The one metric I track for every product
Not MAU. Not revenue. Something simpler - and more honest.
Every product I ship, I give myself one metric to watch. Not a dashboard. Not a funnel. One number that tells me if the product is doing what it's supposed to do.
For GeoMinesweeper, it's games completed per session. For Strait Escape, it's return rate within seven days. For LottoBuddy, it's tickets scanned per user per week. Different products, different signals.
Why one
Because if you track ten things, you're watching noise. When everything moves slightly, nothing is telling you anything. When one thing moves significantly, you have a question worth investigating.
The discipline isn't in choosing the metric - it's in ignoring everything else until that metric changes meaningfully. That's the hard part. There's always another number that looks interesting. Most of them are interesting the same way weather radar is interesting: you can watch it for hours and still not know if it'll rain.
How I pick it
The question I ask is: what's the one behavior that, if it stopped happening, the product is clearly failing? Not the metric that would make an investor deck look good - the metric that reflects the thing the product actually does for users.
That question usually has a short answer. When it doesn't, the product probably doesn't have a clear enough job.
What I've learned from watching it
That most of what I build to improve the metric doesn't move it. And when something does move it, the cause is rarely what I predicted. The metric is a forcing function for honesty.