Segmentation strategy

Every market is a distribution of different types of buyers with different jobs, different constraints, and different willingness to pay. Trying to sell to all of them with the same motion is a way to sell to none of them well. Segmentation is how you map that distribution and pick where to start.

The temptation is to think that good segmentation is about being more precise. It is not. Good segmentation is about being more honest: acknowledging that your product solves different problems for different types of buyers, and that trying to optimize for all of them in parallel guarantees you will optimize for none of them. The segmentation decision is the first GTM decision because every downstream choice—motion, pricing, channel, retention strategy—flows from it.

Bad segmentation: the firmographic trap

The default move is to segment by firmographic: size, industry, geography. A SaaS company might say “we target mid-market HR teams” or “we focus on fintech companies in North America.” These segments are easy to describe and easy to market to. They are also often wrong.

Why? Because firmographic bins do not predict the job, the constraint, or the buying behavior. Two 100-person companies in the same industry have completely different incentive structures if one is pre-product/market fit and burning cash, and the other has found a repeatable motion and is disciplined about efficiency. The first is hiring a product that “helps us move faster.” The second is hiring a product that “reduces our cost per customer acquisition.” The product is the same. The job is different. The buying process is different. The price sensitivity is different.

A retail company with a centralized merchandising team has a different job than a franchiser with distributed buying authority. A software company bootstrapping has a different constraint than one with Series A funding. These structural differences—not the logo on their website—determine whether they will buy.

This is why “mid-market” and “enterprise” are dangerous segmentation labels. They say nothing about whether the buyer can move. They say nothing about what problem they are trying to solve. They are sorting by wallet size, which is a proxy for everything else and therefore a proxy for nothing specific.

The segmentation axes: the diagnostic matrix

Good segmentation isolates one or more of four independent axes. A segment can be defined cleanly on one axis, or it can be the intersection of constraints on multiple axes. The power comes from being explicit about which axis you are using.

Axis 1: The job.

Who is hiring the product to accomplish what specific outcome, and what alternative would they use if you did not exist?

Jira was segmented effectively not by “software teams” but by “teams that need to turn engineering work into visible status for non-engineers.” That is a much narrower job. HubSpot was not segmented to “companies that need marketing tools” but to “small B2B companies that have zero sales infrastructure and need to build one fast.” A spreadsheet or a Rolodex is the alternative. Narrower. Clearer. More convertible.

The job axis is the most powerful because it directly explains why a buyer would switch. But it is also the hardest to articulate precisely. The discipline is to finish this sentence: “A buyer is a match if they are trying to [specific outcome] in a context where the status quo is [the alternative they have today].”

Axis 2: The structural constraint.

What hard constraint governs whether a buyer can adopt this product, independent of desire?

For enterprise software, the constraint is often “we must integrate with our existing system” or “we must pass a SOC 2 audit.” For B2B SaaS, it might be “we must be able to implement without hiring external consultants” or “we cannot spend more than $1,000 a month.” For developer tools, it might be “must run on macOS” or “must be open-source compatible.” For consumer products, it might be “it must work on Android” or “it must be free or under five dollars.”

Constraints are different from preferences. A constraint eliminates adoption regardless of how good the product is. A buyer with a “must integrate with Salesforce” constraint cannot buy a non-integrating product, no matter how much better it is technically. Segmenting by constraint forces you to build a product (or a motion) that works in that constraint, not a product that works only for unconstrained buyers.

The discipline here is: would a buyer with this constraint be unable to adopt even if they loved the product? If yes, it is a constraint. If they could work around it, it is a preference, not a constraint.

Axis 3: The adoption velocity.

How quickly is a buyer able and willing to move from discovery to purchase to activation?

Some segments are eager to buy, willing to try new tools, willing to change their workflow immediately. Others need to be convinced and are slow to switch. B2B examples: startups moving fast (high velocity, weeks to decision) vs. enterprises with frozen budgets and committee approvals (low velocity, 6+ months). Logistics: tech-forward companies (high velocity) vs. traditional family businesses (low velocity). Healthcare: digital-native practices (high velocity) vs. legacy hospital systems (low velocity).

A GTM motion that works for high-velocity adoption (low friction, self-serve, cheap, rapid onboarding) will alienate low-velocity buyers who need time, proof, and relationship building. A motion that works for low-velocity buyers (high touch, long sales cycle, proven ROI, expensive, implementation support) is too slow and too expensive for high-velocity buyers who expect to be live in days. Separate them. Do not try to optimize for both.

The discipline here is to measure time-to-close for different prospect profiles. If one profile closes in 2 weeks and another in 6 months, they are different segments on the velocity axis, and they cannot share a motion.

Axis 4: The buying power.

Who in the organization decides whether to buy, and what is the approval mechanism?

A tool sold to individual contributors (engineers, designers, marketers) will have a different price, feature set, and buying process than a tool sold to managers, who have a different buying process than a tool sold to finance or procurement. An individual contributor in a 50-person company can often spend $1,000 from their own budget with one approval. A manager might spend $10,000 with two levels of approval. A CFO-level decision on a $100,000+ contract requires legal review, procurement processes, board approval, and integration planning. A company with a formal procurement team has a different buying constraint than one where a manager can spend from their budget. Do not try to sell to all three at once.

The discipline here is: who is the final decision-maker, what is the approval chain, and how many people must align? If the approval chain differs, they are different segments.

Asymmetric contrast: the “same company, different segments” test

The most useful diagnostic for bad segmentation is the asymmetric contrast: pick two companies that look identical on firmographics but are radically different on one of the four axes.

Example: Two 50-person B2B SaaS companies, different segments.

Company A is a 50-person B2B SaaS company, Series A, $3M ARR, building a sales intelligence tool. They are 18 months post-product/market fit. Revenue is growing 15% month-over-month organically. They have found the repeatable motion: product-led to engineering teams, expansion to sales leadership. Cash is abundant. They are hiring five people a month. They are disciplined: every hire must pay back in 6-8 months. They are moving fast: a new feature can be built and shipped in 2-3 weeks. They will evaluate a new category in a 3-week trial. If it does not work, they move on. Their job: “reduce friction in our GTM process.” Their constraint: “must integrate with Slack.” Their velocity: high (3-week decision cycle). Their buying power: each department buys independently from their budget.

Company B is a 50-person B2B SaaS company, Series A, $3M ARR, bootstrapped, building an HR management system. They are pre-product/market fit. They are burning cash. They have one large customer paying $100k, two mid-size customers paying $20k each, and a long tail of small customers. They do not know if the product works. Retention is 60% (churn is killing them). They are risk-averse: any new tool must integrate deeply or they cannot support it. Hiring is frozen. They need a tool that delivers proof-of-concept value in 6 weeks or they cannot justify the risk. Their job: “prove the platform works for our largest customer so they do not churn.” Their constraint: “must integrate deeply with our entire stack, including the legacy system we cannot replace.” Their velocity: low (3-month decision cycle minimum, because they cannot risk a new failure). Their buying power: one VP has to approve everything, because cash is tight and risk tolerance is zero.

Same company profile. Radically different segments. If you try to sell the same motion, pricing, and implementation approach to both, you will fail with both. Company A will never convert through a long enterprise sales cycle. Company B will never adopt through a self-serve trial. You are playing a losing game.

This is why “mid-market” and “50-person companies” are not segments. They are bins. Real segments are defined by the job, the constraint, the velocity, and the buying power. And those are often invisible on a firmographic view.

Common founder segmentation mistakes

There are four systematic mistakes founders make when defining segments. Knowing them helps you avoid them.

Mistake 1: Confusing who buys with who is fit.

You say your segment is “VP of Marketing” because they are the person who usually initiates the purchase. But the real segment is “VP of Marketing at a company where demand generation is a bottleneck.” If demand generation is not broken, the VP does not care about your solution, no matter how much they are your target buyer profile. The job is the segment. The persona is secondary.

Fix: Write down not just the job title but the specific problem they are trying to solve and why they cannot solve it today.

Mistake 2: Optimizing for size instead of shape.

You say your segment is “companies with $5M–$50M ARR” because that is where ACV and cost-to-serve pencil out. That is partly right, but it misses shape. A 20-year-old $40M ARR company with a 80% NRR and deep customer relationships behaves differently than a 2-year-old $10M ARR hypergrowth startup burning cash. Same ARR band. Different segments. Different motions.

Fix: Define your segment by the economic stage (pre-PMF, early PMF, scaling, mature) and the growth rate (burning cash, efficient, profitable), not just by the size.

Mistake 3: Assuming the constraint you have is the constraint they have.

You built the product assuming customers have no integration requirements. You sell to a segment where “easy implementation” is the job. Great. But then you encounter a segment where “must integrate with Oracle” is the constraint. You will never own that segment with your current product, no matter how good you are. That is not a sales problem. That is a segmentation problem.

Fix: Ask every early customer: “What almost prevented you from buying us?” The answer is often a constraint you did not know you had to solve for.

Mistake 4: Treating the segment as static.

You segment to “startups” because startups buy with high velocity and are willing to be scrappy. But your segment is moving. As those startups get older, they get slower. They go from 6-week decision cycles to 6-month ones. They add procurement processes. Their job changes from “move fast” to “reduce churn.” If you do not re-segment as your cohort matures, you will wake up one day selling to a segment you no longer understand. Your motion will feel misaligned.

Fix: Measure cohort behavior annually. Has your customer base’s average stage, size, velocity, or buying power changed? If yes, you may need a new segment or a new motion.

How to test segments: the diagnostic sequence

A good segment looks right on paper. A great segment actually buys. The test is rigorous and multi-staged.

Stage 1: Conversion rate (signal-to-noise ratio).

What percentage of prospects in this segment become customers within 90 days?

Track it by segment. If it drops below 5–10% or varies wildly within the segment (some prospects convert at 20%, others at 2%), the segment is not homogeneous. Some members are jobs you can solve; others are not.

The variance is the diagnostic. If all prospects in the segment convert at 7–12% with low variance, the segment is clean. If conversion is 6% for half the segment and 18% for the other half, you have two segments hiding inside one.

Stage 2: Activation rate (do they care?).

What percentage of buyers actually get value from the product within 30 days of purchase?

Activation is different from conversion. A prospect can convert because they were sold by a good salesperson. But do they activate? Do they actually use the product? If your segment has a 40% activation rate while another segment has an 85% activation rate, one segment is better-fit than the other.

The activation drop from conversion to activation is your “buyer’s remorse” signal. If 15% of prospects convert but 12% activate, activation loss is 20%. That is tolerable. If 10% convert and 3% activate, activation loss is 70%. That is a signal that the buyer thought they wanted it but did not actually need it. Wrong segment, or wrong selling story.

Stage 3: Retention and expansion (do they stay?).

Do customers stay for 12+ months? Do they expand? What is the NRR?

A segment where customers churn in six months and never buy additional features is not a segment worth building a business in, no matter how fast they convert. Churn tells you the job was wrong or the constraint makes adoption impossible.

Measure: 12-month net retention rate (are they bigger or smaller in month 12 than month 1?). For most B2B SaaS, good segments show > 90% NRR. Segments below 80% are at risk. Segments above 110% are expanding.

Stage 4: Unit economics consistency (can you build a motion?).

Do the unit economics (CAC, ACV, payback period) hold together within the segment?

If CAC is $5,000 for half the segment and $25,000 for the other half, you cannot build a single motion. You have two segments. If ACV is $12k for one subsegment and $60k for another, the buying committee is different, and you cannot sell to both with one sales motion.

Measure: CAC payback period by segment. If payback is 12 months for segment A and 30 months for segment B, they require different motions. Do not try to blend them.

Stage 5: Message-market fit (can you reach them distinctly?).

Can you explain why this segment should buy in a single sentence, and does that explanation resonate when you test it?

Run an ad to segment A with message 1. Run an ad to segment B with message 2. If message 1 resonates with segment A at 2x the rate of message 2, you have found a segment. If message 1 resonates equally with both segments, you have not separated them on a meaningful axis.

The asymmetry in messaging response is proof that the segment is real and distinct.

If a segment fails any of these tests, it is not a segment. It is a bucket of different types of buyers mixed together. Separate them or abandon the segment.

The naming rule: segment names must predict buying behavior

How you name a segment matters because the name becomes the shorthand everyone uses, and shortcuts become assumptions.

Bad segment names:

• “Mid-market” (size is not predictive)
• “SaaS companies” (industry is not predictive)
• “North America” (geography is not predictive)
• “CMOs” (title is not predictive)

Good segment names:

• “Pre-PMF SaaS, burning cash, high velocity, self-serve buyer” (describes job, constraint, velocity, power)
• “Scaling SaaS, efficient growth, disciplined about headcount, procurement committee” (describes stage, constraint, velocity, power)
• “Large enterprise, mature, SOC 2 required, multiple stakeholders, 6+ month cycle”

The rule: a segment name should be specific enough that someone who has never met a customer in that segment can predict how they will buy. If the name is “mid-market,” a new salesperson has no idea what motion to use. If the name is “Series A SaaS, just hired first sales hire, wants to avoid hard-coded integrations, wants to implement in 4 weeks, buys from the VP of Product,” a new salesperson knows exactly which motion to use.

Stripe named their segments by the buying motion and speed they enabled:

• “Developers, side projects, self-serve, $50-500 MRR”
• “YC companies, technical founder, self-serve, $500-5k MRR”
• “Growth-stage startups, RevOps-led, support, $5k-50k MRR”
• “Mid-market, sales-led, implementation, $50k-500k MRR”
• “Enterprise, field sales, strategic, $500k+ MRR”

Each name contains the job, the buying power, the velocity, and the expected ACV. The name is the shorthand for the motion.

When to multi-segment

Once you have found a segment that works—good conversion, good retention, good unit economics, clear naming—you can expand. But expand carefully.

The mistake is to broaden the existing segment (“mid-market” becomes “mid-market and large enterprise”) instead of finding an adjacent segment that has the same product but a different job, constraint, or velocity.

Stripe did this brilliantly. They started by segmenting to developers building side projects (high velocity, low price, self-serve). Then they segmented to development teams at Y Combinator companies (slightly higher price, still self-serve, faster growth phase). Then they segmented to growth-stage startups (even higher price, support-driven). Then to SMBs (very high touch, sales-led). Then to enterprises (white-glove, field sales). Same product architecture. Different segments. Different GTM motions. Each motion was optimized for the job, constraint, and velocity of that segment. They did not try to sell to all five at once. They did not broaden the “developer” segment into “developers and executives.” They added a new segment with a new motion.

The practical rule: you can support one new segment per 12-18 months in a scaling company. A new segment requires a new motion, new messaging, potentially a new sales team. If you try to add multiple segments simultaneously, you end up with no coherent motion for any of them.

The next decision: product-market fit cannot exist without segment-market fit

Once you have a clear segment, the next question is whether you have product-market fit in that segment. But here is the trap: product-market fit is actually segment-market fit. You do not have PMF in “the market.” You have it in a specific segment. A product can have PMF in segment A (strong retention, high NRR, repeatable motion) and zero PMF in segment B (churn, low velocity, no motion). Most founders do not discover this mismatch until they are already trying to scale into a second segment and find that all their assumptions break.