Market fit before marketing

The product is live, users are scarce, and the first instinct is to buy ads. The logic is clear enough: nobody knows about us, so let’s tell them. Sometimes that’s right. More often, the problem isn’t that people haven’t heard of the product. It’s that the ones who did hear about it don’t come back. Then advertising becomes a pump filling a leaky bucket, which stays exactly as full as you keep paying for.

Why traffic doesn’t save you

A paid channel brings people at a cost that rises over time, because the cheapest and most interested arrive first. If one in a hundred is still around after a month, every retained user costs a hundred times the price of a click. No amount of ad optimization fixes that arithmetic.

Worse, traffic hides the problem. Signup charts go up, the team is happy, and the metric that actually describes the product, how many people came back after a week and after a month, stays where it was. The money runs out before that becomes obvious.

So the order is: first get a small number of people using the product and staying, then turn up the flow.

Signs that fit is emerging

Market fit doesn’t switch on one day. It shows up gradually, and here’s how you can see it.

The retention curve flattens. Take a cohort, everyone who signed up in the same week, and look at what share is still active after one, two, four and eight weeks. Without fit, the curve slides toward zero. With early fit, it drops at first and then levels off: some share of people stays and keeps staying. The height of that plateau depends on the category. What matters is that it exists.

People would be upset if the product disappeared. A method popularized by Sean Ellis: ask active users how they would feel if the product stopped existing tomorrow, with three answers: very disappointed, somewhat disappointed, not disappointed. Watch the “very” share. A commonly cited benchmark is around 40%, but it’s more useful to track whether that share grows from version to version, and who exactly answers that way. Those people are your market.

Organic pull. People arrive because someone told them, not because they saw an ad. Ask at signup how they heard about you, and watch whether “from a friend” and “found it myself” are growing.

Workarounds. Users export data to a spreadsheet because the report they need doesn’t exist, email support asking for something to be done by hand, or stitch your product together with two others into something you never planned. It’s annoying, but it’s a good sign: they need it enough to put up with the friction.

Cheap experiments

Until you have fit, money is better spent on finding out faster where it is.

Ten interviews. Not a survey but a half-hour conversation with people who have the problem you solve. Ask about the past, not the future: how did they handle it last time, what did it cost, what was annoying. “Would you buy this?” is nearly useless, because people are polite. Ten conversations are usually enough to start hearing the same complaints.

A manual version. Before automating, deliver the service by hand to five customers. Matching, reports, recommendations: do them yourself with a spreadsheet and email. It’s slow, but you’ll see what people will pay for and what they actually need, without writing code you’d later throw away.

A pricing test. Show a price earlier than feels comfortable. A pricing page with a pay button, even if the button leads to a waitlist for now, tells you more than ten interviews: interest that vanishes at the sight of a price was never demand. Try different prices on different segments, and look not only at conversion but at who pays.

One channel at a time. Once retention shows up and it’s time to bring people in, pick one channel and take it to a clear result: what does a user who’s still around after a month cost? Five channels in parallel, each on a small budget, give you five unreadable signals.

What to measure from day one

The list is short, and you can build it into even a simple app:

• weekly signups and the source of each;
• the key action, the thing the product exists for, and how many people perform it;
• cohort retention: the share returning after 1, 4 and 8 weeks;
• revenue and revenue per user, if the product is paid;
• reasons for leaving, even if it’s one field on the cancellation form.

Everything else, from heatmaps to twenty-step funnels to attribution models, will be useful later, once there’s something to optimize.

In short

Marketing accelerates what already works. If the product doesn’t retain people, it accelerates spending. First a plateau on the retention curve and people who’d be sorry to lose you, then an acquisition budget.

SUBXCAT can help with that path: setting up measurement, running interviews and experiments, and finding a channel that pays for itself.