B2B Market Segmentation: Cut by the Job, Not the Size

What is B2B market segmentation?

B2B market segmentation is the work of cutting a market into groups that behave differently enough to deserve different treatment. In practice it produces a small number of defined segments, each with its own use case, problem and competitive alternative, from which exactly one is chosen as the ideal customer profile to build the go-to-market around.

The definition is uncontroversial. The cut is where it goes wrong, because there are many ways to divide the same market and most of them produce groups that look tidy and behave identically.

What follows is the version we run in client engagements: how the cut is made, what a segment definition has to contain, how the choice gets made, and the one test that tells you whether the segment is real.

Should you segment by firmographics or by use case?

By use case. A segment is named after the job the buyer is trying to get done and the workflow it sits in, not after a headcount band or an industry code. Firmographics stay in the definition as attributes, included only where they genuinely change the buying behaviour or the message.

This is the opposite of how most segmentation exercises start, and there is data behind the inversion rather than taste. If industry drove behaviour, buyers in different industries would buy differently. Mostly, they do not.

6sense's analysis of industry effects on the B2B buying journey, drawn from its 2024 Buyer Experience Report, is blunt about it: outside manufacturing, differences in buying processes across industries are more subtle than expected. Most buying groups run 11 to 12 members evaluating close to five vendors, while manufacturing groups sit at seven to eight members and three to four vendors. The gap between the earliest and latest industry in terms of when buyers first contact a vendor is 25 to 26 days inside a twelve-month journey, which the report itself describes as practically insignificant. Forrester's State of Business Buying, 2026, from a survey of nearly 18,000 global buyers, points the same way: 13 internal stakeholders and nine external participants shape an average decision regardless of the vertical they sit in. Industry is a filter you apply to a segment. It is a poor thing to build one out of.

Use case behaves differently. Two companies of the same size in the same industry, one replacing a spreadsheet and one replacing an incumbent vendor, are running two different purchases with two different competitive alternatives and two different sets of objections. That difference is worth a separate segment. A twenty-employee gap is not.

What belongs in a segment definition?

Eleven fields, six of which are mandatory. The required half describes the job and the pain, so the messaging can be written from it. The optional half holds the firmographics and the tech stack, which are included only when they change the answer rather than as a matter of routine.

Three of those fields do most of the work and are the three most often missing. The competitive alternative, because a segment defined without it produces messaging that argues against nobody. The sourcing signal, because it is what turns the definition into a list. And why this is winnable, because a segment you cannot win is a market study rather than a plan.

Specificity is the other half of the job. "Enterprise companies" is not a target company. "Fortune 1000 manufacturing companies" is. The test is whether someone who has never met the client could build a list from the sentence.

How many segments should a market segmentation strategy produce?

Between one and three, and never more. One when the focus is already obvious. Two when there are genuinely distinct use cases to choose between. Three when several are viable and the prioritisation needs a real discussion. Beyond three, consolidate rather than adding, because the number is telling you the cut is wrong.

Four or five segments is the most common output of a segmentation workshop and the least useful one. It usually means the market has been sliced by an attribute rather than by a job, and attributes subdivide forever. Use cases do not.

The count also has a practical ceiling nobody enjoys hearing. Every segment you keep needs its own messaging, its own list, its own sequence and its own reporting. Three segments is three of everything.

How do you choose the one segment to focus on?

With a scorecard, in a working session with the client, and with the explicit understanding that the output is one segment rather than a ranking. The segments go in as rows, the criteria that matter for this business as columns, and the discussion that follows is the actual deliverable, not the arithmetic.

Approval matters more than elegance here. The chosen segment goes back to the client in writing and nothing downstream starts until they confirm it, because positioning, messaging, list building and sequencing all read from this one decision. Reopening it in week six costs all four.

The payoff is rarely more customers. It is different ones. More than 8x ARR Growth in less than 3 years through 10-50x larger deals and building a scalable outbound engine is what a segment decision looked like for Classtime, from SalesPlaybook's own client list.

What if the client disagrees with the segmentation?

Then you stop and redo it, rather than negotiating your way through the session. Disagreement at this point is usually a signal that the segmentation was cut on the wrong axis, and continuing produces an ICP nobody owns. Objections here are cheap. The same objection surfacing after launch is not.

The disagreement announces itself in a recognisable way. It sounds like: we have clients that fit two of those segments at once, and you have not even looked at the size of their sales team, which is the factor that actually decides for us.

The response is to take a full step back rather than patch the tables. Explain what makes a segment a segment, reopen the customer research and the kickoff notes together, agree explicitly on the segmentation axis before drawing anything, and then redo the draft on the new axis. It costs a session and buys back the quarter that a wrong ICP would have consumed.

The scenario worth naming: the client has clarity already, the segments land, and there is nothing to redo. That happens, and it should end the session early rather than trigger a search for problems to solve.

When is a segment not a segment?

When you cannot source it. The last check is mechanical: take the definition to the list-building step and try to produce named companies from the observable signals in it. If the only way to build the list is to ask someone which accounts they had in mind, the definition describes a preference rather than a market.

This is also where the segmentation work connects to everything downstream. The addressable number, the working list, the sender capacity that number implies — all of it starts from a definition specific enough to filter against. Get sourcing right and the leads follow. Get the definition vague and no amount of enrichment repairs it.

One decision, made early and written down, is worth more than a thorough analysis that ends in a ranking. Two of those downstream decisions have their own mechanics: the roles a deal has to pass, in the buying center, and the number the chosen segment produces, in a pipeline generation plan. Everything after it inherits the choice: B2B pipeline generation builds the list against it, LinkedIn GTM and AI outbound write the sequences against it, and fractional sales leadership spends much of its first month finding out whether the choice was ever really made.