Segmentation: cutting a market so a choice is possible
Cut a market into groups that buy the same way, test each cut with four questions, and show why an average across "SMBs" prices for nobody.
On this page
A segment is a group of buyers who share a problem, buy the same way and can be reached through the same channel, so that one product, one price and one message can win all of them. Segmentation is the cut that makes the next decision possible: until the market is split, there is nothing to choose between, only an average. This module follows one product throughout, a scheduling tool for small medical clinics at $99 a month and 2% monthly churn, in place of module 0's default $50-a-month customer. Every figure in it is an assumption in USD, there for the reader to replace with their own.
The mistake: "our market is SMBs"
The founder's first investor slide says the market is "small and medium-sized businesses that book appointments": 2,055,000 accounts. Inside that number sit three groups with nothing in common but a calendar.
| Segment | Accounts | Pays a month | Sales cycle | Monthly churn |
|---|---|---|---|---|
| Solo consultants: coaches, tutors, independent advisers | 2,000,000 | $15 | same day, on a card | 6% |
| Small clinics: one to five practitioners and a front desk | 40,000 | $99 | about 3 weeks: one demo, then a trial | 2% |
| Agencies: 10 to 50 staff booking client meetings | 15,000 | $300 | about 90 days, three approvers | 1.5% |
Average them, weighted by accounts, and the "SMB" pays
(2,000,000 × $15 + 40,000 × $99 + 15,000 × $300) ÷ 2,055,000
= ($30.0M + $3.96M + $4.5M) ÷ 2,055,000 ≈ $18.70 a month
and churns (2,000,000 × 6% + 40,000 × 2% + 15,000 × 1.5%) ÷ 2,055,000 ≈ 5.9% a month. Neither figure belongs to any group. Solo consultants are 97% of the accounts, so the average is a solo consultant with a slightly bigger budget. A team that prices at $18.70 and builds a self-serve funnel for 5.9% churn has built for the solos, and has walked away from the clinic that would pay $99 and the agency that would pay $300.
The sales cycle cannot be averaged at all. "Same day on a card" and "90 days with three approvers" are two companies: one runs on paid search, the other on a salesperson.
What makes a cut a segment
Philip Kotler and Kevin Lane Keller's Marketing Management gives the textbook test: a segment is measurable, substantial, accessible, differentiable and actionable. It is also the first step of their sequence, segmentation, targeting, positioning, which is this module's order. In practice the test is four questions, each answered with a number.
- Same problem, at a similar intensity? At a clinic a no-show leaves a practitioner idle and the front desk still paid: assume 12% of appointments, at $80 each. A solo coach has no-shows too, but nobody on payroll is waiting.
- Same way of buying? One price range and one decision process. If one member pays on a card in five minutes and another needs three approvers, they are two segments however alike their problem.
- Reachable through the same channel? Clinic owners read the same practice-management newsletters and attend the same regional association meetings. Solo consultants can be reached only through broad search, where every calendar product bids on the same words.
- Different enough from the next group that the product or message changes? Agencies need meetings routed round-robin across staff and synced to their sales records. Clinics need reminders a patient can answer. One product page cannot sell both.
Size, "substantial", comes last on purpose: a cut that fails any of the four questions is not a segment however large it is.
Cut by need and buying process, not by size
Headcount, revenue and industry code are easy to count, which is why every first slide uses them, but they are proxies. What predicts behaviour is the need and the way the group buys; a size band works only when it tracks both.
"Clinics with one to five practitioners" works because, here, headcount tracks the decision. Below six practitioners the owner is the buyer, there is no IT department, and the alternative is usually paper, a phone and a general booking tool. Above it the cycle lengthens, someone runs procurement, and a practice-management suite is already installed. The boundary sits where the buying process changes, not at a round number.
One test catches a bad cut quickly: write the sentence that would make a member book a demo. For clinics it is "cut no-shows from 12% to 5% without anyone phoning patients". If it fails for half the members, cut again; it means nothing to an agency. The progress a clinic is paying for, fewer empty chairs, is what the product course calls the job to be done, and a segment is roughly the set of buyers hiring for the same job through the same door.
Count bottom-up, and mark what is assumed
The investor slide reached its number top-down: "2M SMBs, and if we get 1%…", a wish multiplied by a population. The bottom-up version counts accounts that pass the four questions, from a register or an association's membership list, and attaches four numbers:
- Accounts: counted, not derived from a share of all businesses.
- ARPU: what members pay today for the alternative, then tested in demos.
- Sales cycle: first contact to signature, which decides self-serve or sold.
- Churn: from your own cohorts; before you have any, from interviews ("how long have you used your current tool?").
Each is labelled measured or assumed. It is the habit of a back-of-envelope estimate: every number can be rebuilt by the reader, and a number an investor cannot rebuild is one they will not believe.
The fifth number, the cost of reaching one account, is what turns a list of segments into a choice. It is the subject of the beachhead arithmetic, which multiplies these four and subtracts it.
When a segment hides a second segment
Even "small clinics" can be too coarse, and the signal is a number that splits in two. Suppose the 400 paying clinics churn 2% a month on average, but 300 churn at 1% and 100 at 5%. The average is the same 2% as before, (300 × 1% + 100 × 5%) ÷ 400, and it describes neither group. Assume 80% gross margin, so a clinic contributes $99 × 0.8 = $79.20 a month. A clinic's lifetime margin is that monthly margin divided by churn: $79.20 ÷ 1% = $7,920 for the first group and $79.20 ÷ 5% = $1,584 for the second, a fivefold gap hidden inside one segment.
The question is whether the 100 share a trait visible before the sale. If they are the one-practitioner clinics, or the ones that came from a $400 suite, that trait is a segment boundary and belongs in sales qualification. If nothing visible separates them, the fix is in onboarding, not in targeting.
The cut by what a clinic uses today, paper, a $29 booking tool or a $400 suite, turns out to matter most of all. It decides who the product is compared with, and it is where Dunford's method starts.
In practice
When an investor or a board asks for the market, give the segments separately, each with its four numbers, and put the sum underneath. "Our first market is 40,000 clinics with one to five practitioners, who pay $99 a month, buy after one demo and churn 2% a month" is a sentence they can check. A single 2,055,000-account market sounds larger and tells them the company has not chosen.
With the team, the segment definition is what sales qualifies against and marketing writes for; if it cannot reject a lead, it is not a definition.
The mistake that costs money is pricing and building the funnel for the average. An $18.70 plan and a self-serve sign-up serve the solo consultants; the clinics see a calendar priced like a hobby, the agencies see no salesperson, and the two groups that would pay five and sixteen times the average price never reach a demo.
Check yourself
1. A fourth group turns up: 6,000 veterinary practices paying about $120 a month after one demo, churning 2% a month, read through their own trade press. Merge them into the clinic segment or keep them separate?
Separate, for now. They pass questions 1 and 2 (no-shows cost an idle vet, an owner buys after one demo at a similar price) but fail question 3: the channel is different, so the cost of reaching them is unknown. Merge them only after one campaign through veterinary press shows the clinic message and a similar cost per demo. If it does, the segment grows from 40,000 to 46,000 accounts at no extra product cost.
2. An investor says the 2,000,000 solo consultants are the real prize. Estimate whether the clinics' demo-led motion can sell to them, assuming a won account costs $800 to acquire that way and 80% gross margin.
A solo consultant contributes $15 × 0.8 = $12 a month and churns 6%, so lifetime margin is $12 ÷ 6% = $200 against an $800 cost to win: every sale loses $600. The segment is real but needs a different company, self-serve and a CAC under $200. Say so, and keep the clinic motion.