The Sean Ellis Test: How to Measure Product-Market Fit
The Sean Ellis test is a one-question survey that measures product-market fit. It asks recent, active users how they'd feel if they could no longer use the product. When about 40% answer "very disappointed," Sean Ellis found a product could grow sustainably. A lower score points to a gap in targeting or in the product to close before spending more on growth.
A city transit agency is thinking about cutting Route 14, so it leaves a card on every seat: how would you get around if this bus stopped running? Some riders write that they'd lose their way to work. Others write that they'd drive, or walk two blocks to the train.
Your product carries both kinds of rider. Signups and logins count them the way the fare box does, one tap each, so founders end up arguing about whether they have product-market fit. Guess wrong, and the growth budget goes to customers who'd shrug if you shut down. The Sean Ellis test leaves the card on the seat and counts the riders who'd be stranded.
How the Sean Ellis test works
Sean Ellis, a startup growth marketer, explains the method in a 2019 post, Using Product/Market Fit to Drive Sustainable Growth. The whole test rests on one question: "How would you feel if you could no longer use [product]?" Users pick one of four answers:
- Very disappointed: the product has become part of how they work.
- Somewhat disappointed: they'd miss it, then find a substitute.
- Not disappointed: they don't really need it.
- N/A, I no longer use it: they've already left.
The score is the share who answer "very disappointed." In his experience, a product can grow sustainably once it reaches around 40%. Below that line, the people who did say "very disappointed" show you where to look next.
Running the survey
A score is only as good as the people you ask. The transit agency puts its card on the bus this week, rather than mailing it to everyone who bought a pass years ago, and Ellis limits the survey to people who've used the product in the last couple of weeks, because anyone who never used it or drifted away months ago has already answered.
- Pull a list of users who've done the product's core action recently, ignoring signups who never got that far.
- Send the one question, plus a few follow-ups on why they'd miss it and who they think would benefit most.
- Wait for at least 30 responses before trusting the percentage. Ellis treats 30 as directional and 100 or more as solid.
- Sort the answers by segment, such as role or acquisition channel, before you read the total.
The follow-ups matter as much as the score. They tell you who your must-have customers are and what words they use for the value, and that's the raw material for your ideal customer profile.
Reading the score
The total is the headline, and the segments are the story. A flat 40% across every segment is a stronger result than 65% in one group averaged down by 20% everywhere else.
| Result | What it means | Next move |
|---|---|---|
| 40% or more overall | Fit is likely in the group you surveyed | Check that it holds across segments and channels |
| Under 40%, one segment above | Fit exists in a narrower market than you're selling to | Narrow your targeting to that segment |
| Under 40% everywhere | The product, or the problem, isn't must-have yet | Study the few "very disappointed" answers before building more |
| Under 30 responses | Too few to read as a percentage | Interview the "very disappointed" users one by one |
A low score isn't a verdict on the company. Ellis describes one business that started at 7% and reached 40% within a couple of weeks by changing its targeting, positioning and onboarding around what its must-have users said. Most of that fix sat in who the product reached and how it was explained.
When the survey isn't enough: the Sean Ellis test measures what people say. Check it against what they do. If your retention curve keeps falling toward zero, a good survey score is a polite answer from people who are leaving.
Sean Ellis test mistakes
The method is simple, so the errors come from how founders run it and what they do next:
- Surveying every signup: lapsed and never-activated users drag the score down and hide the fit that exists.
- Surveying only fans: a list of power users inflates the score just as badly.
- Reading one wave as a verdict: a single survey is a snapshot. Run it again after each real change.
- Scaling on a weak score: more ad spend on a product people wouldn't miss buys churn faster.
Deciding whether you have fit
Go back to Route 14. The agency doesn't need every rider to be stranded, only enough of the right ones, and it needs to know who they are. When 40% of your recent users would be very disappointed, spend on growth with confidence. When the number is lower, find the segment that already cares, and point your marketing there.
If the survey says the product works and growth still isn't coming, look at the go-to-market next. My free marketing diagnostic covers that side: who you're targeting and what you're telling them. Founder problem or funnel problem? helps you tell which kind of gap you're facing, and TAM, SAM, SOM sizes the segment once you've found it.
Count the riders who'd be stranded before you add more buses.
Common questions
Who created the Sean Ellis test?
The Sean Ellis test was created by Sean Ellis, a startup growth marketer who explains the survey and its benchmark in his own writing. He set the 40% line from experience: in his account, a product becomes able to grow sustainably once about 40% of its recent users would be very disappointed to lose it.
How is the Sean Ellis test different from NPS?
The Sean Ellis test and Net Promoter Score measure different kinds of attachment. NPS asks how likely a user is to recommend the product. The Sean Ellis test asks whether the user would miss it. A product can be pleasant enough to recommend and still be easy to live without, which is the gap the Ellis question exposes.
Can a B2B startup with few customers run the Sean Ellis test?
A B2B startup with few customers can run the Sean Ellis test, but the percentage won't mean much below about 30 responses. With a small sample, the founder learns most from the people who say "very disappointed": who they are and what they'd call the product's main benefit.
How often should a startup run the Sean Ellis test?
A startup should run the Sean Ellis test again after any real change to targeting or onboarding, and at least once a quarter while it's still searching for fit. Comparing waves on the same question shows whether the changes moved the score, and a single wave is only a snapshot.