How to read a category instead of copying its leader

The first thing most people do with a category is open the top chart. It's the wrong place to start, and not because the leaders are uninteresting. They're just answering a different question than the one you're asking.

What the leader actually proves

A leader proves that the problem was worth solving, at some point, for someone. It rarely tells you what works now.

Look at when they shipped. Among the hundred most-installed apps in Communication on the US Play store, the median release year is 2015. In Productivity it's 2016, in Tools 2018. These apps were built for a store with fewer products, cheaper installs and different ranking behaviour. Whatever got them to the top is not a strategy you can run today, because the conditions that made it work are gone.

There's a second problem with the top chart: it's stable by construction. Install counts never decrease, so an app that stopped growing years ago keeps its position ahead of one that is growing fast right now. You're reading a monument, not a scoreboard.

"The category is growing" usually isn't about you

Category-level growth is the most misused number in this whole exercise. It hides who the growth actually reaches.

Take Puzzle games. Between two measurements of the US catalogue, ten weeks apart, the category added about 1.2 billion installs — a healthy-looking number. The median puzzle app in the same window grew 1.2%. Both are true. The first describes the category, the second describes what happens to a normal app inside it.

The gap has a shape, and it's steeper than most people expect. In every large category we measured, the top 1% of apps by absolute gain captured somewhere between 65% and 91% of all new installs: 65% in Puzzle, 82% in Tools, 87% in Communication, 91% in Productivity. Everyone else divides the remainder.

So the useful question isn't whether a category is growing. It's whether growth reaches past the first percentile — and how fast the apps in the second and third tiers are moving.

Where the informative apps sit

The instructive apps are the ones growing faster than the background while still small enough that their tactics are visible.

Start with the background rate. Across a panel of 583,000 apps present in both measurements, the median app gained about 4% installs in ten weeks. That's the floor. An app at +5% is standing still in relative terms; an app at +40% with a modest base is doing something specific, and it's usually possible to work out what.

Two filters make this list much shorter and much better:

• Drop the dormant. More than 400,000 apps on the US store haven't been updated in six months or longer. Plenty of them still rank. They look like competitors and behave like furniture.
• Drop the ones you can't learn from. An app growing because it's preinstalled, bundled with hardware, or attached to an offline business isn't running a strategy you can copy. It's running a distribution deal.

What's left is normally a handful of apps. Read their recent reviews, their release notes and what changed in their store listing over the last few months. That's a far denser source of information than the leader's landing page.

Limits

Category boundaries are fuzzier than any chart implies. When we cluster apps by what they actually do rather than by store genre, the US catalogue breaks into roughly 11,600 groups, and only about 4,700 of them contain enough apps to support any statistics. Your category may well be one of the small ones, and then percentiles mean very little.

Install counts also flatter everyone. They're cumulative and they never fall, so a category can look like it's expanding while active usage drains out of it. Growth in installs is evidence that people keep arriving, not that anyone stays.

And none of this tells you whether the second-tier app you're studying is actually making money. It tells you what's moving. Why it moves is a question the store data can't answer.

Figures come from our own dataset of the US Google Play catalogue, measured in late May and early August 2026.