Sean Ellis Test SOP

The Sean Ellis test is a one-question survey that tells you whether your product has reached product-market fit, and you can run it in one to two weeks. You ask active users "How would you feel if you could no longer use this product?" and count the share who pick "very disappointed". 40% is the line. Sean Ellis set it after benchmarking nearly a hundred startups: companies that struggled to grow almost always scored under it, and companies with strong traction almost always scored above it. This SOP shows you who to ask, the exact form, how to read the answers and what to change when you land short.

What is the Sean Ellis test and when should you use it?

The Sean Ellis test (also called the PMF survey or the "40% test") measures how much your users depend on you. Sean Ellis, who led early growth at Dropbox and LogMeIn and later co-wrote Hacking Growth with Morgan Brown, built it as a leading indicator: you learn about fit months before retention and revenue curves can prove it.

Its real strength is the follow-up questions. Rahul Vohra, founder of the email app Superhuman, described in First Round Review how his team started at 22% very disappointed, reached 33% by focusing on the right segment, and climbed to 58% within three quarters by building for the people who already loved the product.

Use this SOP:

• When you have real users, at least 40 people who used the product recently.
• When growth feels flat and you can't tell whether the problem is the product or the marketing.
• Before you raise money or spend on acquisition, because paying to bring people into a product they wouldn't miss burns cash.
• Every quarter after launch, to track whether product changes move the number.

What do you need before you start?

• A list of qualified users. Ellis recommends people who have experienced the product's main use, used it at least twice, and used it in the last two weeks.
• At least 40 expected responses. Vohra writes that results become directionally correct around 40 respondents. With a 20 to 30% response rate, that means emailing 150 to 200 users.
• A survey tool such as Typeform, Tally or Google Forms, and a spreadsheet for coding answers.
• Simple user data (role, company size, plan, signup date) so you can segment later without asking extra questions.

How do you run it, step by step?

1. Pick who gets the survey

Pull everyone who meets the three usage filters above. Leave out people who signed up and never came back: they can't miss something they never used. Survey each person only once, because repeat answers inflate the score.

2. Build the four-question form

Use the template below. Keep the main question word for word and the answer options in the same order. Changing the wording breaks the comparison with the 40% benchmark.

3. Send it by email from a real person

A short note from the founder gets more answers than an automated pop-up. Send it on a weekday morning, send one reminder after three days, and close the survey after ten days.

4. Calculate the headline score

Divide "very disappointed" answers by all answers, leaving out "I no longer use it". This is your score. Write it down before you look at any other column, so the number isn't softened by the comments.

5. Segment the very disappointed group

Sort answers by role, company type or use case. Look at who the very disappointed people are and recalculate the score for that group alone. This is the move that took Superhuman from 22% to 33%: the product already fit one group, and the average hid it.

6. Read the benefits and the blockers

Take the very disappointed group's answers to "What is the main benefit you receive?" and group them into themes. Then read the somewhat disappointed answers from people who name the same main benefit: these are the users you can win over, and their improvement requests are your next build list.

7. Decide and re-survey

Split your next product cycle between doubling down on what fans love and removing what holds back the near-fans. Survey new qualified users each quarter and track the score over time.

Template you can copy

PRODUCT-MARKET FIT SURVEY (2 minutes)

Email subject: Can I ask you 4 quick questions about [Product]?

Hi [first name],
You've used [Product] a few times in the last couple of weeks, so your
answer matters a lot to us. It takes two minutes: [link]
Thank you, [Founder name]

Q1. 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 [Product]

Q2. What type of people do you think would benefit most from [Product]?
Q3. What is the main benefit you receive from [Product]?
Q4. How can we improve [Product] for you?

SPREADSHEET COLUMNS
User ID | Role | Segment | Signup date | Q1 answer | Q2 | Q3 benefit theme |
Q4 request theme | Very disappointed (1/0)

Worked example: a scheduling app for dance studios

This example is illustrative. A founder runs a booking app for small dance and yoga studios with 260 active accounts. She emails the survey to 180 studio owners who logged in at least twice in the past two weeks and gets 54 answers (30%).

Overall, 17 of 54 say very disappointed: 31%, under the line. She segments by studio type. Among the 22 studios running kids' classes with recurring monthly billing, 12 say very disappointed: 55%. Among drop-in adult studios, only 5 of 32 do.

The kids' studios name the same main benefit again and again: parents pay automatically and nobody chases invoices. Nine somewhat disappointed owners name that benefit too and ask for sibling discounts. She narrows her marketing to kids' studios, ships sibling discounts, and plans a product usage analysis to see which features those 22 studios open every week.

What should you measure, and when do you stop?

Stop collecting when you pass 40 qualified answers and new responses stop moving the score by more than two or three points. If the score sits under 25% in every segment after two rounds and two product cycles, the problem sits deeper than features: go back to the problem statement validation SOP.

Which mistakes should you avoid?

• Surveying everyone who ever signed up. Inactive users drag the score down and tell you nothing about the product they skipped.
• Rewording the main question. "Would you recommend us?" or "Do you like us?" measures something else. The 40% benchmark only holds for the original wording.
• Reading the average only. The overall score hides the segment that already loves you. Always segment before deciding.
• Building everything the unhappy users ask for. People who say "not disappointed" are a different market. Ignore their requests for now.
• Running it with friends and early supporters. They answer to be kind and inflate the score.

How can AI speed this up?

Paste the answers to Q2, Q3 and Q4 into an AI assistant and ask it to group them into five to eight themes, with a count and two exact quotes per theme. Ask it to recalculate the score for each segment from your spreadsheet. Then read every very disappointed answer yourself: AI merges themes that sound similar, and the exact phrase a fan uses is what you put on your homepage.

If you want help turning the themes into a build list, the F/MS AI co-founder works through validation tasks with you step by step.

Where does this SOP fit in your validation plan?

The Sean Ellis test opens the Product-Market Fit stage, right after your first users arrive from a landing page test or a concierge test. Run these next:

1. Confirm the survey with behaviour through customer retention analysis: fans should also stay.
2. Compare with an NPS survey if investors or partners ask for a recommendation score.
3. Check whether fans actually bring others with customer referral analysis.
4. Use feature prioritization to rank what the near-fans asked for.

See all 10 Product-Market Fit SOPs. To build and test a startup with deadlines and feedback, the F/MS startup course takes you from idea to first users in four weeks, and the Fe/male Switch startup game lets first-time founders practise these steps in a safe setting.

FAQ

What is a good Sean Ellis score?

40% very disappointed is the benchmark Sean Ellis published. Scores between 30 and 40% mean you are close and segmenting will show where. Hiten Shah's open survey of 731 Slack users found 51%.

How many responses do I need for the Sean Ellis test?

Aim for at least 40 qualified answers. Fewer than 30 gives you a number that can swing ten points with a handful of answers.

Is the Sean Ellis test better than NPS?

They measure different things. The Sean Ellis test asks how much users depend on you, which fits early product-market fit work. NPS asks whether they would recommend you, which suits tracking mature products.

How frequently should I run the Sean Ellis test?

Once per quarter, to new qualified users each time. Running it monthly on the same people trains them to answer and makes the number meaningless.