Full Reference: The Sean Ellis 40% Must-Have Survey

May 13, 2026 - 7 min read

Reference companion to Lesson 5.1 · Your First Customer Is Not a Marketing Problem - who to survey and who to strip, the send email, the per-segment scoring math, the decision tree, the under-40% diagnostic table, the read-by-count guide, and the complete further-reading set. Read the micro-lesson first for the five questions and the score-it steps; return here when your result lands under 40% and you need the diagnostic table.

Who you survey, in full

You need enough responses from people who have used your product recently to spot a segment pattern; a few dozen is the floor. Pull the list from whatever you have:

The MVP database (sign-up table). For a Lovable, Bubble, or Supabase build, export users as CSV.
Your beta waitlist if it converted to active users.
The trial list if you ran paid trials.

If you only have ten users, that is fine. Treat anything under ten responses as directional only. Ten of ten “very disappointed” is a louder signal than 40 of 100. You are not running a peer-reviewed study; you are looking for a dividing line.

Strip out two groups before you send:

Anyone who signed up and never logged in twice. They never used the product, so the question is unanswerable.
Friends and family who you onboarded as moral support. They will all say very disappointed and tell you nothing.

What is left is your sample. Annotate each row with the user’s job title and company size before you send, so the CSV export later can be sliced by segment in one filter.

Send it

Email subject line that works in 2026: “Quick 90-second question about [product]”. Body, three lines:

Hi [first name], building [product] and trying to figure out who really uses it. Would you spend 90 seconds on this? [link]
No pitch. No follow-up. I read every response by hand.
Thanks, [your name]

Send the first batch to your largest user cluster. Re-send a few days later to anyone who has not opened. You will hit a 30-50% response rate on a list under 100, which is enough.

Score it, in full

Export the CSV. Pivot on Q1 by segment from Q5. You are computing one number per segment:

must_have_pct = ("Very disappointed" count) / (total responses excluding "No longer use it")

The “no longer use it” answers come out of the denominator. They are churned users, not should-be-paying users.

Pull three numbers:

Overall must-have %. The headline figure.
Per-segment must-have %. Slice by job title and by company size. One segment will almost always be higher than the average. That is your must-have segment.
Three verbatim Q2-Q3 quotes from your must-have segment. Paste them into a Google Doc. Those quotes are your persona description, your ad copy, and your cold-email subject line for the next lesson.

Three segment cards showing what the blended 28% average hides: B2B Marketers at 64% must-have clear the 40% line, Solo Founders at 14% and Agencies at 0% do not

The decision tree

%%{init: {'theme':'base', 'themeVariables': {'fontFamily':'Caveat, Patrick Hand, Comic Sans MS, cursive', 'primaryColor':'#fff5f5', 'primaryBorderColor':'#cc342d', 'lineColor':'#333', 'primaryTextColor':'#1a1a1a'}}}%% flowchart TD A[Run the 5-question survey] --> B[Overall must-have %] B -->|Under 25% | C[Stop. Product problem. Talk to 5 'very disappointed' users. Find what you missed.] B -->|25-40%  | D[Any segment over 40%?] B -->|Over 40% | E[You have a must-have user. Lesson 5.3 outreach starts here.] D -->|Yes| F[Target that segment. Rebuild the persona on those quotes.] D -->|No| G[Refine the product first. Re-run after the next release.] classDef redbox fill:#fff5f5,stroke:#cc342d,stroke-width:2px; classDef purplebox fill:#fbe9ff,stroke:#a855f7,stroke-width:2px; class C,G redbox; class E,F purplebox;

Re-run cadence: re-run while the must-have rate is climbing, and after every major release once it holds above 40% for two consecutive runs. If a re-run drops, read the “somewhat disappointed” Q2-Q4 verbatims first - the diagnostic is in there.

Reading a small sample honestly

The Sean Ellis 40% threshold is statistically directional at ≥ 10 respondents, useful at 20+, and segment-sliceable at 30+. Under 10 respondents your result is a hypothesis, not a verdict - with 6 “very disappointed” out of 10 the threshold says PASS, but the confidence band is wide enough that real demand could be 20% or 80%. Under 10, segment-slice math does not work and the 25-40% bands do not apply. Read your first-pass count directionally instead, out of your 4-6 onramp users:

0-2 “very disappointed”: directional NO. Book more user sessions before re-running.
3-4 “very disappointed”: directional MAYBE. Book 5-10 more users, re-run.
5+ “very disappointed”: directional STRONG YES. Advance to Lesson 5.3 but caveat your outreach decisions - the segment language is hypothesis, not verified.

Use an under-10 reading to prioritize the next outreach batch, not to advance into Lesson 5.3 with confidence.

What “under 40%” actually means

Under 40% means you have a product problem, not a marketing problem, and the Q2-Q4 verbatims tell you which one.

Pattern	Diagnostic	Fix	Re-entry point
You built for the wrong segment	The product works, but the people you onboarded do not have the pain. Your Q5 slice shows: one segment is at 55%, the rest are at 5%.	Stop selling to the audience and start selling to the segment.	Lesson 5.3 personal-network outreach to the right segment.
You built the right thing, but it is not finished	The Q3 verbatims are hedged (“it is nice to have,” “I would use it if it had X”). The main benefit answers lack conviction.	Go back into the build and finish the thing.	Schedule a Friday demo with the next release.
The pain is real, but your product is not the relief	The Q4 verbatims name a workaround that is already 80% of the job (a spreadsheet, an existing tool, a person they pay).	Either niche into the 20% the workaround does not cover, or pivot.	Lesson 2.5 validated-problem statement.
The product solves the pain, but the workflow is too long	Users say “very disappointed” but session logs show they bailed before the payoff. Funnel collapses between signup and the “30-minute save” moment.	UX cut, not a strategy pivot. Shorten the path to the first win.	Retest after shortening the funnel; re-run the 40% test after the next UX release.

When founders should skip the test

Condition	What to do instead
Under 10 users	Run Lesson 2.4 outreach (with the list-building method from 2.3 if you don’t already have a 30-name list) and book 10 more user calls before re-attempting the test. The test requires 10-30 users who actually touched the MVP to be meaningful.
Pre-launch	Use the Mom Test interview script instead. The 40% test asks “if you could no longer use the product” - if the user never used it, the answer is meaningless.

Advanced (optional)

Layering on segment isolation for 100+ users: After you run the 40% test once and close your first paid pilot, read Sean Ellis’s original Hacking Growth and the Superhuman PMF Engine. Both combine the 40% test with structured segment-isolation workflows. The main path in the lesson is enough for the Module 5 decision; the advanced version becomes relevant after your first 10 customers ship.