Most founders talk about product-market fit like it's weather. It arrives or it doesn't. You feel it or you don't. That framing is useless when you're trying to decide what to build next quarter.
There's a better option. A product-market fit survey turns a vague feeling into a number you can track, segment, and improve. It takes about a day to set up and it costs nothing. And the output isn't just a score. It's a ranked list of what to build, who to build it for, and who to stop listening to.
Here's how to run one properly.
What is a product-market fit survey?
A product-market fit survey asks your existing users one core question: how would you feel if you could no longer use this product? The percentage who answer "very disappointed" is your PMF score. Above 40% is the widely used benchmark for having found fit.
The test comes from Sean Ellis, who ran early growth at Dropbox, LogMeIn, and Eventbrite. He surveyed close to 100 startups and noticed a pattern. Companies that couldn't find durable growth almost always scored under 40%. Companies with real traction almost always cleared it. It's not a law of physics, but as leading indicators go it's better than almost anything else you can measure in a week.
The reason it works is that it measures dependency, not enthusiasm. People will tell you they love your product to be polite. Very few people will claim they'd be devastated to lose something they barely use.
Who should you survey (and who should you exclude)?
Survey people who have actually experienced your core product, and nobody else. Signups who never activated will drag your score toward zero and tell you nothing useful about the product you actually shipped.
Sean Ellis's original screening rule is a good default. Include users who:
- Used the product at least twice
- Used it within the last two weeks
- Completed the core action, whatever that is for you (sent an email, published a page, ran a report)
If you've got fewer than 40 people who clear that bar, don't run the survey yet. You'll get a number that swings 15 points depending on who happened to answer. Go do customer interviews instead, then come back when you have a real base.
Practical target: 100 to 200 responses. That gives you enough to segment. Under 40 responses, treat the score as directional at best.
One more thing. Don't survey the people you personally onboarded and text every week. They'll answer for the relationship, not the product. If those users are a big chunk of your base, tag them and look at their responses separately.
What questions should the survey include?
Four questions. The first gives you the score, the next three give you the roadmap. Anything beyond that and your completion rate falls off a cliff.
Question 1: How would you feel if you could no longer use [product]?
- Very disappointed
- Somewhat disappointed
- Not disappointed (it isn't that useful)
- N/A, I no longer use it
Question 2: What type of people do you think would most benefit from [product]?
Open text. This is the one everyone skips, and it's the most valuable question in the survey. Your users will describe your ideal customer profile in their own words, better than any positioning workshop you could run.
Question 3: What is the main benefit you receive from [product]?
Open text. The answers become your homepage copy. Not the benefit you think you deliver, the one people actually cite when nobody's selling to them.
Question 4: How can we improve [product] for you?
Open text. Critical: you'll segment these answers by the response to question 1, which is where most of the value lives.
Keep it to these four. Rahul Vohra's team at Superhuman ran essentially this set, and the whole thing takes a user about 90 seconds. Resist the urge to bolt on NPS, a satisfaction scale, and three demographic dropdowns. Every extra field costs you responses, and responses are the whole point.
How do you actually send it?
In-app is best, email is fine, and a link on Twitter is worthless. You want responses from your real users, not from whoever's bored on a Tuesday.
The mechanics matter less than people think. Typeform, Google Forms, a simple embedded widget: all fine. What matters:
Timing. Trigger it after a user completes the core action, not on login. Someone who just got value from you is in a position to give you a real answer. Someone who just opened the app is thinking about something else.
Subject line, if you're emailing. "Quick question about [product]" outperforms anything that sounds like a marketing campaign. Send it from a real person, ideally the founder.
Length signal. Tell them it's four questions. People will start a survey they believe has an end.
Follow-up. One reminder, four days later, to non-responders only. That single reminder usually adds 30 to 50% more responses. Two reminders start to annoy people.
Typical response rates for in-app prompts to active users run 10 to 25%. Email to active users runs 5 to 15%. Plan your base size accordingly: if you want 150 responses from email, you need somewhere around 1,500 qualified users on the list.
What does your PMF score actually mean?
Above 40% means you have fit and your problem is growth. Between 25 and 40% means you have something real but it isn't sharp enough yet. Below 25% means the product or the audience is wrong, and shipping more features won't fix it.
Here's the part that gets lost. A low score isn't a verdict. Superhuman's first score was 22%. Rahul Vohra didn't shut the company down. He segmented the data, rebuilt the roadmap around two specific insights, and the score reached 58% within about three quarters. That's the actual point of the survey: it's a diagnostic, not a grade.
A few things to watch out for when you read your number:
| Situation | What it usually means |
|---|---|
| Score above 40% but flat growth | You have fit with a segment too small or too hard to reach |
| Score 25-40% with one loud segment at 60%+ | You're serving two audiences, pick one |
| Score below 25% across all segments | Wrong problem, wrong audience, or the product doesn't work yet |
| Score jumped 15 points with no product changes | Your sample changed, check who answered |
Track the score quarterly, not monthly. It moves slower than you want it to, and measuring it too often just creates noise you'll be tempted to react to.
How do you turn the answers into a roadmap?
Segment every open-text answer by how that person answered question 1. Then build for one group and ignore another. That single move is what separates a useful survey from a slide in a board deck.
Three buckets, three different jobs:
The "very disappointed" group. Read their answers to question 3 and find the common benefit. That's your product's real value proposition. Read their answers to question 2 and you've got your ICP in customer language. Now read their answers to question 4 and treat every request as high priority, because these are the people who already love you and you want to keep them.
The "somewhat disappointed" group. This is where the growth is. These people see something in the product but something's blocking them. Sort their question 4 responses into two piles: requests that would move them toward what the "very disappointed" group already values, and requests that would pull the product in some other direction. Build the first pile. Skip the second. Superhuman's version of this was to focus only on the somewhat-disappointed users who also cited the same main benefit as the very-disappointed group, which cut out a lot of noise.
The "not disappointed" group. Ignore them. This feels wrong and it isn't. They told you the product isn't for them. Building for them will make the product worse for the people who love it, and it almost never converts them anyway.
If you're mapping this against your broader positioning and go-to-market plan, it helps to have the whole picture in one place. A spreadsheet works, Notion works, and a planning tool like Foundra walks first-time founders through connecting customer research to the positioning and GTM sections it feeds. Whatever you use, the rule is the same: survey insights should change your roadmap document, or you wasted everyone's time.
When should you run your first one?
Run it once you have 40+ users who've hit your core action twice in the last two weeks. Before that, the number is noise and interviews will teach you more.
Rough timing by stage:
- Pre-launch or under 40 active users. Don't survey. Do 10 to 15 customer discovery interviews instead. You need texture, not statistics.
- 40 to 100 active users. Run it, but treat the score as a rough baseline. Focus on the open-text answers, which are useful at any sample size.
- 100+ active users. Run it properly, segment it, and set a quarterly cadence.
- After a major pivot or repositioning. Run it again eight to twelve weeks after the change, once people have had time to use the new thing.
One caveat worth naming: this survey works best for products people use repeatedly. If you sell something transactional, a wedding planning tool, say, or tax software, "how would you feel if you could no longer use it" doesn't map cleanly onto the buying behavior. You'll get better signal from repurchase rate and referral behavior. Know when the tool fits.
Key takeaways
- The PMF survey measures dependency, not enthusiasm. That's why it beats NPS as an early-stage signal.
- 40% "very disappointed" is the benchmark, but a low score is a starting point, not a death sentence. Superhuman went from 22% to 58%.
- Only survey users who've experienced the core product. Screening is what makes the number mean anything.
- Four questions, no more. The three open-text ones are where the roadmap comes from.
- Segment every answer by the question 1 response. Build for the somewhat-disappointed, protect the very-disappointed, ignore the rest.
- Run it quarterly, not monthly. And don't run it at all under 40 qualified users.
If you want more on the validation side of this, the Foundra key reads library has walkthroughs on customer discovery interviews and smoke tests that pair well with this survey.
FAQ
What is a good product-market fit survey score?
Above 40% of respondents answering "very disappointed" is the standard benchmark, based on Sean Ellis's survey of roughly 100 startups. Between 25 and 40% suggests partial fit with a specific segment. Below 25% usually means the product, the audience, or the problem needs to change.
How many responses do I need for a PMF survey?
Aim for 100 to 200 responses so you can segment meaningfully. You can get directional signal from 40, but below that the score swings too much to be useful. Prioritize response quality over volume: 50 responses from truly active users beat 300 from cold signups.
How is a PMF survey different from NPS?
NPS asks whether someone would recommend you, which measures social willingness. The PMF survey asks how they'd feel losing the product, which measures dependency. Early-stage users will recommend a product they don't really use. Very few will claim they'd be devastated to lose it.
How often should I run a product-market fit survey?
Quarterly is right for most startups. The score moves slowly, and measuring monthly produces noise you'll be tempted to overreact to. Also run one eight to twelve weeks after any major pivot, repositioning, or pricing change.
Can I run a PMF survey before launching?
No. The survey depends on people having used the product, so there's nothing to measure pre-launch. Use customer discovery interviews, landing page smoke tests, and waitlist conversion rates instead, then run the survey once you have 40+ users hitting your core action.
What if my score is under 25%?
Read the open-text answers before touching the roadmap. Look for any segment scoring meaningfully higher than the average: a job title, a company size, a use case. If one exists, that's your beachhead and the fix is narrowing your focus. If no segment clears 40%, the problem is more fundamental and you're looking at a pivot, not a feature sprint.
Top comments (0)