Product-Market Fit: Signals, Tests and Limits
Product-market fit is evidence that a defined market repeatedly chooses and receives value from your product. Look for retention, customer pull, willingness to pay and a delivery model that can support the relationship. No single survey score, growth rate or funding event proves it.
Treat fit as a working conclusion about a segment and use case. It can weaken when customer needs, competition, pricing or the product changes.
Define the market you are testing
Name the customer, the recurring job and the alternative they would otherwise use. A product can fit one segment while confusing another. Aggregating both can hide the strongest opportunity or create a false sense of broad demand.
Distinguish the buyer from the user. An enterprise buyer may renew a contract while employees barely use the product. Conversely, enthusiastic users may lack a budget or purchasing authority. Both product value and buying behavior matter to the business model.
Look for several kinds of evidence
| Signal | What it may indicate | What to check next |
|---|---|---|
| Repeated use at a natural cadence | Ongoing utility | Is use voluntary and meaningful? |
| Customers pay or renew | A valued commercial relationship | Is it sustained by discounts or switching barriers? |
| Unprompted referrals or expansion requests | Customer pull | Are requests from the target segment? |
| Customers invest time in adoption | A problem worth solving | Does the result justify the work? |
| Consistent reasons for choosing the product | A clear value proposition | Do actual behaviors match the explanation? |
Michael Seibel’s account of product-market fit emphasises strong customer demand and warns against treating hiring or fundraising as proof. That is a useful corrective to vanity signals, rather than a universal numerical test for every type of company.
Examine retention by cohort
Choose a meaningful activity and interval. Weekly return may suit a weekly planning tool. It would be a poor test for software used to prepare an annual filing. Define whether the unit is a person, team or paid account.
Group customers by the time they started and compare them at the same age. Inspect changes by segment, acquisition source and product version. A healthy total can be sustained by new arrivals while every cohort loses most of its customers.
For a new product, long-term evidence may simply not exist yet. State that limit. Early usage and a short pilot can justify another experiment without establishing durable fit.
Hypothetical example: three user segments
A fictional reporting tool attracts freelancers, small agencies and internal enterprise teams. The top-line dashboard shows growing sign-ups. Cohort analysis reveals that agencies return for each monthly client report, freelancers use it once and enterprise teams cannot complete their approval process.
The team interviews retained and departed customers in each group. Agencies value a repeatable reporting format and are willing to pay for it. Freelancers mainly wanted a one-time export. Enterprise interest is real, but supporting the required permissions would create a substantially different product.
The team focuses its next release and acquisition test on agencies. It checks whether newer agency cohorts also return, whether paid renewals persist and how much manual help each account needs. It does not claim fit across all three segments because the combined sign-up graph rises.
Guard against false positives
Founder-assisted use, promotional offers, mandatory deployment and novelty can all create promising early activity. They are not necessarily worthless signals, but they need context. Record discounts, implementation help and any conditions that may disappear later.
Surveys can reveal what customers value and what they would miss. Their interpretation depends on who responds, their experience and the wording. A selected group of power users cannot represent everyone who tried and abandoned the product.
Decide what the evidence supports now: continue discovery, concentrate on a segment, improve the core journey or cautiously expand. Use product feedback prioritization to choose the next change, and revisit MVP scope if the original hypothesis no longer fits the observed market.
Build a segment-level evidence sheet
For each plausible segment, record the job, frequency of need, current alternative, activation, repeated meaningful use, payment behavior and operating effort. Keep the sample size and observation period beside every result. This prevents a small but promising group from being described as an established market.
Include what customers do without you. If an agency exports data to a spreadsheet after every report, ask why. The workaround may reveal an important missing capability, a preferred review method or a boundary where the product is already sufficient. Do not assume every external tool represents a feature opportunity.
Use the customer interview notes template to compare recent events consistently. Ask retained customers what would be difficult without the product and departed customers what they use instead. Keep the evidence separate from the team’s explanation of it.
Interpret retention curves carefully
A cohort table can show the proportion of a starting group that completes a defined useful activity at successive intervals. Specify whether you mean activity in each interval, any return after that point or an active paid subscription. These are different definitions and can produce different-looking curves.
If an illustrative monthly agency cohort begins with twenty accounts and twelve create a report in the third month, activity retention for that interval is 60% under that definition. The calculation is not a benchmark. It also says nothing by itself about payment, report quality or whether the eight inactive agencies no longer needed a report.
Inspect the denominator and timing. Annual contracts can delay visible cancellation. New accounts may not have reached the first natural renewal occasion. An apparent flattening in a small cohort can reflect few observations. Use the retention metrics calculator for its stated customer and revenue conventions, and keep product-activity retention defined separately when it measures something else.
Use surveys as one evidence stream
A survey asking how disappointed customers would be without the product can reveal perceived importance, especially when followed by a question about the value they would miss. It cannot establish fit without attention to eligibility, response bias and actual behavior.
Do not apply a circulating percentage threshold as a universal pass mark. A survey of enthusiastic recent users differs from a survey including abandoned accounts. Report who was invited, who responded, what experience they had and the exact question. Investigate discrepancies between strong stated affection and weak voluntary renewal.
Design the next discriminating test
Choose a test that separates plausible explanations. If customers leave before setup, observe onboarding. If they complete setup but do not return, investigate whether the job recurs or whether the result is inadequate. If they return but refuse to pay, examine buyer incentives, alternatives and the commercial offer.
For the fictional agencies, a useful next test might be a limited paid pilot over several reporting cycles with ordinary support. Record the implementation help and price conditions. If agencies renew only with continuous bespoke founder work, the team has evidence of demand for a service but has not yet demonstrated the proposed product model.
Decide how much expansion the evidence supports
Keep three judgments separate: the segment values the outcome, the product can deliver it reliably and acquisition can reach suitable customers economically. Strength in one does not establish the others. A tightly supported pilot may validate the problem while leaving distribution unresolved.
Use the experiment plan to specify the next uncertainty and review date. Expand the segment or channel gradually enough to notice when the new customers differ. Continue monitoring customer pull and delivery burden after an encouraging result; product-market fit is a relationship with a market that can change.