Metadata audit

Copying the same English text into every country means missing the real search terms there. The audit shows which field is empty in which storefront in one table; a gap that cannot be measured turns into work that never gets done.

Localization is not translation: it is rewriting in the words people in that market type into the search box.

The audit's output is recovered characters. What goes into the room is decided by the words you actually rank for - that is, by the cusp band.

You can run the same audit without an account, for a single metadata set, in the free tool.

Apple gives you a 100-character keyword field per app, and in practice half of it is spent. The reasons are dull and mechanical - which is exactly why all of it is recoverable. The audit does not estimate; it counts: which word repeats where, and how many characters come back.

The largest item is repetition. Writing a word that already appears in your app name into the keyword field a second time gains nothing; Apple has already indexed it. If your app name is long, you may be spending a real share of the field on repeats without noticing.

The second item is formatting. Every space after a comma is one character doing no work; across twenty words that is nineteen characters, which is another phrase. Leading and trailing spaces, the double commas left behind when a word is deleted, and punctuation that is not part of a word all fall in the same item.

The third is stem duplication. Apple does not keep the singular and plural of the same stem apart, so writing both burns the longer one's characters. In languages with rich suffixes this item finds noticeably more than it does in English.

The fourth is writing a phrase whole. Words in the keyword field combine freely with one another, so writing the parts separately gets the combinations indexed too. Writing the phrase whole spends the space between the words and blocks the other combinations at the same time.

Apple's store is open in 175 countries and regions, and each has its own keyword field. Most of those fields are indexed independently. Leaving one country empty spends all 100 characters there.

That waste has no visible sign. Your app is live there, its page renders, it simply does not appear in search. The only thing that surfaces it is that your ranks in that country never formed - and because nobody looks, it goes unnoticed for months.

The audit reports country coverage on its own line: how many fields are filled, how many are empty, and which of the empty ones already receive downloads. That last part matters, because filling a field for a country you get no downloads from is not the priority.

The second country item is copying. Pasting one country's field into another spends one of the two: the same language being searched differently in different countries is the rule, not the exception. The audit flags fields that are identical.

A metadata score produces no work. "Three words from your title repeat in the keyword field, 27 characters come back" produces work. So the output is a list: what was found, where, and how many characters it returns.

The rows are sorted by recovery. The top row is the correction that opens the most characters, and it is usually app-name repetition. There are also unsortable rows - a subtitle that has become a word list, a country field left empty. They carry no character figure but their decision value is high.

The right way to use it: apply the top three rows, ship one release, wait a week, then compare ranks. Applying all of it at once makes it impossible to tell which correction did what.

What you put in the characters you opened is a separate decision. Opening characters is not the gain by itself; what you write there is. Choose from your pool, starting with the phrases in the 11-30 band.

For a number from an audit tool to deserve trust, you have to be able to arrive at it yourself. Every item here can be done by hand in half an hour.

Put the app name and the keyword field side by side, lower-case both by the country's own folding rules, and mark the shared words. Each shared word's length plus one comma is recoverable. Then count the spaces after commas, look for two forms of the same stem, and split any part that contains a space.

Add it up and the number you get matches the tool's. If it does not, either a step was skipped or the letter folding was applied wrongly - both are the most common mistakes.

What the tool does is repeat that arithmetic across 175 storefronts every day. It does not hide the method; it only takes on the scale - and a tool that automates something you could do yourself is one you can check.