Product-Value Fit (Real PMF)

Ask a team “what does your product do?” and you will likely get a category name: “We're a CRM.” “We're an HR platform.” Then ask the same question of their best customers. The answers do not resemble the category; they sound more like this: “It cut month-end reconciliation from two days to half a day.” “It got new reps on the phone in their first week.”

The category name is the story you tell yourself. What the customer says is the real description of fit. Product-value fit is the work of finding that second sentence and turning it into a shared language for engineering, sales and marketing.

When you need it

After you start seeing PMF signals, but before you scale. Concrete signs:

• Some customers love the product, others merely try it, and you cannot explain the difference.
• Sales, marketing and product use different definitions of the “ideal customer.”
• Onboarding shows everyone the same flow, but only some users reach the “aha” moment.
• A new hire cannot describe the product in one sentence.

Where it is too early: if no user group shows clear attachment yet, trying to define fit is speculation. First find the trace of fit; defining it comes after. The discipline of searching for proof of fit is the subject of the PMF piece.

What most teams miss

Teams treat fit as a yes/no fact. Product-value fit is really a map of boundaries. The product works wonderfully for some people, under some conditions, for a particular job, and is weak elsewhere. A team that never draws that boundary tries to tune both product and sales for everyone and gets lost in the average.

Clayton Christensen's Jobs-to-be-Done framework helps here because it describes a situation, not a person. A common template runs: “When I am in [a situation], I want to [make progress] so that [an outcome] happens.” The same person hires your product in one situation and not in another. Fit is built with the situation the person is in, not with the person.

Fit is not a customer label, it is a description of a situation: who, at what moment, trying to finish which job.

Picture a scene. Two teams use the same invoice automation. The first is an accounting office trying to close hundreds of invoices at month end with two people; the second is an agency sending thirty invoices a month. Same product, same features. For the first it is a workload saver, for the second a nice-to-have. Their retention curves probably diverge too. Unless you define them separately, the average picture misdescribes both.

This framework also sharpens the boundary on the marketing side. A team that can answer “who are you not for?” filters out the wrong customer up front, which lowers churn and support load. Telling sales “everyone is a prospect” inflates the pipeline in the short term and the cancellation rate in the medium term.

The difference from the PMF piece: PMF asks “is it there?” Product-value fit asks “what exactly exists, for whom, and why?” If you cannot answer the second, you cannot reproduce the first.

The playbook

1. Define your best users by behavior

“Best” does not mean the biggest payers or the loudest voices. Identify users with high retention, who put the product at the center of their workflow and recommend it unprompted. They are usually a small slice of the total, and the point is to find what they share: which role, which scale, which existing tool, which trigger. If they share nothing, there is no fit to define yet, which is itself a valuable finding.

2. Decode the moment of purchase

Why did these users come to the product that day? What did they drop from their previous solution, what event triggered the move? In every interview ask “what happened that made you start looking?” The trigger event is the first half of your situation description, and it often decides your marketing message too: people do not search for categories, they search for solutions to something that just happened to them.

3. Separate progress from cost

What progress did the user expect and what costs did they accept: learning effort, data migration, convincing the team, changing a habit? Map both sides of value. Where cost is high, fit forms only if the progress is very large; otherwise users abandon the trial.

4. Write the fit sentence

“[Specific role], after [specific trigger], wants to do [specific job] despite [specific obstacle]; our product delivers [concrete outcome].” Have two different team members read it and check whether they picture the same customer.

5. Spread fit into product, price and onboarding

Redraw onboarding around the shortest path to the outcome in the fit sentence. For segments that do not fit, either open a separate path or route them out politely. Tie pricing packages to the value of this situation, not to the number of features.

What to measure

• Activation rate by segment: The share of users matching your fit sentence who reach first value. A leading indicator.
• Retention gap between fit and non-fit segments: A test of whether your definition actually separates anything.
• Sales win rate by fit score: Tag opportunities by fit level; if win rates diverge, the definition works.
• Support load per customer: Non-fit customers are usually the ones who ask for the most help.

Pitfalls

Demographic descriptions. “Ages 30 to 45, manager” is not a situation. Demographics are the shadow of behavior; the real description is what they are trying to do.

Calling everyone a fit. Stretching the boundary under sales-target pressure. The non-fit customer you win today comes back in six months as both churn and a reference you would rather not give.

Writing the definition once and filing it. As market and product change, the fit sentence ages. Retest it every quarter with your newest best customers.

How AI changes this

In AI products, early usage can swell with curiosity. People sign up to try something new, play for a couple of months and leave, so the early retention curve can make fit look stronger than it is.

A sturdier piece of evidence is the level where the retention curve flattens: the user base that remains once the curious are gone. According to a16z's analysis, the curves it examined mostly flatten around the third month, and where flattening starts depends on how broadly the product is used and how it is paywalled. The sample is a selected set of dozens of AI companies with more than $1M ARR. Your own cohort matrix shows where your curve flattens.

Rule of thumb. Split the curve into segments that match your fit sentence and those that do not; read the level and month where flattening starts separately for each.

Our reading: a small but flattening core can be a more valuable trace of fit than a large, constantly flowing crowd of users. Early volume does not show how far the core can grow.

Start this week

1. Pick your eight best customers and ask each: “What was the event that made you start looking for a product like this?”
2. Draft one fit sentence from the answers and have one person each from sales, product and marketing read it, to see whether they picture the same customer.
3. Evaluate the three customers you lost in the last 30 days against that sentence: were they non-fit, or a fit with a different problem?