Your listing and the ASO score

The metadata page, the coverage matrix, density warnings, and what the 0-100 audit checks.

Read-only, on purpose

The metadata page shows the live Play listing exactly as Play returns it. The panel never writes to Play and never keeps a second copy of your text.

A savable draft was removed for that reason: with a draft in the panel and the real text in Play, nobody could tell which one the store actually shows. Analysis runs on the published text, which is the only version that ranks.

Listings are read once a night. Read now fetches immediately if you have just changed something and do not want to wait.

Coverage matrix

For your twenty strongest target keywords, the matrix shows whether each appears in the title, the short description and the long description. Matching is by root in the app's own language, so an inflected form counts.

Weight runs title, then short description, then long description. A term missing from all three is not being targeted at all, whatever its score says.

Density warnings

Repeating a keyword past a point stops helping and starts reading like spam to both users and Play. When a term takes up too much of the long description, it is listed with its count and percentage.

Density is only computed for terms that actually appear, because it is meaningless otherwise.

The ASO score

A 0 to 100 audit of the listing with an itemised list of what costs points, sorted by how much each one costs — so the top of the list is the next thing to fix.

Checks cover the things that can be read from a listing: title length and whether the strongest term is in it, short description length and coverage, long description length and structure, whether the listing has been translated for the store, ratings and screenshots.

The score is always computed from the last full read of the listing. Search results also leave a short record of each app without its descriptions, and scoring that record once dropped a listing from 81 to 28 overnight. If an app has not been read in full yet, the score says the listing has not been fetched instead of showing a number.

Your competitors' scores are computed the same way and shown next to yours. The comparison is more useful than the number: a 62 in a category where everyone scores 55 is a different situation than a 62 where everyone scores 80.

Try your copy before you publish it

Under the live listing there is a card called Try your copy. It opens filled with your published text, so you are editing what Play shows today rather than starting from an empty box. Change a title, short description or long description and it measures your version the way the coverage matrix measures the published one: which of your tracked keywords each field would cover, which ones you would gain, which you would lose, and a density warning if a term takes over the long description.

The comparison stays hidden until you actually change something. On a freshly opened card your candidate is the live text, so every difference would read ±0 — and a card that has not been touched would be indistinguishable from one that found nothing worth reporting.

It needs no AI key and it is drawn whether AI is on or off. With AI switched off it is the only writing aid on this page. Nothing is saved: not to Play, not to a draft table, not to your browser. You copy what you like into Play Console yourself.

Two things to read carefully. Coverage is weighted by popularity, and a tracked keyword whose popularity has not been measured counts as average rather than as zero — the card says how many did. The comparison is against the last full read of your listing, so if the listing has never been fetched in full there is nothing to compare against and the card says so rather than scoring you against an empty text.

The input limits are deliberately above Play's — 100, 300 and 8000 characters against Play's 30, 80 and 4000 — so you can paste something over the limit and see exactly how much Play would truncate.

Below the comparison, Try your copy lists tracked keywords with measured popularity that don't appear in the title or short description — matching word stems, so an inflected form counts as present — ordered by popularity, with your latest rank. Each row shows how many characters adding the missing words would take (words already in the field don't count) and whether that fits in the title, only in the short description, or nowhere.

Keywords whose popularity hasn't been measured yet are only counted, never recommended, and the list stays hidden until your live listing has been read in full.

Score trend

The ASO score is recorded once a night, and the Score trend card on the score page shows the last 90 days with the 7-day and 30-day change. Before this the score was recomputed on every visit and stored nowhere, so there was no way to tell whether editing your title had moved it.

Only the total is kept, not the breakdown of the individual checks. The checks and their weights change as the product changes, and a frozen breakdown would quietly disagree with the live card.

An app whose listing has not been fully read yet is skipped rather than written down as a zero — a zero derived from a partial record would say your text is bad when the truth is that the listing has not been fetched. History is kept for 180 days, and one recorded day draws no line: a trend needs a second night.

What a listing change moved

On the Activity page, each change to your own listing (title, short or long description, icon, screenshots) shows how your tracked keywords ranked in the 14 days before and the 14 days after it: the average rank change and how many keywords it rests on, keywords that entered or left the top 10, and the change in estimated visibility.

If fewer than 14 days have passed, the row says how many days are left; if there were no rankings before the change, or too few keywords were measured in both periods, it tells you why instead of showing numbers. When two listing changes fall within the same weeks, the row flags the overlap, because their effects can't be separated.

The row also carries a control group: over the same two windows and the same keywords, how far your competitors moved. Net is your movement minus theirs. When both move the same way, what moved was the category, not your text — and that is usually the honest answer. When competitors were ranked in both windows on too few keywords, the row says there is no control group rather than implying they stood still.

This is a before/after measurement, not proof that the change moved your rankings: Play, competitors and seasonality move rankings too. The control group narrows that gap without closing it.

Drafting with AI

Below the read-only text there is a workspace that asks your AI provider for an alternative title, short description or long description. The prompt spells out Play's rules — no exaggerated claims, no competitor names, character limits respected.

Generated text is never saved. You copy the one you like into Play Console yourself. Without an AI key the workspace is still there, and Generate answers with a note saying no key has been added; only switching AI off for the workspace removes it. Either way everything above it keeps working.

Each alternative is measured on our side and shows its length against Play's limit — 24/30, say — with a clear over-limit mark and its own Copy button. A model counting its own characters is often wrong. If the provider cuts an answer off at its length limit, the unfinished alternative is dropped rather than shown half-written.

→ Notifications and the inbox