The Sean Ellis Test
A one-question survey that reveals product-market fit before your retention data can
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 4
- Confidence
- 97%
Ask a random sample of engaged users a single question — how disappointed they would be if they could no longer use your product — and count the share who answer 'very disappointed.' Roughly 40% or higher signals product-market fit. It is a leading indicator you can read on day one, long before retention cohorts mature.
Origin
Created by Sean Ellis while consulting for early YC-connected startups. It began as a way to filter customer feedback to only people who genuinely cared; the 40% threshold emerged from observing a pattern across many Silicon Valley startups he shared the question with. He first used it at Xobni (Inbox backwards) because senior-management customers gave lukewarm satisfaction scores, and flipping the question to loss-aversion produced more honest answers.
Core principles
- 01Product-market fit means someone would genuinely miss your product if it disappeared
- 02Loss-aversion framing ('how would you feel if you could no longer use this') gets more honest answers than satisfaction framing
- 03The score is a leading indicator; retention is the lagging, ground-truth confirmation
- 04Only survey people who have actually experienced the product's value
- 05To be a must-have, a product must be both valuable AND unique — commodity use cases produce 'somewhat disappointed' answers because an easy alternative exists
How to run it
- 1
Take a random sample of activated users
Survey people who have hit your activation moment — used the product two-plus times, ideally within the last week or two so they haven't churned. Not homepage visitors, not fresh signups, not demo-watchers.
Pro tip When measuring the effect of an onboarding change, only survey the cohort that went through the new onboarding, so it acts as a clean experimental group.
- 2
Ask the core question
Ask: 'How would you feel if you could no longer use this product?' with choices very disappointed / somewhat disappointed / not disappointed / N.A. I've already stopped using it.
Watch out Don't run it on one-off products (a movie, a single workshop) — 'how would you feel if you could no longer attend the workshop you just attended' is nonsensical. Use NPS as your filter question there instead.
- 3
Read the 'very disappointed' percentage against ~40%
40%+ very-disappointed is the fit signal. Treat the number as a rallying target the whole team aligns on before pouring fuel on growth, not a precise scientific line (39 vs 41 is noise).
Pro tip Culture shifts the bar — Nubank uses 50% (Brazilians answer generously); a pessimistic market like Hungary might make 30% good enough.
Watch out Needs a real sample. ~30 responses is a workable minimum; four of ten very-disappointed is directional but not something to go to market on.
- 4
Confirm with retention cohorts while you grow
Once the score clears your target, start building growth but watch retention cohorts. If you are churning out the very people who said they'd be very disappointed, retrench and re-examine whether you truly have fit.
Watch out A high score late in a heavily-invested journey (e.g. after a user has built out a website or set up an event) can be inflated by switching costs, not pure utility. BlackBerry-style, a must-have attribute (the keyboard) can also vanish overnight when the market shifts.
In the wild
Sean committed to six months of growth work, ran the test, and got back only 7% very-disappointed. Rather than quit, he used the signal to move the score (see the 'Move the Score' framework).
→ Reached 40% in two weeks, 60% six months later; the company hit a billion-dollar valuation years later.
Sean expected a low score for what he saw as a commoditized website builder facing Wix and Weebly, but it returned ~90% very-disappointed — the highest he'd seen — driven by the investment users had sunk into building their sites (the Hooked 'investment' effect).
→ The signal still guided the work; the previously-flat business resumed significant growth over the next 12 months.
Common mistakes
Surveying the wrong people
Running it on homepage visitors, fresh signups, or people months past use distorts the reading. You want recently-active users who reached the aha moment.
Treating 40% as a finish line
Hitting the number isn't 'we have fit, go go go.' Its real power is diagnostic — it tells you who your must-have users are and points you at what to double down on.
Using a satisfaction question instead
Satisfaction framing produces lukewarm, polite answers; the loss-aversion framing forces people to reveal whether the product is actually a must-have.
Is it for you?
Best for
Early-stage founders and growth leads with at least an MVP and a modest base of activated users who need to know if they have product-market fit before scaling spend
Not ideal for
One-off / single-consumption products (a movie, a one-time event, a workshop) where 'no longer use it' is meaningless — use NPS there
From the transcript
“how would you feel if you could no longer use this product and I give them the choice very disappointed somewhat disappointed or uh even…”
“if you 40 40% or more of people say they'd be very disappointed if they can no longer use the product you essentially have product…”
“I would say it's a leading indicator of product Market fit the the the lagging indicator is do they actually keep using it”
“what I recommend is a random sample of people who've uh really used your product”
“to be a must have it needs to be both uh valuable and unique”
From the episode
The original growth hacker reveals his secrets
Sean Ellis (author of “Hacking Growth”)