Reading the scores

Popularity, difficulty, gap, opportunity and relevance — what each is, and why a blank is not a zero.

They are estimates

Google Play publishes no search volume. Every popularity number here is derived from observable behaviour: whether the term appears in autocomplete, how high it appears, how short the seed that surfaces it was, how many different continuations exist, and how many apps compete for it.

The values are relative to each other, not absolute. A tool that shows search volume as a count is guessing; read these as a ranking, not a measurement.

The five numbers

• Popularity — how much demand the signals suggest, 0 to 100.
• Difficulty — how strong the apps already ranking for it are, from their ratings counts. High means you are up against established apps.
• Gap — popularity minus difficulty. A quick read of where demand exceeds competition.
• Relevance — how well the term matches your own listing, with root matching in the app's language: an inflected form counts, filler words do not.
• Opportunity — popularity, difficulty and relevance combined for this app. A popular, easy term with nothing to do with your app converts nobody.

Popularity and difficulty are global: the same keyword has the same values for everyone. Gap and opportunity depend on the app and are computed as you look at them.

Relevance and opportunity are read against the last full read of your listing. Search results also carry a short record of your app without the descriptions; scoring against that would make a good listing look irrelevant.

A blank is not a zero

Difficulty stays blank until enough of the apps in the first ten have a known ratings count. This matters: measured across the pool, in the easiest difficulty band none of the top ten had a known ratings count, while in a hard band four in five did. Showing a number there would have measured our own data coverage, not competition, and labelled the most contested term in a category easy.

Popularity stays blank when the keyword has no suggestion evidence and has not been probed — typed in by hand, taken from a listing, from AI or from a competitor. It is unmeasured, not low. Gap and opportunity stay blank with it — on the Discover tab as well as on Tracked — because both are built on popularity.

Relevance and opportunity stay blank until your listing has been read in full once, rather than showing a zero.

The gap fills by itself. Discovered keywords are probed, and the nightly refresh works through the pool oldest-first.

Score detail and reporting

The badge on each row opens the breakdown: every signal, what it contributed and why. It is the fastest way to see whether a number rests on real evidence or on one weak signal.

If a value looks wrong, the flag on the row reports it. Long keywords once scored higher than short ones because of a formula error that ran for months, and it stayed hidden because users had no way to say so.

Recompute scores on the Opportunities page rescores everything you track from signals already stored. It makes no Play request; use it after a change rather than waiting for the night.

→ Rank tracking