The Sean Ellis test asks users one question: 'How would you feel if you could no longer use this product?' with the answers very disappointed, somewhat disappointed or not disappointed. If at least 40% say very disappointed, you probably have product/market fit. Growth marketer Sean Ellis set the 40% benchmark after comparing nearly a hundred startups.
Key takeaways
- Ask: 'How would you feel if you could no longer use [product]?' Score the share who answer 'very disappointed.'
- 40% or more suggests product/market fit. Under 40%, the product isn't yet a must-have.
- Superhuman scored 22%, then 33% by focusing on its best-fit users, then 58% within three quarters.
- Survey people who've actually used the product recently, and segment the results. Averages hide your best customers.
- Re-run the test at least once a year. What fits today may not fit tomorrow.
It was 2017, and Rahul Vohra had a problem.
He and his engineers had been building Superhuman, "the fastest email experience in the world," since 2015. Two years of work. And they still hadn't launched.
The team wanted to launch. Rahul wanted to launch. But he couldn't shake the feeling that the product wasn't quite there.
He went looking for a way to measure that feeling. Most of what he read described product/market fit after launch, when it's often too late to fix a poor first impression. Then he found a line from Marc Andreessen that summed up what he was living through: "You can always feel when product/market fit is not happening."
So Rahul turned the feeling into a number. He sent every pre-launch user a short survey.
Only 22% said they'd be "very disappointed" if they could no longer use Superhuman.
That number told him his gut was right. It also gave him something to work on. I tell the full story in Chapter Ten of the book, because it's the best example I know of turning product/market fit from a vague idea into a metric.
What is the Sean Ellis test?
The Sean Ellis test is a short survey that measures how much your users depend on your product. The key question is:
"How would you feel if you could no longer use [product]?"
- Very disappointed
- Somewhat disappointed
- Not disappointed
Superhuman added three open questions, and I'd use all four:
- How would you feel if you could no longer use [product]?
- What type of people do you think would most benefit from [product]?
- What is the main benefit you receive from [product]?
- How can we improve [product] for you?
A ready-to-copy version is in the Sean Ellis PMF survey template.
Where does the 40% benchmark come from?
Growth marketer Sean Ellis ran this question across nearly a hundred startups and compared the results. Companies that struggled to find growth almost always had fewer than 40% of users saying "very disappointed." Companies with strong traction almost always beat it.
So 40% became the line. It's not a law of physics. It's a benchmark drawn from experience, and a very useful one.
PMF score = Users answering "very disappointed" ÷ Total responses × 100
How to run the Sean Ellis test, step by step
Step 1: Pick the right people
Survey users who have actually experienced your product, recently and more than once. Not people who signed up and vanished, and not your mom. (She'll say "very disappointed" whatever you build.)
Step 2: Get enough answers
Aim for at least 40 responses before you read much into the score. Fewer than that is a conversation, not a statistic, so follow up with calls.
Step 3: Ask all four questions
The first question gives you the score. The other three tell you who your best customer is, what they value and what's missing.
Step 4: Score it, then segment it
Don't stop at the overall number. Split the answers by customer type, plan, industry, how they found you and how long they've been a customer.
Scoring a survey (illustrative)
You survey 120 recent users of a bookkeeping subscription.
- 42 say very disappointed
- 50 say somewhat disappointed
- 28 say not disappointed
Overall score: 42 ÷ 120 = 35%. Below the line.
Now segment. Among the 50 tradespeople who responded, 26 say very disappointed: 52%. Among the 70 other businesses, only 16 do: 23%.
You don't have a product/market fit problem everywhere. You have fit with tradies and a vague offer for everyone else. That's a much more useful finding than "35%."
Score your Sean Ellis survey
Paste your survey answers into the free Product/Market Fit Survey Scorer and get your 'very disappointed' score, segment breakdown and next steps.
What Superhuman did next: the four-step engine
The survey was only the start. Here's how Rahul used it, as described in the book and in his own account to First Round Review.
1. Segment to find your supporters. Rahul looked at who answered "very disappointed" and built a detailed persona of them. Focusing only on users who matched that persona lifted the score from 22% to 33% straight away, without changing the product.
2. Find out why people love it. He asked the fans what the main benefit was. For Superhuman, the fans loved its speed, focus and keyboard shortcuts, so that's where he doubled down.
3. Find out what holds the others back. He looked at the "somewhat disappointed" users whose main benefit matched the fans' and asked what stopped them from loving it. He set aside the users who wanted something else entirely.
4. Split the roadmap. Half the roadmap went to doubling down on what fans loved. The other half went to fixing what held the near-fans back. He published the roadmap so users could see the improvements coming.
Within three quarters, Superhuman's score reached 58%. For what happened to the company after that, read how Superhuman's obsession with fit ended in an acquisition.
The bit most people skip
The genius is in step 3. Rahul didn't try to please everyone who was lukewarm. He only listened to the lukewarm users who wanted the same thing his fans wanted. Everyone else was a different market. When you're small, ignoring the wrong feedback is as important as acting on the right feedback.
When the Sean Ellis test can mislead you
The test is a guide, not a verdict. Watch for these traps:
- Surveying only fans. If you only send it to your most engaged customers, you'll get a flattering number.
- Tiny samples. Ten answers can swing from 30% to 60% on one person's mood.
- Ignoring behavior. For a subscription business, the hard evidence is retention. If people say "very disappointed" but cancel after two months, believe the cancellations. Look for a retention curve that flattens out instead of falling to zero.
- Mistaking niche fit for a big business. Superhuman found fit with a passionate, premium-paying group. In July 2025 Grammarly agreed to acquire it, with Superhuman reported at about $35 million in annual revenue. Fit with a niche can build a valuable company without guaranteeing standalone scale against giants like Google and Microsoft.
How does the Sean Ellis test compare with other fit signals?
The survey measures what people say. Pair it with signals that show what they do. In the book I describe product/market fit as something your audience tells you, one way or another. Here are the main ways they tell you.
| Signal | What it shows | Strength | Weakness |
|---|---|---|---|
| Sean Ellis "very disappointed" score | How dependent users feel | Fast, cheap, works before launch | It's an opinion, and samples can be small |
| Cohort retention curve | Whether customers keep paying | Hard evidence of value | Takes months of data |
| Organic growth and referrals | Whether customers bring others | Shows real enthusiasm | Hard to separate from marketing spend |
| Customers asking to pay | Willingness to pay before you ask | The strongest early signal there is | Rare, and easy to miss |
| Shorter sales cycles | Whether buyers say yes faster | Clear in B2B | Needs a sales process to compare against |
GitHub is the classic "asking to pay" story. During its free private beta, users wrote in asking how they could pay for it. Years later, the same company found a second wave of willingness to pay with Copilot. See how GitHub's users who asked to pay now pay for AI by the credit.
For a subscription business, the retention curve is the signal I trust most. Plot each month's new customers and the share still paying at month 1, 2, 3 and onward. If the curve keeps falling toward zero, you don't have fit, whatever the survey says. If it flattens out, a core group has found something they won't give up. The math is on the churn rate page.
Using the test outside software
You don't need an app to run this. Change the wording to suit:
- A pilates studio: "How would you feel if you could no longer come to our classes?"
- An accounting firm: "How would you feel if you could no longer use our monthly bookkeeping service?"
- A coffee club: "How would you feel if the coffee club ended?"
Then segment exactly as above. In my experience, service businesses often find they have strong fit with one type of client and are merely tolerated by the rest. That's a pricing and marketing insight as much as a product one.
Customer Pain-Point AnalyserPaste reviews, emails or tickets and get pain points and jobs-to-be-done.Open the free tool →What to do next
- Send the four-question survey to recent, active users this week.
- Score it and segment it with the free Product/Market Fit Survey Scorer.
- Write a persona of your "very disappointed" customers.
- Split your roadmap between what fans love and what holds near-fans back.
- Re-run the test every quarter while you're working on fit, and at least once a year after.
For the bigger picture, including how Netflix, Looker, Patreon and GitHub knew they'd found fit, read product/market fit and the product/market fit examples.
Frequently asked questions
How many responses do I need for the Sean Ellis test?
There's no magic number, but I wouldn't read much into fewer than 40 responses. Below that, treat the survey as a structured conversation rather than a statistic, and follow up the interesting answers with calls. With a few hundred responses you can segment by customer type and still trust each group's score.
Can I run the Sean Ellis test before launch?
Yes, and that's exactly what Superhuman did. Rahul Vohra sent the survey to every pre-launch user because he wanted to know whether he had fit before the first impression was made. You need people who have genuinely used the product, so run it on beta users, pilot customers or a waiting-list group you've let in early.
What should I do if I score below 40%?
Don't panic, and don't launch a marketing push yet. Segment the results to find which customers say 'very disappointed' and focus on people like them. Ask those fans what they love. Ask the 'somewhat disappointed' users who match that profile what's holding them back. Then split your roadmap between doubling down on the first and fixing the second.
Does the 40% benchmark work for B2B and service businesses?
It works as a guide anywhere customers can genuinely imagine losing you. In B2B, survey the people who use the product day to day, not only the buyer who signed the contract. For a service business, change the wording to fit ('if you could no longer use our bookkeeping service') and pair the result with your retention numbers.
How is the Sean Ellis test different from NPS?
Net Promoter Score asks how likely customers are to recommend you, which measures goodwill. The Sean Ellis test asks how they'd feel if you disappeared, which measures dependence. A product can be pleasant enough to recommend without being a must-have. For product/market fit, dependence is the more useful signal.
Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
Robert Coorey, MBA
Robert is the author of The Subscription Playbook and Feed a Starving Crowd and a Director at BPAG Accountants in Sydney. More about Robert.
Sources
- First Round Review, How Superhuman Built an Engine to Find Product/Market Fit
- Reuters via Cyprus Mail, Grammarly eyes inbox dominance with Superhuman deal (7 Jul 2025)
Facts and prices checked 27 October 2026. Spotted something out of date? Tell us.