How to Measure Product-Market Fit (Beyond the Sean Ellis Survey)

The Sean Ellis Survey Is a Starting Point, Not a System

Sean Ellis proposed a simple test: ask users "How would you feel if you could no longer use this product?" If 40% or more say "very disappointed," you have product-market fit. The test is elegant. It is also dangerously incomplete.

I have seen this survey produce misleading results across 20+ consulting engagements with SMBs and startups, at a financial social media startup before its acquisition, and now on a GenAI squad at a major finance technology company. The failure mode is always the same: the survey tells you what a self-selected group of respondents says they feel. It does not tell you what your entire user base actually does.

At one consulting client (a B2B SaaS tool with ~3,000 users), 44% of survey respondents said "very disappointed." The team celebrated. Six months later, monthly retention had dropped from 68% to 51%. The users who filled out the survey were the power users. The users who were quietly leaving never responded.

The 40% threshold is a useful gut check. But PMF measurement requires behavioral evidence, not survey sentiment. Here are five signals that tell you whether product-market fit is real, sustained, and defensible.

Signal 1: Retention Curve Flattening

If you only measure one thing, measure this. A retention curve plots the percentage of a cohort that comes back over time. Every product loses users after signup. The question is whether the curve eventually flattens or keeps decaying toward zero.

A curve that flattens means a core user base is forming. Users who survive the initial drop-off are sticking. A curve that never flattens means every user is eventually leaving, which is the clearest possible evidence that you do not have product-market fit, regardless of what any survey says.

At the startup I worked at before the acquisition, we tracked retention obsessively. Early on, our D30 curve was still declining with no sign of a floor. NPS was fine. The Ellis survey passed the 40% threshold. But the retention curve told the real story: users liked the idea of the product more than they liked using it. After an onboarding overhaul, D30 retention stabilized at 31% and the curve flattened clearly at that level. That single chart became the most important artifact in the acquisition data room.

The metric to watch is period-over-period retention, not cumulative retention. If users who survived to week 4 are returning at 90%+ rates in weeks 5 through 8, the curve has flattened. You have a retained base. I wrote a complete guide on how to build and interpret this analysis: [Retention Curve Analysis](/insights/retention-curve-analysis-guide).

Signal 2: Organic Acquisition Rate

Products with real product-market fit acquire users without paying for all of them. Not because organic growth is morally superior to paid acquisition, but because organic acquisition is a behavioral signal. It means existing users are telling other people about the product, or prospective users are searching for the problem you solve and finding you.

Track the ratio of organic to paid new users over time. If that ratio is flat or declining while you scale spend, you are buying growth, not earning it. If the ratio is increasing (or if organic is growing independent of spend), users are pulling other users in.

At the startup, organic acquisition was 22% of new signups in month 6. By month 14, it was 41%. We had not changed our SEO strategy or referral incentives meaningfully. Users were sharing the product in group chats and forums because it solved a real problem. That shift in the organic ratio was a stronger PMF signal than any survey result we collected during the same period.

The distinction matters because paid acquisition can mask the absence of PMF for months or even years. If you turn off paid spend and signups drop to near zero, the survey may say users love you, but the market is not pulling you forward on its own.

Signal 3: Usage Frequency Deepening

Product-market fit is not a binary state. It deepens over time as users integrate the product more tightly into their workflow. Usage frequency deepening measures whether retained users are increasing their engagement, not just maintaining it.

Segment your active users by tenure (1 month, 3 months, 6 months) and compare their average sessions per week or core actions per week. In a product with real PMF, longer-tenured users use the product more frequently, not less. They have found more value over time and built habits around it.

At the fintech company where I work now, users in their first month average 3.1 sessions per week. Users in their sixth month average 4.7 sessions per week. That deepening curve is one of the strongest PMF indicators in our entire metrics stack, because it means the product becomes more valuable with continued use, not less.

The opposite pattern is a red flag even when retention looks stable. If your retained users are gradually using the product less frequently, they are drifting toward churn. Retention might hold at 60% for several quarters, but if usage frequency is declining within that retained base, the floor will eventually give way. I covered the mechanics of measuring usage frequency trends in [usage frequency analysis](/insights/usage-frequency-analysis), but the PMF signal is simple: retained users should use the product more over time, not the same amount or less.

Signal 4: Willingness to Pay

This signal applies whether you are pre-revenue, freemium, or already charging. Willingness to pay is the behavioral test of whether users value your product enough to exchange money for it, and it is one of the clearest separators between "users like this" and "users need this."

For pre-revenue products: run a pricing page test or a "buy now" button before you build billing. Measure click-through rate. At one consulting client, we put up a pricing page with three tiers before any payment infrastructure existed. The click-through rate on the "Start Trial" button was 11.3%. That number, combined with retention data, gave us more confidence in PMF than three months of user interviews.

For freemium products: track the free-to-paid conversion rate over time and by cohort. A rising conversion rate in successive cohorts means your product is getting better at demonstrating value worth paying for. A flat or declining rate means free users are satisfied staying free, which means the paid tier is not solving a problem worth paying to solve.

For products already charging: monitor expansion revenue. Are existing customers upgrading to higher tiers, adding seats, or increasing usage-based spend? Expansion revenue above 120% net dollar retention is the gold standard signal that paying customers are getting increasing value. Below 100%, you are contracting, which means paying customers are getting less value over time regardless of what they tell you in surveys.

The reason willingness to pay matters as a PMF signal (distinct from revenue as a business metric) is that it filters out casual interest. Users will engage with free products out of curiosity. They will respond to surveys positively because they have no reason not to. They will only pay when the product solves a problem they cannot ignore.

Signal 5: Expansion and Word-of-Mouth Without Prompting

The final signal is the one you cannot manufacture: unprompted expansion and advocacy. This is distinct from organic acquisition (signal 2), which measures whether new users find you organically. This signal measures whether existing users voluntarily expand your footprint without being asked.

Examples of unprompted expansion: a user adds colleagues to the account without a referral incentive. A customer mentions you in a conference talk you did not sponsor. A prospect mentions in a sales call that they heard about you from someone at a different company. A user writes a blog post or forum thread recommending you without being part of an ambassador program.

These signals are harder to instrument than the other four, but they are not impossible to track. Tag inbound leads by source and monitor the "heard about you from someone" category. Track seat expansion that occurs outside of sales-initiated upsell motions. Monitor brand mentions in communities where you are not actively marketing.

At the startup, we noticed users in finance-related Reddit threads recommending the product unprompted. We had no Reddit strategy. We were not seeding those conversations. Users were recommending the product because it solved a specific problem in their workflow. That organic advocacy accelerated after we improved onboarding (and retention), confirming that the signals compound: when retention is strong and usage deepens, word-of-mouth follows.

This signal is the closest behavioral analog to what the Sean Ellis survey tries to measure. The survey asks "would you be disappointed." Unprompted advocacy shows you who is so un-disappointed that they actively recruit other users on your behalf. The behavioral version is harder to collect but impossible to fake.

Building a PMF Measurement System

These five signals are not independent. They form a causal chain:

1. Retention curve flattens (a core user base forms) 2. Usage frequency deepens (retained users get more value over time) 3. Willingness to pay increases (deeper value translates to economic commitment) 4. Organic acquisition grows (satisfied, paying users attract new users) 5. Unprompted expansion emerges (strong product-market fit becomes self-reinforcing)

If you are early stage, start with signal 1. If your retention curve is not flattening, the other four signals will not materialize, and no survey result changes that reality. If retention is healthy, layer on usage frequency and willingness to pay. Organic acquisition and unprompted expansion are lagging indicators that confirm PMF, not leading indicators that predict it.

The Sean Ellis survey is useful as a periodic temperature check, the same way [NPS has a limited role](/insights/nps-overrated-alternatives) as a sentiment overlay once your behavioral metrics are in place. But the survey cannot be the foundation. PMF measurement is behavioral. It is longitudinal. It requires cohort analysis, not cross-sectional snapshots.

The framework above does not require sophisticated tooling. It requires event tracking (which you already need for product analytics), cohort segmentation (which any modern analytics tool supports), and the discipline to look at trends over time rather than single-point metrics. If you need a [North Star metric](/insights/north-star-metric-framework) to anchor the system, choose the one that sits at the bottleneck of your current stage. For most early-stage products, that is retention. For products past initial traction, it is usage frequency or expansion revenue.

PMF Is a Gradient, Not a Threshold

The biggest conceptual mistake teams make is treating product-market fit as a binary state: you either have it or you do not. In practice, PMF is a gradient. You can have strong retention but weak willingness to pay (common in free consumer products). You can have high willingness to pay but declining usage frequency (common in enterprise products sold top-down where end users did not choose the tool). Each combination points to a different diagnosis and a different intervention.

The five behavioral signals above give you a multidimensional view. The Sean Ellis survey gives you a single number. For the most consequential strategic question a startup faces, one number is not enough.

---

I help teams measure product-market fit with behavioral data, not surveys. [lester@gradientgrowth.com](mailto:lester@gradientgrowth.com)