The product-market fit survey and the interviews that should follow it

The product-market fit survey asks your recent users one main question: "How would you feel if you could no longer use [product]?" The answers are "Very disappointed," "Somewhat disappointed" and "Not disappointed." You then look at the share who chose "very disappointed."

The question comes from Sean Ellis. According to Rahul Vohra's account of how Superhuman used it, Ellis benchmarked the survey across nearly a hundred startups and found that companies struggling to grow almost always had fewer than 40% of users answer "very disappointed," while companies with strong traction almost always had more. That 40% line is a practitioner's benchmark from his comparisons, not a statistical threshold. Treat it as a rough signal.

The score tells you where you stand. It does not tell you why, which is what the follow-up interviews are for.

Who to send it to

Survey people who have actually used the core of your product recently. Superhuman followed Ellis's recommendation to include users who had used the product at least twice in the previous two weeks. Vohra writes that you start to get directionally correct results at around 40 respondents.

Leave out people who signed up and never came back, and don't survey the same person twice. Superhuman avoided repeat surveys so it would not distort its tracking over time.

If you have no product or only a handful of users, you cannot run this survey yet. Talk to people who do the job your product is for instead. The customer discovery interview script is a better starting point at that stage.

The questions to ask

Superhuman's version had four questions:

1. How would you feel if you could no longer use [product]? (Very disappointed / Somewhat disappointed / Not disappointed)
2. What type of people do you think would most benefit from [product]?
3. What is the main benefit you receive from [product]?
4. How can we improve [product] for you?

Keep it that short. The three open questions give you the words people use and the reason behind their first answer.

How to read the results

Start by grouping every response by the answer to the first question. Hiten Shah's open research on Slack did this with 731 responses, of which 51% chose "very disappointed." He then compared what the must-have group said about benefits, integrations and improvements with what everyone else said.

Vohra's process went further:

• Very disappointed: Read their answers to question 2 to see who the product works for. Happy users often describe themselves. Read question 3 to find the main benefit.
• Somewhat disappointed: Split this group by whether they named the same main benefit. Those who did are close; read their answers to question 4 to find what holds them back.
• Not disappointed: Superhuman chose not to act on this group's requests, reasoning that they were far from loving the product.

When Superhuman first ran the survey, 22% of users said "very disappointed." Narrowing the analysis to the kinds of users in that group raised the figure to 33%, and after three quarters of product work it reached 58%, according to Vohra. That is one company's account of its own results, not an expected outcome.

Follow up with interviews

A survey answer is a sentence. You still need the story behind it. Shah's write-up recommends adding an opt-in to the survey and then talking to selected respondents. NN/g's guidance on surveys makes the same division: surveys suit questions about how many people hold an attitude, and questions about why belong to qualitative methods.

Suppose, as a hypothetical, you run an invoicing tool for small construction subcontractors. A third of respondents say "very disappointed," and many name "getting paid faster" as the main benefit. Several "somewhat disappointed" users named the same benefit but asked for lien waiver tracking. Pick a few from each group and ask:

• Very disappointed users: "Tell me about the last invoice you sent. What did you do before you used us?" You are learning what the product replaced and what would make them leave.
• Somewhat disappointed users who named the same benefit: "Walk me through the last time you needed a lien waiver. What happened?" You are checking whether the request describes a real, recurring job or a nice-to-have.
• Very disappointed users in a segment you did not expect: "How did you find us, and what were you using before?" A surprise segment may be a better market than the one you planned for.

Ask about specific past events instead of reactions to features. Our customer interview questions and feature request guide have wording you can reuse.

What the score cannot tell you

The survey measures the attitudes of the people who answered. It says nothing about users who ignored it, and a score built from a founder's friends or a very early community may look better than the market will be. It also cannot tell you whether a new segment will pay or how large it is.

Use the result to decide who to talk to next. If the interviews point to a segment you have few users in, you may need to reach people outside your user base. You could start from a directory page such as finance professionals in construction or describe the role in a people search. Instant Expert finds people whose work matches your question. You review them, it sends the invitations, and you pay for each call that gets booked.