One conversion rate, two completely different users
Hey,
I spent part of this week inside an app’s conversion data. The founder wanted to know why growth felt stuck. Installs were healthy. The conversion rate looked fine. Nothing screamed problem.
That “fine” was the problem.
A blended conversion rate is an average. And an average is two or more different things pretending to be one.
When I cut this app’s users by how deeply they used the product, the single number fell apart. Users who barely engaged, one or two actions in, converted poorly and churned fast. Users who went deep, who set the product up to actually run their work, converted at a far higher rate and almost never left.
Same app, same pricing. Two completely different populations inside one number.
The aggregate number wasn’t wrong. It was hiding one cohort that was succeeding and another already halfway out the door.
The line between the two groups had nothing to do with plan tier or how long someone had been a customer. It was depth of use. How far past the obvious first step they got.
That changes the question worth asking.
The blended conversion rate tells you almost nothing on its own. What actually tells you something is how conversion changes as users go deeper. Find the threshold where they start to stick, and every onboarding decision and every email you send has one job. Pull more users across it.
Pulling the data is the easy half. The harder half is reading it correctly. Knowing which depth metric actually predicts retention for your product, where that threshold sits, and what to change so more users reach it. That’s judgment, and it’s the part most founders never get to, because the blended number told them everything looked okay.
If you want that done for your app, that’s what a sprint covers, from the segmentation through to the onboarding changes that follow from it.
Ohad
