What product-market fit actually is

Not a score and not a moment. Retention, unprompted return and how hard people fight to keep using it.

What is it?

Product-market fit means you have built something a specific group genuinely wants, evidenced by their behaviour rather than their compliments.

The honest signals: they come back without being prompted, they tell other people, they complain loudly when it breaks, and they would be genuinely annoyed if you took it away.

Why does a founder care?

Because almost everything else is downstream of it. Hiring, fundraising, marketing spend and scaling all assume PMF. Doing them before it just makes a wrong thing bigger and more expensive.

And because founders very reliably convince themselves they have it. Signups, press, a funding round and enthusiastic feedback can all coexist with nobody actually needing what you built.

Example

Two companies, both at 500 signups.

Company A: 500 signups, 40 weekly active, retention flat at 8% by week 8, no referrals, churn 12% monthly. Users are polite in interviews. This is not PMF, whatever the signup number suggests.

Company B: 180 signups, 95 weekly active, retention flat at 52% by week 8, a third arrived by word of mouth, and two customers emailed within an hour when the service went down.

B has far fewer users and unambiguously has fit. The tell is the retention curve flattening — it means a stable group has genuinely adopted it — and the fact that people noticed and cared when it broke.

The outage is genuinely one of the best signals available. Nobody emails about a product they were indifferent to.

The common mistake

First-time founders often measure signups and press, which are the easiest numbers to move and the least informative. Both can be bought.

The second mistake: treating PMF as binary and permanent. It is a matter of degree, it applies to a specific segment, and it can be lost — through a competitor, a market shift or your own changes.

The third: assuming more features produce PMF. If retention is flat at 8%, adding features usually moves it to 9%. The problem is normally the segment or the core value, not the feature count.

How it works

Step 1: Plot the retention curve by cohort

Group users by join week and follow each. You are looking for the curve to flatten rather than fall to zero.

Step 2: Check whether it flattens above zero

A flattening curve means a stable group adopted it. A curve reaching zero means nobody did, however high it started.

Step 3: Measure unprompted return

Usage without emails or notifications driving it. Prompted usage measures your marketing, not your product.

Step 4: Ask the disappointment question

'How would you feel if you could no longer use this?' A commonly used threshold is 40% answering 'very disappointed' — treat it as a signal, not a certificate.

Step 5: Watch what happens when it breaks

Silence during an outage is the clearest negative signal there is.

Step 6: Measure per segment

You may have strong fit with one group and none with another. The average hides the finding.

When to use this

Continuously from your first cohort onwards, and definitely before scaling spend or hiring.

When not to use it

Do not assess PMF with under about thirty users or fewer than eight weeks of data. Cohort curves need time to flatten, and early enthusiasm is not durable.

Do this now

Apply this to your own startup in My Full Journey (free account).