---
title: "Product/market fit survey: questions, template and scoring"
description: "The Sean Ellis question, who to send it to, Superhuman's four questions, how to read the 40% threshold, and a survey template to copy."
url: "https://raconte.ai/en/blog/product-market-fit-survey"
---

# The product/market fit survey: the Sean Ellis test and the interviews that explain it

The Sean Ellis question, who to send it to, Superhuman's four questions, how to read the 40% threshold, and a survey template to copy.

October 22, 2026

Godefroy

Updated on November 22, 2026

Key takeaways

*   The product/market fit survey asks one question: how would you feel if you could no longer use the product? A product has found its market when at least 40% of active users answer "very disappointed".
*   Send the survey to people who have used the product at least twice in the last two weeks. Around 40 answers are enough to see the direction.
*   The score shows whether people would miss the product, never why. Three open questions give the why: who benefits most, the main benefit, and what to improve.
*   Read the "very disappointed" answers first. They describe the customer to build for and the benefit you must protect.
*   Typed answers to open questions come back short. In a five-minute voice interview, the people who agree to a follow-up explain their answer.

The product/market fit survey, also called the Sean Ellis test, asks your active users one question: “How would you feel if you could no longer use \[product\]?”, with three answers, “very disappointed”, “somewhat disappointed” and “not disappointed”. When at least 40% answer “very disappointed”, your product has found a market that would miss it. The score shows whether people would miss the product. It never shows why, which is what you need to decide what to build next. This page gives you the question, who to send it to, the follow-up questions that helped Superhuman take its score from 22% to 58%, how to read the results, and a template to copy. It ends with the step most teams skip: asking the “very disappointed” users to explain their answer out loud.

We build [Raconte](/en), a voice interview tool, so we have an opinion on that last step. The survey itself works fine as a plain form.

## The product/market fit survey in one question

Sean Ellis, the marketer who ran early growth at Dropbox and LogMeIn and later coined the term “growth hacking”, wrote the question to find out whether a startup was ready to spend money on growth. Here it is, with its answers:

> How would you feel if you could no longer use \[product\]?
> 
> *   Very disappointed
> *   Somewhat disappointed
> *   Not disappointed
> *   N/A, I no longer use \[product\]

Your score is the share of “very disappointed” among people who still use the product. Leave the “N/A” answers out of the count, since they come from people who have already left. Ellis set the threshold at 40% after comparing the results of nearly a hundred startups. The ones that struggled to grow sat below it, while most of those that grew fast were above it. He later called the number “a bit arbitrary”, so read 38% and 42% as the same result.

The question asks about loss, which is why it tells you more than “how satisfied are you?”. Plenty of people are satisfied with a tool they could replace in an afternoon. The “very disappointed” answer picks out the people who have no easy substitute.

## Who to send it to and when

An answer only means something when it comes from someone who has used the product enough to miss it. Rahul Vohra, founder and CEO of the email app Superhuman, [described his survey in detail](https://review.firstround.com/how-superhuman-built-an-engine-to-find-product-market-fit/) in First Round Review in 2018. He sent it to users who had used the product at least twice in the last two weeks. A good default comes down to three checks:

*   the person has reached the core of the product (sent emails, published a first report, whatever your product’s main job is)
*   they have used it at least twice
*   they have used it in the last two weeks

Leave out people who signed up yesterday, who cannot know yet, and people inactive for months, whose answers measure your churn. Vohra writes that results start to be “directionally correct” at around 40 respondents. Below that, one unhappy team can move the score by ten points. With only 100 eligible users, send the survey to all of them.

Send the survey to each user once, after a few weeks of real use, then run it again every quarter or after a major release with the users who became eligible since. The trend matters more than any single score, so keep the wording identical from one run to the next. If you already send an [onboarding feedback form](/en/blog/customer-feedback-form) in the first week, keep the two separate: nobody can miss a product they opened three days ago.

## The four questions of the Superhuman survey

Vohra kept Ellis’s question and added three open ones:

1.  “How would you feel if you could no longer use Superhuman?” (very disappointed, somewhat disappointed, not disappointed)
2.  “What type of people do you think would most benefit from Superhuman?”
3.  “What is the main benefit you receive from Superhuman?”
4.  “How can we improve Superhuman for you?”

Each open question has a job. When question 2 asks users to describe your best customer in their own words, the “very disappointed” ones tend to describe themselves. The answers to question 3 name the benefit you must never break. The answers to question 4 list what stands between hesitant users and “very disappointed”.

When Superhuman first ran the survey, in the summer of 2017, 22% of users answered “very disappointed”. The team singled out the very disappointed users and built a profile of them from their answers to question 2: busy professionals such as founders, executives and managers who spent hours a day in email. The score was 33% among the users who matched that profile, before anyone changed a line of code. Three quarters of product work later, it reached 58%.

## How to read the results

Read the three groups in this order.

1.  Start with the very disappointed. Look at who they are (role, company size, use case) and at the benefit they name. This group is your market. At Superhuman, their answers kept coming back to speed, focus and keyboard shortcuts.
2.  Then split the somewhat disappointed by their main benefit. Those who name the same benefit as the very disappointed are close to loving the product. Their answers to question 4 tell you what still holds them back. For Superhuman, that group mostly asked for a mobile app. The ones who value something else want a different product, so set their requests aside.
3.  Leave the not disappointed for last. Vohra ignored them entirely, since their requests would pull the product toward people who will never love it.

Then compute the score again on the segment that matches the profile. A product at 25% overall can reach 45% for one segment. Market to that segment and build for it. Superhuman split its roadmap in two: half went to what the very disappointed loved (more speed, more shortcuts, more automation), half to removing what held back the somewhat disappointed who valued speed.

For a benchmark outside Superhuman, Hiten Shah, co-founder of Crazy Egg and KISSmetrics, [surveyed 731 Slack users](https://hitenism.com/slack-product-market-fit-survey/) and found 51% would be very disappointed without it.

![How to read the three answer groups of a product/market fit survey](/.netlify/images?url=_astro%2Fread-results-en.G3lonSWp.jpg&w=1376&h=768&dpl=6b031bf5959953076155d961)

## Why the score alone misleads

The score raises the alarm when too few users would miss the product, and stops there. Four limits are worth knowing before you present it to anyone.

*   It measures a forecast. People guess how they would feel, usually badly. Justin Jackson, co-founder of the podcast host Transistor, [makes the case](https://justinjackson.ca/product-market-fit-survey) with his own story: he expected to be upset when Microsoft shut down the Sunrise calendar app, switched to Google Calendar and was fine within days.
*   The sample flatters you, for three reasons: only active users get the survey, the most engaged answer first, and users who already left have no say. Check the score against your retention curve before you celebrate.
*   The threshold fits young products best. Ellis drew it from startups. Jackson adds that the survey is less reliable on a mature product, where habit keeps people in place long after the enthusiasm has gone.
*   Typed answers to the open questions come back thin. Ask “what is the main benefit you receive?” in a text field and you get “speed”, “saves time”, or an empty box. Those words are enough to sort answers into groups. They never say what the person was doing when the product saved them time, or what they used before, the two things you need to write a positioning or rank a roadmap.

Rob Fitzpatrick, author of _The Mom Test_, gives the general rule behind the forecast problem and the thin answers: ask about what people did, never about what they would do. The open questions of the survey only half follow it. The follow-up interview covers the other half.

## The signals that prove product/market fit beyond the survey

The survey records a forecast: how users would feel without the product. To prove product/market fit, check that forecast against two signals of what they actually do.

*   Retention levels off. For each monthly cohort, plot the share of users still active one, two and three months after signing up. Brian Balfour, founder of the growth training company Reforge, [sums up the test](https://www.lennysnewsletter.com/p/how-to-know-if-youve-got-productmarket): if the curve “flattens off at some point, you have probably found product/market fit”. A curve that keeps sliding towards zero means people try the product and leave, whatever the survey says.
*   New users come without ads. They arrive through word of mouth, direct visits and search. Marc Andreessen, co-founder of Netscape, [described a product without fit](https://pmarchive.com/guide_to_startups_part4.html) in 2007: customers “aren’t quite getting value out of the product, word of mouth isn’t spreading, usage isn’t growing that fast”.

When retention and word of mouth agree with a score above 40%, you have a case you can show a board. When the score contradicts them, trust retention and word of mouth, since they record what people did.

Use the follow-up interview as a last check. When most of the very disappointed tell the same story (the same moment, the same tool left behind), the fit holds for a segment you can name. When every story differs, the 40% may come from politeness.

## Running the follow-up as a short interview

You get the best answers to questions 2 to 4 once you dig into what the person just said: “You said speed. When did you last notice it?” A form cannot follow up like that. It takes a conversation, though five minutes are enough.

Add one last question to the survey: “May we contact you for a five-minute follow-up about your answers?” Then interview the people who said yes, starting with the very disappointed, then the somewhat disappointed who name the same main benefit as the very disappointed. Keep the follow-up short and anchored in the past:

*   “Tell me about the last time \[product\] got you out of a problem. What were you doing?”
*   “What did you use before? What made you switch?”
*   “If \[product\] disappeared tomorrow, what would you do on Monday morning?”
*   For the somewhat disappointed: “You mentioned \[their request\]. When did you last need it, and what did you do instead?”

The Monday morning question asks for a concrete plan instead of a hypothetical feeling. People describe what they would do on Monday more honestly than how they would feel. If someone says “go back to Gmail, I guess”, that is a “very disappointed” worth less than the tick suggested. To push further on an answer without putting words in the person’s mouth, repeat their own word as a question or ask what happened next. These two [follow-up techniques for open questions](/en/blog/open-ended-questions-qualitative-research) fit easily in five minutes.

Interview one segment at a time and compare answers inside it. When eight founders out of ten describe the same moment (the inbox cleared before the first meeting, say), you have the sentence for your homepage, in your customers’ words. Those answers also belong in your [voice of the customer program](/en/blog/voice-of-the-customer), next to support tickets and sales calls.

```mermaid
graph TD
  A["Send the survey to active users"] --> B["Compute the score on active users only"]
  B --> C["Profile the very disappointed"]
  C --> D["Interview the ones who agreed to a follow-up"]
  D --> E["Interview the somewhat disappointed who share the main benefit"]
  E --> F["Split the roadmap: protect the benefit, remove the blockers"]
  F --> G["Run the survey again next quarter"]
  G --> B
```

In [Raconte](/en), you create one interview per segment from a short brief that lists these questions and the points to follow up on. Then you invite the respondents who said yes by email, or share a single link. Each person answers by voice when it suits them. When an answer stays vague, the AI asks for an example. You get back the transcript, the audio, a summary and the sentiment of each answer. Raconte does not group answers across interviews yet. To compare the answers of a segment, read the transcripts side by side, or let an AI assistant read them through the API or the MCP server. A five-minute interview costs $0.60 past the 60 free minutes each month.

## A product/market fit survey template to copy

The template has seven questions and takes two minutes to answer. Keep Ellis’s question first, so that no other question colours the answer.

#

Question

Answer format

Why it is there

1

”How would you feel if you could no longer use \[product\]?”

Single choice: very disappointed, somewhat disappointed, not disappointed, N/A I no longer use it

Gives the score

2

”What type of people do you think would most benefit from \[product\]?”

Free text

Describes your best customer in users’ words

3

”What is the main benefit you receive from \[product\]?”

Free text

Names the benefit to protect

4

”How can we improve \[product\] for you?”

Free text

Lists what holds hesitant users back

5

”What would you use instead if \[product\] were no longer available?”

Free text

Names your real alternative, including “nothing” or “a spreadsheet”

6

”What is your role?”

Single choice among your four to six main roles, plus “other”

Lets you segment without guessing from question 2

7

”May we contact you for a five-minute follow-up about your answers?”

Yes / No

Tells you who to interview

Questions 5 to 7 are our additions to the Superhuman survey. Question 5 shows who you actually compete with, which is often a spreadsheet or an old habit.

For the invitation, one line is enough: “We want to understand what \[product\] does for you. Seven questions, two minutes.”

![The seven questions of the product/market fit survey template](/.netlify/images?url=_astro%2Ftemplate-en.BG8vcMxy.jpg&w=1376&h=768&dpl=6b031bf5959953076155d961)

## Where to start

Pick the users who used your product at least twice in the last two weeks, send them the seven questions, and wait for 40 answers. Compute the score on active users, read the very disappointed first, and recompute it on the segment that matches their profile. Then call ten of the people who said yes to question 7, before you touch the roadmap.

If you want those follow-ups to run without booking ten calls, [Raconte](/en) runs them by voice. You write the brief, send the link, and read the transcripts as they come in. Every organisation gets 60 free interview minutes a month, with no card required, then you [pay per active minute](/en/pricing). Before your product has users, the survey has nothing to measure. Run [customer discovery interviews](/en/blog/customer-discovery-interviews) instead.

## Frequently asked questions

### What is the Sean Ellis test?

The Sean Ellis test is a one-question survey that asks active users how they would feel if they could no longer use the product: very disappointed, somewhat disappointed or not disappointed. When at least 40% answer “very disappointed”, the product has reached product/market fit. Sean Ellis, a growth marketer who worked at Dropbox and LogMeIn, set the threshold after comparing nearly a hundred startups.

### Is 40% still the benchmark?

Yes, 40% is still the threshold teams compare themselves to, though Ellis himself called it “a bit arbitrary”. Treat it as a guide rather than a pass mark: 38% and 42% are the same result. Your own trend from one quarter to the next, and the gap between your segments, tell you more than the distance to 40%.

### How many responses do I need?

Around 40 responses are enough to see the direction, according to Rahul Vohra, who used the survey at Superhuman. If you compare segments, aim for 40 per segment. With fewer eligible users than that, send the survey to all of them and treat the score as a rough indication.

### What is the difference between the product/market fit survey and NPS?

[NPS](/en/blog/net-promoter-score) measures how likely customers are to recommend a product, on a scale from 0 to 10. The product/market fit survey asks how they would feel if they lost it. Someone can recommend a tool they would replace without regret, so the two scores often disagree.

### Can I run the survey before launch?

No. The question only works with people who have used the product at least twice in the last two weeks. Before launch, interview potential customers about the problem itself and how they deal with it today.

### How do you prove product/market fit?

Put the survey next to behaviour. A score above 40% “very disappointed”, cohort retention curves that level off instead of sliding to zero, and new users who arrive by word of mouth rather than through ads make a solid case together. Then interview the very disappointed users: they will tell you which segment the fit holds for, and why.

### How do you test demand for a product?

Ask potential customers about the last time they had the problem, what they did about it and what it cost them. The people who already spend time or money on a workaround are your demand. A pre-order or a paid pilot proves more than enthusiasm, since compliments cost nothing. Run these conversations as [customer discovery interviews](/en/blog/customer-discovery-interviews) and keep each one on past facts.

Ready to launch your first interview?

Describe your interview, share the link: an AI agent runs it by voice with each participant. You get back the transcript and analysis.

[Try for free](/login)

## Continue reading

[![Customer feedback form: 4 templates by moment](/.netlify/images?url=_astro%2Fthumbnail.BylD6fd7.jpg&w=1200&h=630&dpl=6b031bf5959953076155d961)

September 27, 2026

## Customer feedback form: 4 templates by moment

Four customer feedback form templates to copy, for after a purchase, a support ticket, onboarding and a cancellation, with the reason for every question.



](/en/blog/customer-feedback-form)[![Customer discovery interviews: method and 15 questions](/.netlify/images?url=_astro%2Fthumbnail.w6TuGzWB.jpg&w=1376&h=768&dpl=6b031bf5959953076155d961)

September 24, 2026

## Customer discovery interviews: method and 15 questions

What customer discovery is, its four phases, 15 questions that keep people on facts, and how to hold a weekly interview cadence without drowning in logistics.



](/en/blog/customer-discovery-interviews)[![Open-ended questions: 45 examples for qualitative research](/.netlify/images?url=_astro%2Fthumbnail.x5EdDQ3f.jpg&w=1376&h=768&dpl=6b031bf5959953076155d961)

August 22, 2026

## Open-ended questions: 45 examples for qualitative research

A bank of 45 questions grouped by research goal, a qualitative questionnaire to copy, and how to write open-ended survey questions people answer.



](/en/blog/open-ended-questions-qualitative-research)
