How to Run a Product-Market Fit Assessment: Steps, Examples and Checklist

When you need this document

Run a fit assessment during Validate, when accumulated signals need converting into a judgement: before scaling spend on marketing or sales, before committing a bigger roadmap, or when growth feels harder than it should and the team suspects the product is being pushed rather than pulled. It is a periodic reading rather than a one-off gate. If you are testing one specific hypothesis, a validation report is the sharper instrument; the fit assessment weighs the whole accumulated picture.

Gather these first

• Retention and engagement data over enough time to show whether usage endures
• Commercial signals: conversion, willingness to pay, renewal and expansion behaviour
• Acquisition mix — how much growth is organic pull versus paid or founder-driven push
• Qualitative evidence: user interviews, churn conversations, unsolicited feedback
• A definition of the target segment the assessment is scoped to
• The threshold agreed in advance for what would count as fit for this product

Step by step

1. Define what fit means for this product before reading any evidence

   Product-market fit is not one universal number, and that is exactly why the bar must be set first. Decide which signals matter for your product's shape — retention curves for a habitual tool, renewal and expansion for enterprise software, repeat purchase for a marketplace — and what level each must reach to count as fit. Setting the bar after seeing the data is the assessment's original sin; every result can be argued into 'encouraging' once it is known.

   What good looks like: The signals and thresholds are written down, agreed, and dated before the evidence is assembled.

2. Scope the assessment to a specific market

   Fit is a relationship between a product and a particular market, so name the market: the segment, the use case, the buyer. A product can have strong fit with UK mid-market operations teams and none with the enterprise buyers the pitch deck promises. Assessing 'the product' against 'the market' in general produces a blended answer that is true of no one and actionable by no one.

   What good looks like: The assessment names the segment it is judging fit against, and other segments are assessed separately or explicitly excluded.

3. Read the behavioural evidence first

   Start with what users do, because behaviour lies less than sentiment: do they return without prompting, does usage deepen over time, do cohorts flatten into a retained core or decay to zero? Look at the shape of retention, not just the level — a curve that flattens says some group has woven you into their life; one that slides forever says the product is a visit, not a habit. Behavioural evidence sets the frame the other signals must be read within.

   What good looks like: Retention is examined by cohort and by shape, and the reader can see whether a durable retained core exists.

4. Weigh the pull signals against the push

   Distinguish growth you are buying from growth the market is giving you. Organic sign-ups, word-of-mouth referrals, inbound interest and users hacking around missing features are pull; paid acquisition, founder-led sales heroics and discount-driven conversions are push. A product can show respectable top-line growth entirely on push — which is precisely the situation a fit assessment exists to expose before scaling multiplies the cost of it.

   What good looks like: The growth picture is decomposed into pull and push, and the assessment says plainly which is doing the work.

5. Add the stated evidence, listened to sceptically

   Layer in what users say: interviews, churn conversations, and disappointment-style questions such as how users would feel if the product vanished. Treat stated enthusiasm as a weak signal unless behaviour corroborates it — people are generous in interviews and honest in their usage logs. The qualitative layer's real value is explanatory: it tells you why the behavioural numbers look the way they do, and which segment the strongest feeling comes from.

   What good looks like: Qualitative findings are used to explain behavioural patterns, not to overrule them, and are segmented by who said what.

6. Hunt the disconfirming evidence and segment everything

   Before concluding, actively look for what undermines the happy reading: the churned users' actual reasons, the cohort that never activated, the segment where every metric sags. Then break every aggregate apart — fit hides in averages. Strong fit in one narrow segment blended with none elsewhere produces mediocre overall numbers, and the correct response to that picture (focus) is the opposite of the response mediocre averages usually get (broaden).

   What good looks like: The assessment contains evidence that resists the preferred conclusion, and no key metric is reported only as an average.

7. Deliver a verdict and the action it implies

   Judge the evidence against the pre-set bar and say one of three things: fit in a named segment, partial or emerging fit with specifics, or not yet. Then attach the consequence — scale what is working, focus on the segment showing pull, iterate the proposition, or stop — with an owner and a date to reassess. An assessment that ends 'signals are mixed but promising' has collected the evidence and declined the job.

   What good looks like: The verdict names a segment, follows from the pre-set thresholds, and comes with a specific next action and reassessment date.

Common mistakes

• The bar for fit is set after the results are in, so whatever the numbers show gets interpreted as nearly there. — Write the signals and thresholds down before assembling evidence, and judge against that record. If the team cannot agree a bar in advance, that disagreement — not the data — is the first thing to resolve.
• Averages hide the answer — blended metrics show mediocrity, when one segment has strong fit and the rest have none. — Segment every metric by cohort, use case and buyer type. The most valuable output of a fit assessment is often the name of the one segment where the product is genuinely pulled.
• Stated enthusiasm outweighs behavioural silence — glowing interview quotes carry the verdict while retention quietly decays. — Let behaviour lead and words explain. If users say they love the product but do not return, believe the logs and use the interviews to find out what the affection is actually attached to.
• Push growth is mistaken for pull — sales heroics and paid acquisition producing numbers the market itself would never generate. — Decompose growth by source before reading it as evidence of fit. The question is not whether the numbers rise but whether the market is doing any of the lifting.
• The verdict is a hedge — 'mixed but encouraging' — so scaling decisions proceed by momentum rather than evidence. — Force one of three conclusions: fit in a named segment, emerging fit with named gaps, or not yet. Each has a different consequence, and refusing to choose quietly selects 'carry on as before'.

Before you call it done

• Signals and thresholds for fit were agreed and recorded before evidence was gathered
• The assessment is scoped to a named segment, not the market in general
• Retention is analysed by cohort and shape, showing whether a durable core exists
• Growth is decomposed into pull versus push, with the balance stated plainly
• Qualitative evidence explains the behavioural picture rather than overruling it
• Disconfirming evidence was sought and is presented, with no metric reported only as an average
• The verdict names a segment and comes with an action, an owner and a reassessment date

Frequently asked questions