Resource · The PMF Method

The 40% benchmark,
explained in plain terms.

Everything FitSignal measures is built on published, well-documented work: Sean Ellis’s disappointment survey and the 40% benchmark, and the process Rahul Vohra built around it at Superhuman in 2018. This guide explains how it works, where the 40% figure comes from, and what to do with your result.

< 25%
Keep searching

The market does not need this yet, at least not this audience. Change the segment, the problem, or the product before optimizing anything.

25–40%
Close: segment harder

Somebody here loves you. Find the persona already above 40%, focus the product on them, and let the rest go for now.

40%+
Strong signal: compound it

Most companies with strong traction scored above this line. Protect what your core loves and fix the blockers of the almost-convinced. A high score is encouraging rather than conclusive, so pair it with retention and repeat use.

The core idea

Measure disappointment, not satisfaction.

Satisfaction questions invite politeness. The Superhuman method instead asks users to imagine loss: “How would you feel if you could no longer use this product?” People who would be very disappointed have organized part of their work or life around you. That’s what fit actually is. Ellis defined the 40% mark after comparing results across nearly 100 startups, and he was candid that the line itself is “a bit arbitrary.”

Two habits keep the number meaningful. Survey users after real usage, meaning at least two product sessions, typically about two weeks in. And be deliberate about repeats: asking the same person again does not give you an independent reading, so a trend built on repeat answers is harder to attribute to product change. FitSignal applies the usage rule automatically and surveys each user once per project by default. Recurring waves are available as a setting when you want them, which is the normal configuration for NPS.

The instrument

Seven questions, each with a job.

Q1
How would you feel if you could no longer use it?
The PMF question. Very / somewhat / not disappointed. The only scored answer.
→ your score
Q2
Why did you choose that answer?
Context for everything else. The sentence behind the checkbox.
→ context
Q3
What would you use as an alternative?
Your real competitive set, usually not who you think it is.
→ competitors
Q4
What is the main benefit you receive?
From “very disappointed” users only: this is why people love you. Protect it.
→ love cloud
Q5
What type of person benefits most?
Your users describe your ideal customer better than you can.
→ ICP
Q6
How can we improve for you?
From “somewhat disappointed” users: the blockers worth fixing, ranked by the AI analysis.
→ roadmap
Q7
What’s your job title?
Feeds persona assignment, so every other answer can be segmented.
→ personas
After the score

The four-step improvement playbook.

1
Segment to find your beachhead

Slice the score by persona. Somewhere in your data is a segment already above 40%. That’s who you’re building for now.

2
Name what your core loves

Use the love cloud (Q4, very disappointed only). This is the main benefit. Every roadmap decision must protect it.

3
Fix the right blockers only

Listen to “somewhat disappointed” users who already want your main benefit, and politely ignore the rest. That’s what the linked blocker cloud and impact ranking compute for you.

4
Re-measure the next cohort

Ship, then survey newly eligible users. Watch the trend line rather than any single week’s number, and keep your eligibility rule stable so a move in the score means something.

Sources & further reading
  • Rahul Vohra: “How Superhuman Built an Engine to Find Product/Market Fit” (First Round Review, 2018). The original framework this product implements.
  • Sean Ellis: the original “very disappointed” survey question and the 40% benchmark, defined by comparing results across nearly 100 startups.
  • Fred Reichheld / Bain & Company: the Net Promoter® methodology behind our NPS® surveys.

Theory’s done. Measure for real.