Free calculator · No signup

Product-Market Fit Calculator (Sean Ellis Test)

Enter how many people chose each answer to "How would you feel if you could no longer use the product?" See your score, how sure you can be of it and how it compares with the 40% benchmark.

Your survey answers

The question: How would you feel if you could no longer use the product? Enter how many people chose each answer.

Often worded as: not disappointed, it isn't really that useful.

Left out of the score, because these people can no longer imagine losing the product. Enter 0 if your survey did not offer this answer.

Your product-market fit score

Very disappointed

38.8%

Below 40%, but with this many responses the true figure could still be anywhere from 28.8% to 49.7%. Collect more responses and study the very disappointed group.

Your score, its likely range and the 40% line
Your scoreLikely range (95%)40% benchmark
Responses counted
80
Likely range (95%)
28.8% to 49.7%
Very disappointed answers needed for 40%
32 of 80
Somewhat disappointed
47.5%
Not disappointed
13.8%
Score if you count everyone who answered
36.9%

Short answer

The Sean Ellis test asks current users how they would feel if they could no longer use your product. Your score is the share who answer "very disappointed". Ellis treats more than 40% as the sign of a must-have product, and in his benchmark of about 100 startups, the ones that struggled to grow almost always scored below that.

Updated · Fonda editorial team ·How we write these pages

How to use the product-market fit calculator

  1. 1

    Very disappointed:

    how many people said they would be very disappointed if they could no longer use your product.

  2. 2

    Somewhat disappointed:

    how many chose the middle answer.

  3. 3

    Not disappointed:

    how many said they would not be disappointed, often worded as "it isn't really that useful".

  4. 4

    I no longer use it (optional):

    Sean Ellis's original survey offers a fourth answer for people who have stopped using the product. Enter that count here. It is shown for reference but left out of the score. Enter 0 if your survey did not offer it.

The result updates as you type. The headline is your share of very disappointed users. Below it you see a likely range for that share, a chart against the 40% line, how many very disappointed answers you would need to reach 40% with the same number of responses, and the score you would get if you counted everyone who answered.

The product-market fit score formula

Formula

Score = very disappointed / (very disappointed + somewhat disappointed + not disappointed). Likely range (95% Wilson score interval), with p = score, n = responses counted and z = 1.96: centre = (p + z x z / 2n) / (1 + z x z / n). Half-width = z / (1 + z x z / n) x square root of (p x (1 - p) / n + z x z / (4 x n x n)). Range = centre - half-width to centre + half-width.

Why "no longer use it" is left out. The question asks people to imagine losing the product. Someone who has already stopped using it cannot answer that, which is why Sean Ellis's survey gives them a separate N/A answer and is sent only to people who have used the product recently. His post does not spell out the denominator, so the calculator also shows the score with everyone counted. If the two differ a lot, many of your users have already left, and your churn rate deserves a look.

What the likely range means. You surveyed some of your users, not all of them. The range shows where the true share of very disappointed users could plausibly sit if you asked everyone. With 95% confidence, ranges built this way contain the true figure 95 times out of 100, as long as the people you surveyed were picked at random. The fewer responses you have, the wider the range.

The calculator uses the Wilson score interval rather than the familiar plus or minus formula (score plus or minus 1.96 x the square root of p x (1 - p) / n). The plus or minus version can give a range below 0%, or no range at all, when nobody or everybody says very disappointed, which is common with small samples. The NIST handbook notes that the Wilson lower limit cannot be negative and that its performance does not depend strongly on the sample size or the score.

A worked example

Say you surveyed the active users of a small budgeting app, as an example. 31 said very disappointed, 38 somewhat disappointed, 11 not disappointed and 4 said they no longer use it. These are the calculator's starting values.

Scroll sideways to see the whole table.

StepCalculationResult
Responses counted31 + 38 + 1180
Score31 / 8038.8%
Wilson centre(0.3875 + 3.8416 / 160) / (1 + 3.8416 / 80)0.3927
Wilson half-width1.96 / 1.048 x square root of (0.3875 x 0.6125 / 80 + 3.8416 / 25,600)0.1044
Likely range0.3927 - 0.1044 to 0.3927 + 0.104428.8% to 49.7%
Needed for 40%40% of 80, rounded up32 very disappointed
Score counting everyone31 / 8436.9%

The score is one response short of 40%, and the likely range runs from about 29% to 50%. So this survey cannot tell you whether the app is above or below the benchmark. The next step is to collect more responses and, more usefully, to look at who the 31 very disappointed users are.

What does the 40% benchmark mean?

In a 2009 post, Sean Ellis wrote that if over 40% of your users say they would be very disappointed without your product, there is a great chance you can build a successful business on it. If most answers are somewhat disappointed, the product is a nice to have, and he advised keeping spending low while you improve the core experience. First Round Review reports that he set the 40% line after comparing nearly 100 startups.

Treat 40% as a signal, not a pass mark. It is one survey question, and a score just above or below the line means little if your likely range is wide. It says nothing about whether the market is big enough or whether people will pay, which is what pricing research and a product-market fit review of revenue and retention are for.

How to run the Sean Ellis survey

  1. 1

    Pick the right users.

    Ellis recommends a random sample of people who have experienced the core of the product, used it at least twice and used it in the last two weeks. Superhuman used the same rule: at least twice in the last two weeks.

  2. 2

    Ask the main question.

    "How would you feel if you could no longer use [product]?" with four answers: very disappointed, somewhat disappointed, not disappointed (it isn't really that useful), and N/A, I no longer use it.

  3. 3

    Add three open questions.

    Superhuman asked what type of people would most benefit from the product, what the main benefit is to you, and how the product could be improved for you.

  4. 4

    Collect enough answers.

    In a 2009 interview, Ellis said he looks for around 30 responses as a minimum. Superhuman's founder wrote that results become directionally correct at around 40. Use the likely range above to judge your own sample.

  5. 5

    Survey free and paid users apart.

    On his blog, Ellis advised that with a freemium product you should survey free and paid users separately.

If you do not have users yet, this test is not for you yet. Talk to potential customers first, using a guide to customer interviews, and come back once people have used a working version.

How to segment by the very disappointed group

The overall score matters less than who gives each answer. Rahul Vohra, Superhuman's founder, described a method in First Round Review that you can copy.

  1. 1

    Group the answers.

    Sort every response by its answer to the main question, and note each person's role or type from the first open question.

  2. 2

    Find who is very disappointed.

    Look at the types of people in the very disappointed group. Superhuman found founders, managers, executives and business development people, and set other types aside for a while. Its score for that narrower group rose from 22% to 33%.

  3. 3

    Learn what they love.

    Read the main benefit the very disappointed users name. That is the thing not to break.

  4. 4

    Fix what holds others back.

    Among somewhat disappointed users, Superhuman listened closely to those who named the same main benefit and politely set aside those who did not.

  5. 5

    Measure again.

    Recalculate the score for your chosen group as you ship changes. Put each group's counts into the calculator to compare them.

The very disappointed users are your best guide to your ideal customer profile. Their answers to "who would most benefit" often describe it in their own words.

What the calculator leaves out

The likely range only covers random sampling error. It cannot fix a biased sample: if only your keenest users reply, the score will be too high. It also assumes each person answered once. The calculator does not weight answers, compare segments for you or track your score over time, so keep a simple record each time you run the survey. A high score is a good sign, not a promise of growth.

Frequently asked questions

Should I count people who no longer use the product?

The calculator leaves them out of the score, because the question asks people to imagine losing something they still use. It shows the score with everyone counted as well, so you can report either. Say which one you use, and use the same one every time you repeat the survey.

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

Sean Ellis has said he looks for around 30 responses at minimum, and Superhuman's founder found results became directionally correct at around 40. More responses narrow the likely range. To plan a survey for a set margin of error, use the survey sample size calculator.

Is the Sean Ellis test the same as NPS?

No. Net Promoter Score asks how likely people are to recommend you on a 0 to 10 scale, while the Sean Ellis test asks how they would feel if they lost the product. One measures willingness to recommend, the other how much people depend on the product. You can run both, and the NPS calculator scores the first.

What should I do if my score is below 40%?

Do not spend more on growth yet. Find the people who did say very disappointed, learn what they have in common and what they value most, and improve the product for that group. Superhuman started at 22% and reached 58% about three quarters later by doing this.

Can I run the test before I launch?

Not in a useful way. The question only works for people who have used the product, ideally at least twice in the last two weeks. Before launch, test demand with interviews and pre-orders. Fonda's Validate phase has you talk to real customers before it gives you a go, refine or pivot Verdict.

Sources

  1. Sean Ellis, Startup Marketing: Free Customer Development Help, Survey.io (May 2009, checked 2026-11-20)
  2. Venture Hacks: Sean Ellis interview on product/market fit (December 2009, checked 2026-11-20)
  3. Startup Marketing: Getting to Product-Market Fit, with replies from Sean Ellis (July 2010, checked 2026-11-20)
  4. Rahul Vohra, First Round Review: How Superhuman Built an Engine to Find Product/Market Fit (checked 2026-11-20)
  5. NIST/SEMATECH e-Handbook of Statistical Methods: Confidence intervals for a proportion (checked 2026-11-20)

Related tools and guides

Add this calculator to your website

Free to use on any site. Paste this code where you want the calculator to appear, and keep the credit link.

<iframe src="https://fonda.co/embed/product-market-fit-calculator" title="Product-Market Fit Calculator (Sean Ellis Test)" width="100%" height="720" style="border:0;max-width:960px" loading="lazy"></iframe>
<p style="font-size:13px"><a href="https://fonda.co/tools/product-market-fit-calculator">Product-Market Fit Calculator (Sean Ellis Test)</a> by Fonda</p>