How to Measure Product-Market Fit in SaaS, Honestly

Product-market fit is the most consequential thing a SaaS team can measure and the one most often judged by vibes. Founders declare it after a good month, investors ask for it in a pitch, and growth teams pour spend into a funnel long before the product has earned retention. The honest way to measure it is to combine two signals that fail in opposite directions: a survey that captures what users say, and a retention curve that captures what they do. Used together they are hard to fool; used alone each has a well-known blind spot.

This piece lays out both measures, how to instrument them without lying to yourself, and how to read the result as a continuous diagnostic rather than a trophy. The goal is a number you can defend to a skeptical board member and, more importantly, one that tells you where to spend next: on the product until fit arrives, or on the funnel once it has.

The 40% survey: what users say

The attitudinal half of the measurement is the Sean Ellis test, which asks active users one question — "how would you feel if you could no longer use this product?" — and counts the share who answer "very disappointed." The widely used benchmark is that 40% or more "very disappointed" indicates product-market fit. As one guide summarizes it, the threshold "is empirically correlated with sustainable growth" in Ellis's dataset (Stackmatix, Sean Ellis Test guide). That correlation is the whole value of the number: it is a cheap, fast proxy for a demand signal that otherwise takes months of retention data to confirm.

The survey's validity lives or dies on who you ask. Respondents must have used the product recently and experienced its core value, or the answers describe people who never engaged rather than people who did. Qualify on behavior — active in the last two weeks, past the activation milestone — before you send it. This is the same discipline that makes an activation metric honest: the population you measure determines whether the number means anything. A 55% result from unqualified signups is a vanity figure; a 38% result from genuinely activated users is a real, if sobering, signal.

The retention curve: what users do

The behavioral half is the cohort retention curve, and it is the more honest of the two because it cannot be talked up. Group users by the period they signed up, then plot the share of each cohort still active at day 1, 7, 30, 90 and beyond. A product with fit produces a curve that declines, decelerates, and flattens into a plateau — the retained core that keeps coming back. A product without fit produces a curve that keeps sliding toward zero, which no amount of top-of-funnel spend will fix, because you are refilling a leaking bucket. One PMF guide puts it plainly: a curve that "does not flatten" means the product "is not delivering enough value."

Read the shape before the number. A flattening curve at 25% can indicate better fit than a smoother-looking curve that is still declining at month six, because the flattening is the signal and the absolute level depends on your usage cadence. Hold the measurement definition steady while you read it — bounded versus unbounded retention, and which event counts as "active," change the curve's shape dramatically, which is why the choice of retention definition has to be fixed before you draw conclusions. The related day-7 retention diagnostic is a useful early tripwire, but it is a leading indicator, not a substitute for the full curve.

Why one signal is not enough

The two measures exist to catch each other's failures. The survey can pass while retention fails: a burst of enthusiasm from a narrow early-adopter niche produces a high "very disappointed" score even as the broader cohort curve slides, because the enthusiasts are a small slice who do not represent the market you are trying to win. Relying on the survey alone, you would declare fit and scale spend into a product that cannot hold the users it acquires — the most expensive mistake in SaaS growth.

Retention can also look adequate while the survey warns of fragility. A curve propped up by contractual lock-in, switching costs, or a lone must-have feature can plateau respectably while users would happily leave if a credible alternative appeared — exactly the sentiment a low "very disappointed" score exposes. Triangulating the two, plus organic referral volume, gives a picture no single metric can. This is the same logic behind choosing a north-star metric that pairs a usage signal with a value signal, and behind tracking leading indicators that move before revenue does.

Instrumenting the measurement

To run the survey repeatably, trigger it on a behavioral condition rather than a calendar — fire it the first time a user crosses your activation threshold and has returned at least once, so every respondent has genuinely experienced the core value. Store the response alongside the user's cohort and activation status so you can segment the 40% number by plan, acquisition channel, and signup month. A single blended score hides the story; the useful view is whether the segments you want to expand into are above or below the bar.

For the retention curve, the instrumentation requirement is a clean, consistent "active" event and stable cohort assignment. Decide once what counts as active — a meaningful action, not a login — and keep it fixed, because changing the definition mid-stream makes old and new cohorts incomparable. Build the curve on acquisition cohorts and watch newer cohorts against older ones: if recent signups retain worse than earlier ones, fit is eroding in the segment you are currently acquiring, even if the blended curve still looks fine.

Reading the result and acting on it

The point of measuring fit is to decide where the next dollar and the next sprint go. Below the bar — survey under 40% in your target segment and a curve that will not flatten — the answer is to keep working the product and resist the temptation to scale acquisition. Spending into a product without fit converts cash into churned users and buys you a worse version of the same diagnosis a quarter later. The measurement is giving you permission to not scale yet, which is often the most valuable thing it does.

Above the bar, the measurement licenses the opposite move: pour fuel on the funnel, because the product now holds the users it acquires and growth spend compounds instead of leaking. But keep the instruments running. Fit is not a finish line; expanding into a mainstream segment with different needs can pull a newer cohort's curve below the older one's long before aggregate revenue notices. Continuous measurement turns product-market fit from a one-time declaration into an early-warning system you can steer by.

A practical setup for this quarter

Stand up both halves in a single sprint. Add the one-question survey to your product, gated behind your activation event, and start collecting responses tagged by cohort. In parallel, define your "active" event, build the cohort retention curve, and plot the last twelve months of signups. Within a few weeks you will have one attitudinal number and one behavioral shape, segmented the same way, which together answer the question founders usually guess at.

From there, make it a standing ritual: re-run the survey quarterly on current actives, refresh the curve monthly, and always read the newest cohorts against the oldest. The discipline is cheap and the payoff is large, because every major growth decision — when to raise, when to scale spend, when to go back to the product — depends on an honest answer to whether you have fit, and honest answers come from measuring what users do, not only what they say.

Frequently asked questions

The questions below cover the Sean Ellis 40% survey, why it must be paired with a cohort retention curve, the plateau that signals fit, and how to measure product-market fit continuously rather than as a one-time event.