Your later users won't be as engaged or lucrative as your early users, and that's probably more of a feature than a bug

I've got inferential stats on the mind since I've been thinking a lot about experimentation at work, but it does in general make me wonder if this is a part of how growing companies get out in front of their skis.

In terms of analysis framing, this may be limited in its actionability since it can be hard to get statistical baselines on your target market. You mostly see TAMs discussed in terms of size, not in terms of the characteristics of their average member

Pivots in company direction could also be viewed through a stats lens. If you find out your product resonates with a different audience than your original ideal customer profile, perhaps you've discovered that you're sampling from a different population than you thought.

Perhaps product market fit simply means you've got a representative sample of your total addressable market?

In many cases, early adopters are outliers. Their behavior is more extreme than what your average customer will look like in the long term, which means your average and per user metrics eventually revert to the mean--the population mean.

If you think of your total addressable market as a (statistical) population, your product's customer base is a sample of that TAM. You continue to draw samples from that population as you grow, and over time, you get a better sense of what the population mean actually is

But something that's easy to miss about those enthusiastic early adopters is that they're not normal users, which I mean in both the colloquial sense and in the statistical sense.

It can be a nerve-wracking thing to watch--many a founder or startup leader has seen this trend and worried it means that users are becoming lower quality, or that devoted early adopters are losing interest.

If you've worked on an early stage or growing product, you've probably had the experience of watching your average and per user metrics gradually worsen as your user base grows.