The method that defines the audience by the job they are trying to get done, not by who they are. Two buyers with different titles, industries, and company sizes belong in the same segment if they are hiring a product for the same job in the same situation. Two buyers with identical firmographics belong in different segments if their jobs differ.

The unit of segmentation is the job, not the customer.

What this method does

Firmographic segmentation groups buyers by attributes you can read off a CRM: industry, headcount, role, region. Those attributes are easy to filter on and weak at predicting behavior. JTBD segmentation groups buyers by the progress they are trying to make and the situation that triggered the search. People with the same job behave alike under marketing pressure, which is the only kind of similarity that makes a message land.

The output is an audience defined as a triggering situation plus a blocking decision, sharp enough that L2 can write a single sentence against it.

Steps

1. Anchor on the L0 problem

Whose working week actually contains the problem L0 named?

Start from the problem statement, not from your customer list. The customer list is biased by who you already reached. List the kinds of people who hit this exact problem, in behavioral terms, before any attribute creeps in.

2. Find the trigger

What event pushes this person from coping to looking?

The trigger is the moment the problem becomes urgent enough to act on. Interview recent switchers, anyone who changed their behavior in the last six months, and ask when they first thought about looking for something different. The first thought is more diagnostic than the purchase, because it surfaces what set the change in motion. A segment without a trigger is a need description, and need descriptions cannot explain timing.

3. Name the job and the blocking decision

What progress are they trying to make, and what decision is currently stuck?

State the job in the buyer’s own words, stripped of product language. Then name the decision they are mid-stream on and what is blocking it. “Should we replace the manual process, and we cannot because no one trusts the new numbers enough to defend the switch.” The blocker is the thing L2 will have to dissolve, so it has to be specific.

4. Group by job, then check the spread of attributes

Do the people in this segment share a job even though their firmographics differ?

Collect the buyers who share the trigger and the job into one segment. Now look at their attributes. A healthy JTBD segment usually spans several industries and company sizes, because the job cuts across them. If your segment happens to be firmographically uniform, that is fine, but it should be a consequence of the job, not the thing you started from.

5. Cut to one segment and pressure-test it

If you had to win one segment first, which, and can someone name real people in it?

Pick one segment to lead with. Describe it to someone in that world. If they can name two or three real people who fit, it is sharp enough. If they say “basically everyone in X,” the job or the trigger is still too generic. Return to step 2.

Worked example

Using a fictional RevOps tool called RevSync:

Segment: teams who own the board pipeline number and have just been burned by it. Trigger: a board or QBR deck went out with a pipeline figure that did not survive scrutiny in the room. Job: walk into the next high-stakes meeting with one number they can defend. Blocking decision: whether to keep hand-reconciling Salesforce against the warehouse, which no one has time for, or trust a tool to do it.

Notice what this segment ignores: industry, headcount, whether the title is “RevOps,” “Sales Ops,” or “Finance.” A 40-person startup and a 4,000-person enterprise can both be in it, because the job and the trigger are the same. A firmographic segment (“RevOps leaders at Series B SaaS”) would have split that audience and missed the Finance buyer who has the same job.

Common pitfalls

• Reaching for firmographics first because they are filterable, then bolting a job on afterward. The job has to lead, or it is just an ICP with extra words.
• A segment with no trigger. Without the “when,” you cannot explain why this person acts now rather than next year, which is the whole point.
• Naming the job in product language (“they need reconciliation software”). That is a feature wish. The job is the progress, not the product.
• One segment that quietly contains three jobs. If three different triggers map to it, you have three segments. Split them.
• A segment so narrow it describes one company. The job should be shared by a market, not a logo.

Validation checklist

Before treating the segment as final, run it against every item below. If any fail, it is not done.

• The segment is defined by a job and a trigger, not by industry plus headcount plus title.
• The trigger is a specific event, not a generic state of dissatisfaction.
• The blocking decision is named and specific enough to point at.
• You can name two or three real people in the segment.
• The segment spans firmographics where the job does, rather than being firmographically uniform by default.
• L2 could write one sentence against this segment without asking you a follow-up question.

When to use this method

Most B2B marketing, and especially: new or fuzzy categories, products that span several buyer types, demand-led motions where the message has to do the qualifying, and any situation where the buying decision is driven by a triggering event rather than by which vendor scores highest on a fit model.

Where it disagrees with the rival

ICP leads with the account and asks which companies are worth pursuing. JTBD segmentation leads with the person and asks which decision is stuck. ICP is stronger when account selection is the binding constraint, as in sales-led and account-based motions with long buying groups. JTBD is stronger when the message has to do the work of finding and moving the buyer. The honest reconciliation is sequence, not victory: in an account-based motion, ICP picks the accounts and JTBD picks the message inside them.

When to pick the rival instead

When the deal is large, the buying group is long, the motion is sales-led, and the real question is which accounts to aim at rather than which decision to dissolve, lead with ICP and use JTBD only to sharpen the message once the accounts are chosen.

Self-check

Can you now use JTBD segmentation?

You should be able to:

• Explain why the job, not the customer’s attributes, is the unit of segmentation.
• Locate a triggering situation from a switch interview rather than guessing it.
• Name a blocking decision specific enough for L2 to write against.
• Recognize when a segment is secretly firmographic and rebuild it around the job.
• Say when account fit should lead instead, and hand off to ICP without treating it as a defeat.

If any of these are unclear, return to the relevant step above, or walk back to L0, since a segment that will not sharpen usually rests on a problem that was never finished.