Most users don’t churn. They disappear.
After talking to a few founders running cohorts / courses, I noticed something strange.
People don’t usually make a clear decision to quit.
They don’t say ''this isn’t for me''.
They just stop showing up.
No message.
No signal.
No event.
Just.. silence.
The problem:
Most tools only show retention after the fact.
By the time you see the drop in a cohort chart - the user is already gone.
There was a moment where they drifted.
But nothing captured it.
So I started thinking:
What if the real problem isn’t retention -
but the fact that absence isn’t recorded?
Curious if others have seen this:
Do your users explicitly quit?
Or do they just silently disappear?
Trending on Indie Hackers
I've seen this exact pattern with store owners tracking customer behavior, and honestly it applies to SaaS users too.
The counterintuitive thing I learned: silence IS the signal. You just have to baseline what "normal" looks like per user first.
When I was building out engagement tracking, I noticed that active users who eventually stick around actually complain more. They file tickets, they ask questions, they tag you in stuff. The users who quietly nod along and never push back? Those are your ghosts-in-waiting.
The practical move that worked for me: track the delta between someone's average engagement cadence and their current behavior. If a user who logged in 4x/week drops to 1x, that's a screaming red flag — even though "1x per week" looks fine in aggregate.
You're right that cohort charts show this too late. But I'd push back slightly on "no signal, no event." The signal exists, it's just the absence of activity, which most tools don't treat as an event worth tracking. What does your current setup look like for measuring per-user engagement baselines?
This is a really good point. The baseline + delta framing makes a lot of sense, especially that it’s not just ''no activity'', but a change relative to what was normal for that person.
What I’ve been thinking about is slightly earlier than even that , not just fewer logins or actions, but the micro shifts before they show up as measurable drops.
Like:
Those don’t always register in typical engagement metrics, but they seem to be where the shift actually starts.
Right now I’m just looking at this manually with a few people - no real system yet, just trying to understand if that earlier layer is consistent enough to matter.