Working · 2026-11PrimaryGrowth opsProcess

PMF survey and the 40% test

Ask users how they would feel if they could no longer use the product and treat 40% answering 'very disappointed' as the PMF bar.

Also called: Sean Ellis test, very disappointed survey

Claim and evidence

The claimWithin just three quarters of our work to improve the product, the score nearly doubled to 58%.

Superhuman · review.firstround.com · 2018-11-12

What the numbers show

Share of Superhuman users who would be very disappointed without the product

22%58%

40%

share answering 'very disappointed' that Sean Ellis found to be the threshold (a benchmark, not a result)

review.firstround.com

Why it works

Diagram, not data
Ask active users thedisappointmentquestionAt or above the bar:scaleBelow the bar: keepimprovingSegment answers andtrack over time
Ask active users thedisappointment questionAt or above thebar: scaleBelow the bar:keep improvingSegment answers and track overtime

The question measures dependence rather than satisfaction. Per Ellis's findings quoted by Superhuman, companies below the line struggled to grow.

How to run it

  1. Survey active users with the single disappointment question
  2. Count the share answering very disappointed
  3. Compare against the 40% benchmark
  4. Segment answers by user type
  5. Track the score over time

Before / after

Before
How do you like the product?
Submit
Satisfaction rating says little about need
After
How would you feel if you could no longer use the product?
Very disappointed
Somewhat disappointed
Not disappointed
Who would benefit most from it?
Submit
One question that measures dependence

Generalised wireframe of the pattern, not a copy of any company's screen.

Ways to run it

Design options to test, not promised results.

  1. Option ASingle-question pulse

    Send only the disappointment question to recently active users and count the top answer.

  2. Option BSegmented score

    Break results down by user type and focus on the segment closest to the bar, as Superhuman did.

  3. Option CRecurring survey

    Repeat the same survey on a fixed rhythm and watch the trend instead of one reading.

Timeline

  1. 12017 Superhuman — Used the survey as a PMF score and worked to raise it
  2. 22026-11 status checked: Working

Dated examples

  1. 2017

    Superhuman

    Used the survey as a PMF score and worked to raise it

    Score rose from 22% in summer 2017 to 58% within three quarters.

    “Within just three quarters of our work to improve the product, the score nearly doubled to 58%.”
    ✓ quote checked word-for-word on the page

    review.firstround.com · 2018-11-12

Does it still work?

Workingas of 2026-11

Still the most-used quick PMF check; the 40% line comes from Ellis's own sample, not a universal law. source

Risks

  • Small samples give unstable scores
  • Surveying the wrong segment hides the real signal

Sources

  1. review.firstround.com · First Round Review · Rahul Vohra · 2018-11-12
    “Ellis found that the magic number was 40%.”
    ✓ quote checked word-for-word on the page
  2. review.firstround.com · First Round Review · Rahul Vohra · 2018-11-12
    “Ask your users how they’d feel if they could no longer use your product.”
    ✓ quote checked word-for-word on the page

Change log

  • 2026-11-20 — harvest v0.1 (ops-ai)