Product-Market Fit Is a Relationship, Not a Milestone

Product-market fit is often discussed as a gate: find it, declare it, then scale. That shorthand is useful for rejecting premature growth. It becomes dangerous when the declaration outlives the evidence.

A product can fit one use case but not another, delight practitioners while failing the buyer, or retain an early segment that is too small to support the business.

The same product can lose fit as prices, alternatives, regulation, workflows, or customer expectations change. Nothing in the codebase needs to break for the relationship with the market to weaken.

Treat product-market fit as a relationship you can describe and test, not a badge the company earns forever.

Name the relationship before measuring it

Marc Andreessen’s classic formulation describes product-market fit as a good market served by a product capable of satisfying it. The frequently neglected word is market.

“Mid-market companies” is rarely precise enough. A market definition should tell the team whose behaviour it is studying and why those people might choose, use, and pay for this product.

Write the current fit hypothesis in four parts:

1. Product: the promise and experience being evaluated, including the version or plan when those differences matter.
2. Market: the segment, buyer, user, and use case—not merely an industry label.
3. Behaviour: the observable result that would indicate recurring value.
4. Timeframe: a period long enough for the relevant need and repeat behaviour to occur.

A useful statement might read:

For finance teams closing several entities each month, the reconciliation workflow reduces manual investigation enough that teams complete every close with it and renew after a full planning cycle.

That statement is still a hypothesis. Its value is that retention, interviews, pricing evidence, and losses can all be examined against the same unit of analysis.

Research-Driven Opportunity Sizing offers a compatible way to set explicit boundaries around the people, problem, and attainable value.

Read several signals, not one score

No single metric can prove product-market fit. Each signal illuminates part of the relationship and hides another part.

A serious review combines evidence about the problem, reached value, repeated value, payment, and reach. Contradictions are useful. They show where the current story is too simple.

Problem pull

Problem pull is stronger than positive reaction to a concept. Look for recent attempts to solve the problem, material consequences, recurring workarounds, budget already spent, or a deadline that makes inaction costly.

Enthusiasm in an interview is weak evidence when the problem is hypothetical. A clumsy spreadsheet used every week may be stronger evidence than praise for a polished prototype.

How to Approach Customer Discovery Systematically shows how to investigate past behaviour and context without turning an interview into a purchase forecast.

Reached value

Activation matters only when its event represents a result the customer cares about. Creating a workspace, inviting a colleague, or clicking a feature may be setup rather than value.

Define the shortest credible path to the promised outcome. Then inspect who reaches it, how long it takes, what prevents it, and whether assisted onboarding is concealing product friction.

For a multi-sided product, value may require several actors. A buyer completing setup is not proof that the end user has benefited.

Retention appropriate to the use case

Retention is powerful because repeated voluntary behaviour can reveal durable value. It is also easy to misread.

The return interval should match the job. Daily retention is irrelevant for annual tax software. Account retention may hide that the practitioners who need the product have stopped using it.

Define the starting event, return event, denominator, cohort age, and observation window. Compare cohorts only after each has had the same opportunity to return.

Do not compress all customers into one curve. Segment by use case, acquisition promise, plan, maturity, or another distinction that could explain a different value pattern.

A stable aggregate can conceal two opposing movements: strong retention in one segment and rapid deterioration in another.

Willingness to pay and continue paying

Payment is evidence that someone can capture enough value to exchange money for it. It is not automatically evidence that the user experiences the promised outcome.

Study the entire commercial behaviour: trial conversion, discount dependence, renewal, expansion, contraction, payment delays, procurement objections, and the alternatives used in negotiations.

In B2B products, ask who receives value, who controls the budget, and whose risk determines renewal. Fit can fail at any of those interfaces.

Stated willingness to pay is useful for exploring language and ranges. A purchase, renewal, or credible budget trade-off is stronger evidence.

Credible reach

A product may create exceptional value for a small group the company cannot identify or reach reliably. That can be a worthwhile niche, but it is not the same commercial proposition as a large, accessible market.

Reach evidence includes a recognisable segment, a repeatable way to find it, a message tied to the actual use case, and acquisition economics compatible with the business model.

Keep the distinction clear: a campaign spike proves that a channel produced attention. It does not prove that the resulting customers reached or retained value.

Separate fit from temporary growth

Growth can come from promotion, a platform feature, sales effort, migration incentives, an unusual news cycle, or customers buying ahead of a deadline.

Those mechanisms are not bad. They answer a different question.

To distinguish demand from durable value, trace cohorts from acquisition promise through activation, repeat use, payment, and renewal. Compare channels only after their cohorts have had time to mature.

Look for a coherent chain:

• the message attracts people with the intended problem;
• those people reach the promised value;
• enough of them return when the need recurs;
• the buyer can justify continued payment;
• serving them supports the economics of the product.

If one link is weak, name the uncertainty. “We have strong acquisition but unclear repeat value” is a better operating statement than “we almost have PMF.”

Fit belongs to a segment and use case

An average can create the appearance of fit where no single customer group has a coherent experience.

Suppose a planning product serves team leads and portfolio managers. Team leads value quick updates. Portfolio managers value comparison, permissions, and auditability.

They may share an account while evaluating different promises. Combining their usage can make adoption look broad even if neither workflow is satisfactory.

Review fit at the smallest segment that could support a distinct product or go-to-market choice. Then ask whether the differences justify separate experiences or whether the product can serve both through one coherent system.

Do not segment until every cohort is too small to interpret. A segment earns separate analysis when it changes the expected behaviour, the value mechanism, or the decision the team would make.

This also prevents a common expansion error: treating enterprise as a larger version of small business. The user, buyer, implementation burden, risk, and definition of value may all change.

The 40% survey is an indicator, not a verdict

The Sean Ellis survey asks how users would feel if they could no longer use the product. In Superhuman’s published account, the team used the share answering “very disappointed” as a leading indicator.

That account describes a 40% benchmark and explains how Superhuman segmented responses to find the users who valued the product most.

It does not establish a universal law. The result depends on who receives the survey, what experience they have had, when they are asked, how many respond, and what “very disappointed” means in that product category.

Use the survey diagnostically:

• include people who have experienced the product’s core value, not every sign-up;
• inspect response and selection bias;
• compare segments rather than celebrating one blended percentage;
• read the reasons behind the answer;
• test whether the survey pattern agrees with behaviour and payment.

A high score with weak retention demands investigation. A lower score in a low-frequency, high-consequence product may mean the question is a poor proxy for value.

The survey can sharpen a hypothesis. It cannot remove the need to understand the market.

Entering an adjacent market resets the question

An adjacency may share technology, brand, or a feature set with the core product. It does not inherit product-market fit.

A new segment can have a different trigger, workflow, buyer, risk tolerance, price model, support burden, and set of alternatives. Reusing the product reduces build cost; it does not prove demand.

State what is being carried over and what must be learned again. Existing evidence may support confidence in the capability while offering little evidence about the new market.

Treat the adjacency as a fresh hypothesis:

• Which problem is urgent in this segment?
• Which part of the existing promise transfers?
• What changes in adoption, value, and purchase?
• Which current strengths create an advantage?
• What evidence would justify a larger commitment?

Expansion works best as staged learning, not as a relabelling exercise.

Detect fit that is eroding

Fit usually weakens in the details before the company-level metrics make the problem undeniable.

Watch new cohorts and important segments for changes in reached value, repeat behaviour, renewal reasons, discount pressure, support burden, and the alternatives named during losses.

Qualitative evidence matters here. Customers may still renew while describing the product as a legacy dependency rather than a preferred way to work. That is a different kind of retention risk.

Separate product change from market change. A declining cohort may reflect worse onboarding, a new customer mix, a competing substitute, changed regulation, or a job that occurs less often.

Do not schedule a ritual “PMF check” and assume the calendar creates insight. Review fit when meaningful evidence changes or before a commitment that depends on the relationship remaining true.

Useful triggers include a new segment, pricing change, acquisition promise, major workflow redesign, competitor shift, or sustained divergence between cohorts.

A hypothetical choice: discover, strengthen, or scale

Consider a fictional compliance product used to prepare evidence for external audits. The scenario is invented to show how the signals lead to different decisions.

Mid-sized customers repeatedly complete the evidence workflow, renew without heavy discounts, and describe a clear reduction in coordination work.

Smaller customers sign up after a successful campaign but rarely reach a complete audit package. Interviews suggest that many face the problem only occasionally and prefer support from an adviser.

Enterprise prospects express interest, yet procurement uncovers requirements for controls and data residency that the current product does not meet.

Calling the whole company “post-PMF” would hide three distinct decisions.

• Scale the validated mid-market motion, while watching service quality and cohort retention.
• Discover whether the small-business problem and delivery model justify a product bet.
• Strengthen enterprise fit only if the required controls align with strategy and economics.

Product Growth Strategies helps turn that diagnosis into a deliberate choice about market penetration, development, product expansion, or diversification.

A product-market-fit review

Before increasing investment, ask:

• Can we name the product, segment, use case, behaviour, and timeframe under review?
• What evidence shows the problem is urgent enough to pull a solution?
• Which event represents reached value rather than completed setup?
• Is retention defined for the natural frequency of the job?
• Which cohorts or roles behave differently, and why?
• What do purchase, renewal, contraction, and discounting reveal?
• Can we identify and reach enough of the relevant market credibly?
• Could promotion or one channel be masking weak repeat value?
• Does survey evidence agree with observed behaviour and payment?
• Which signal has weakened, and what competing explanations remain?
• Are we deciding to discover, strengthen, or scale?
• What evidence would change that decision?

Product-market fit is not a finish line. It is a claim about a particular product and market at a particular time.

Make the claim narrow enough to test. Then let the disagreements between signals improve the decision.

Sources

• The only thing that matters — Marc Andreessen, Stanford copy
• How Superhuman Built an Engine to Find Product-Market Fit — First Round Review
• How the Retention Analysis chart calculates retention — Amplitude

Read next

Product Growth Strategies explains how to choose a growth direction once the evidence supports a specific product–market relationship.