Product-Market Fit Survey

In Brief

The Product-Market Fit Survey is a single multiple-choice question that asks existing users, “How would you feel if you could no longer use this product?” with four answer options ranging from “very disappointed” to “no longer use.” If 40 percent or more of qualified respondents pick “very disappointed,” that is a strong quantitative signal that the product has achieved product-market fit for that segment. The output is a percentage score you can track over time, paired with follow-up questions that reveal what users value most and where the product is falling short.

Description

“Product/market fit” is a term coined by Marc Andreessen in his 2007 post The Pmarca Guide to Startups, part 4 — The Only Thing That Matters. Andreessen defined it as “being in a good market with a product that can satisfy that market” and described the lived experience of a startup that has it: customers buy as fast as you can build, usage grows as fast as you can add servers, and money piles up in the checking account. When you don’t have it, customers don’t quite get value, word of mouth is flat, sales cycles drag, and reviews are tepid. PMF is the single thing that matters because almost nothing else — team, market, product polish — can rescue a startup that lacks it, and almost nothing else can sink one that has it.

Having achieved product-market fit, you can safely say that you have proven the value hypothesis: the key assumption that underlies why a customer is likely to use your product. PMF is also a proxy for customer excitement — customers recommend you more frequently when you have it.

In 2009, Sean Ellis turned Andreessen’s qualitative description into a quantitative test. Ellis had helped run early growth at Dropbox, Eventbrite, LogMeIn, and Lookout, and noticed that the startups whose growth efforts compounded shared a common pre-condition: when he surveyed their existing users with a single question — “How would you feel if you could no longer use this product?” — at least 40 percent answered “very disappointed.” Below that threshold, paid acquisition usually struggled to compound; above it, marketing investment tended to scale. Ellis tested the survey across roughly one hundred startups before publishing the 40 percent benchmark in his Startup Pyramid essay.

The 40 percent number is a heuristic, not a hard cutoff. Rahul Vohra’s 2018 First Round Review account of how Superhuman built a “PMF engine” using the Sean Ellis test traces a deliberate climb from 22 percent → 33 percent → 58 percent over a series of release cycles by analyzing the qualitative follow-ups, identifying a “high-expectation customer” segment (a phrase Vohra credits to brand strategist Julie Supan), and aiming the product roadmap squarely at the things that segment said were missing. Brian Balfour’s essay The Never Ending Road To Product Market Fit argues PMF is not a single moment but a series of escalating tests — a leading-indicator survey, a leading-indicator engagement metric, a flattening retention curve, and a “trifecta” of growth, retention, and monetization — each of which must hold before scaling. Dan Olsen’s The Lean Product Playbook frames PMF as the top of a five-layer pyramid (target customer → underserved needs → value proposition → feature set → user experience), with the survey as one diagnostic among several.

This page covers the survey itself: how to qualify respondents, run the question, and read the score.

How to

Prep AI Prompt

1. Define the user qualification filter. Sean Ellis’s original guidance is to send the survey only to users who have (a) experienced the core of the product, (b) used it at least twice, and (c) used it in the last two weeks. Surveying casual or lapsed users contaminates the score with people who never had a chance to depend on the product. If your analytics platform doesn’t track these conditions, define them as best you can — first-week churn is normal and you don’t want to count it as “not disappointed.”
2. Decide standalone or embedded, then pick the channel and tool. A standalone single-question survey (in-app intercept, email link, or a dedicated form) keeps the PMF question uncontaminated by surrounding questions. Embedding it in a longer customer survey is fine if you place the PMF question early. Common channels: in-app prompt, email to active users, intercom-style chat. Common tools: Google Forms, Typeform, Sprig, the dedicated pmfsurvey.com template.
3. Set a target sample size. Aim for at least 30 qualified responses before treating the score as anything more than directional, and 100+ before treating it as a real signal. Below 30, a handful of answers swings the percentage by 10+ points.
4. Draft the follow-up question battery. Vohra’s account adds three follow-ups that turn the score from a number into a roadmap: What type of person do you think would benefit most from this product? (segmentation cue and Julie Supan’s high-expectation-customer framing), What is the main benefit you receive from this product? (the value proposition in your users’ words), and How can we improve this product for you? (the missing-feature gap, especially valuable from “somewhat disappointed” respondents). Add a fourth — Why did you choose that answer? — so you can read the qualitative pattern under the quantitative score.
5. Pre-commit to segmentation analysis. Decide before launch which segments you’ll cut by (acquisition channel, plan tier, role, company size, signup cohort). The headline percentage often masks one segment well above 40 percent and another well below — the high-expectation customer segment is typically the one to build for. Choosing the cuts upfront prevents post-hoc data dredging.

Analysis AI Prompt

If 40 percent or more of qualified respondents picked “very disappointed,” that is the threshold signal Sean Ellis found correlates with a startup that can scale acquisition and grow. Below 40 percent, the product is not yet a “must have” for enough of the surveyed segment, and pouring marketing budget on top is likely to be inefficient.

The headline number is the start of the analysis, not the end:

• Read the qualitative follow-ups by quantitative bucket. Compare the language “very disappointed” users use to describe the main benefit, against the language “somewhat disappointed” users use. The very-disappointed cohort tells you what the product really is in users’ minds; the somewhat-disappointed cohort tells you what’s missing for them to cross over.
• Segment. Cut the score by the segments you pre-committed to in prep. A 28 percent overall score that hides a 55 percent segment is a different problem than a uniform 28 percent — the first calls for narrowing the target customer; the second calls for fundamental product change.
• Identify the high-expectation customer. Per Vohra, the segment that scores highest on “very disappointed” is the segment whose expectations you should design for. Building for the median user dilutes the product; building for the high-expectation segment tends to lift the median anyway.
• Plan the next roadmap. Vohra recommends a roughly 50/50 split: half the roadmap doubling down on the benefits the very-disappointed cohort named, half closing the gaps the somewhat-disappointed cohort named.
• Track over time. Re-run quarterly. PMF is not a single moment but a series of measurements that should hold (and rise) as the product and audience evolve, per Brian Balfour’s Never Ending Road framing.

AI tools (Dovetail, Marvin, generic LLMs with the raw responses pasted in) can auto-categorize the open-ended follow-ups and surface the language patterns that distinguish “very disappointed” respondents from “somewhat disappointed” ones. That language analysis directly informs your positioning and marketing copy. AI can also auto-segment quantitative responses by acquisition channel, usage behavior, or demographic profile to find which specific segments have crossed 40 percent even when the overall score has not.