend: a segment with high value and high acquisition cost may still be the best investment.

Lifecycle stage. Prospect, first-time buyer, repeat customer, lapsed. The same person moves through these, and each stage wants a different message; the lifecycle marketing guide covers the post-purchase stages.

Firmographics (for business buyers): industry, company size, role, technology in use, buying process. Often the only practical basis for targeting in B2B channels, and a weak predictor of need on its own. The tutorial on building an ideal customer profile from CRM data shows how to find which of these fields actually separate the accounts that stayed and grew from those that left.

Demographics and geography. Age, income, household, location. Easy to target, easy to measure, and usually the weakest predictor of why anyone buys. Useful as a constraint (the product is only sold in these regions) more than as a definition.

Psychographics. Attitudes, values, interests. Powerful for messaging, difficult to target directly; platforms approximate them through interest and behavioural signals.

Most workable segmentations combine two bases: a need-based definition (what they want) with a behavioural or firmographic one (how to find them).

What makes a segment usable

A segment is worth defining only if it passes four tests:

TestQuestionFails when…
DistinctDo these people respond differently from others, in a way we can act on?the segment gets the same message, offer and channel as everyone else
ReachableCan we find and reach them through available channels, at acceptable cost?the definition is real but no targeting option approximates it
SubstantialIs it large enough, or valuable enough, to justify its own treatment?serving it costs more than it can ever return
MeasurableCan we tell whether we are winning in it?no data source identifies members, so results cannot be attributed

A segment that fails "reachable" in digital channels can sometimes be served through content (people find the page written for their situation) when it cannot be served through targeting.

Choosing which segments to target

Targeting is a resource decision. For each usable segment, estimate its size, its value, the cost of winning it and the fit with what the business does well, then choose. The common patterns:

  • Concentrated: one segment, served deeply. Right for small businesses and new entrants; a clear position, a focused plan, a dependence on one group.
  • Differentiated: several segments, each with its own message and sometimes its own offer. Right when the segments are each substantial and the business can run parallel plans without diluting them.
  • Mass with variations: one core offer, light variations in message by segment. Right when needs are similar and the differences are mostly in how people find the product.

The go-to-market guide covers the sequencing question that follows: which segment first, and when the next one is added. When the chosen segment is an industry, the guide to marketing to an industry vertical covers the proposition, proof and presence a business-to-business supplier needs to win it.

Illustrative example — not a real result

Segmenting a meal-planning app's market

Research finds three need-based groups: people managing a medical diet who want precision, busy households who want speed, and fitness-focused users who want macronutrient tracking. Behavioural data adds a fourth cut: a large group of lapsed users who tried the app once. The medical group is distinct and valuable but small and hard to reach through advertising; it is served through content for its specific conditions. The household group is the largest and most reachable, and becomes the primary paid target with a speed-led message. The fitness group overlaps heavily with existing users of competitor apps and is deferred. The lapsed group gets a re-engagement email flow rather than acquisition spend. Four segments, four different treatments, one of which is "not now".

From segments to audiences

In digital channels a segment has to become something a platform can target. The translation is rarely exact:

  • Behavioural segments map to customer lists, website and app audiences, engagement audiences and the Lookalike Audience expansions built from them.
  • Need-based segments map to search terms (what someone with this need types), content topics, and interest or in-market categories that approximate the need.
  • Firmographic segments map to professional-network targeting and to account lists.
  • Demographic segments map directly but are usually too broad to use alone.

The mismatch matters: a platform audience is an approximation of the segment, and results should be read as results for the approximation. Checking how well the audience matches (through conversion rate by audience, survey questions, or CRM matching) is part of the measurement plan. One segment can also hold two audiences when the person who pays is not the person who uses the product, and the guide to marketing when the buyer is not the user sets out how to reach and measure each role.

Segment profile template

One page per segment the business will treat differently, filled from research rather than imagination.

FieldWhat goes here
NameShort, descriptive, used consistently across teams
Defining need or jobThe situation and goal that make this group distinct
Size and valueHow many, how much they are worth, how much they cost to acquire
How to recognise themBehavioural, firmographic or demographic markers that identify members in data
How to reach themChannels, platform audiences, search terms, content topics that approximate the segment
What they valueAttributes they judge, in order
ObjectionsThe reasons they do not buy, paired with proof that answers each
VocabularyTheir words for the problem, the category and the benefit
TreatmentWhat the business does differently for them: message, offer, channel, product
Owner and review dateWho keeps the profile current, and when it is next checked

Segmentation checklist

0/7

Segmenting with little data

A new business has no behavioural data and a small research budget; segmentation still matters. Start from the job: list the situations in which someone would buy, group situations that need the same thing, and treat each group as a hypothesis. Pick one to serve first using the usable-segment tests, with "reachable" weighted most heavily, because a segment you cannot find is not a segment yet. Then let the first customers correct the hypothesis: interview them, read how they describe themselves and the problem, and watch which messages they responded to. Within a few dozen customers the real segments usually look different from the first guess, which is the point of starting small.

Keeping segments alive

A segmentation is a model of the market at one moment. Segments merge, split and change value as the product, the competitors and the platforms change. A light annual review — do the segments still differ, are the targets still right, has a new group appeared in the data — keeps the model honest; the annual planning guide includes it. Tagging research findings by segment, as the tutorial on building a research repository does, lets that review start from everything already learned about each one.

Common mistakes

  • Segments nobody treats differently. A deck with six personas and one campaign.
  • Demographics mistaken for needs. "Women aged twenty-five to forty" is a targeting constraint, not a segment.
  • Too many segments. Beyond what the team can serve, segmentation becomes reporting.
  • Refusing to choose. Targeting every segment to avoid losing any produces a message for none.
  • Ignoring value. The easiest segment to reach is often the least valuable; the segment that complains most is often the smallest.
  • Treating platform audiences as the segment. Interest categories are the platform's guess, not the customer's need.

Segmentation sets up positioning: once the segment is chosen, the question becomes what the brand should mean to it.

Frequently asked questions

What is the difference between segmentation and targeting?

Segmentation divides the market into groups; targeting chooses which groups to serve. Both come before positioning, which decides what the brand should mean to the chosen group. Together they are the STP sequence.

How many segments should a business have?

As many as it will actually treat differently — in message, offer, channel or product. Segments that get the same treatment are one segment. Most small businesses work with two or three; large ones with a dozen, each with an owner.

Can segmentation be done from analytics data alone?

Behavioural segmentation can (new versus returning, high-value versus low-value, by product category). Need-based segmentation, which is usually the more useful kind, requires research into why people buy, which analytics cannot see.

Terms used on this page

Tools to try

Frameworks

Where this page is used

Also related

Related