The keyword pool

The pool's identity is the phrase folded for its storefront. Turkish and Azerbaijani lowercase a capital “I” to “ı”; applying that rule everywhere turns “AI” into “aı” and quietly destroys one of the store's most valuable stems.

So every stage that folds text - n-gram extraction, metadata audit, cusp suggestion - works knowing the country. Otherwise a phrase folded one way here and another way there becomes two separate keywords.

Every candidate's popularity is measured and the ones below the floor are permanently dropped, so the same dead term is not queried again every week. A candidate that could not be measured stays blank rather than being given a popularity of zero.

Keywords collected for unpublished apps count against no plan limit: an app that is not in the store has no rank to measure and therefore no daily cost. Thinking should be free.

Nearly every keyword tool feeds from the same three sources: search suggestions, category charts and app titles. That pool is open to everyone, so everyone's list looks alike - and that is exactly where competition is densest.

rankcusp grows the pool from five channels. The first four are the places everyone looks: your own metadata, rival metadata, search suggestions and category charts. The fifth is rival reviews - and that is the place nobody looks.

All five need to be used. Most teams settle for the first two and end up with a pool trapped inside their own marketing language. A narrow pool does not produce a good list; a good list comes from a wide pool and a hard cut.

While growing the pool, do not filter. Leaving a phrase out because it "looks too hard" is deciding without measuring - and difficulty is not something intuition estimates well.

Keyword research is treated as a finding problem: find the right words, write them down, done. But hundreds of phrases can be found for any app, and most of them are either not searched, out of reach, or do not describe what the app actually does.

Without elimination what happens is this: a fifty-phrase tracking list is built, nothing is touched, and three months later the list is forgotten. A six-phrase list gets looked at every week and actually worked on.

The order of elimination matters. First the rows whose popularity could not be measured are set aside - no decision can be made about them, because blank does not mean zero, it means unknown. Then the phrases describing something the app does not do: ranking for work you do not do means poor ratings, and a rating drop weakens your strength across every phrase.

What is left is split into bands: the first 10 are defended, 11-30 are spent, above 30 is investment. Three groups needing different work, and keeping them in one list is what makes the list unreadable.

Phrases in the pool run from one word to four, and the length decides how the phrase behaves.

One-word phrases are the most searched and the hardest. Competing there means competing with the category's largest apps; a new app getting in with metadata is not realistic. Two-word phrases are the balanced zone: searched enough, and far less contested. Most of the cusp band is this length, and metadata moves work best here.

Three-word phrases are searched less but their relevance is very high. An app ranking for one is doing exactly what the searcher asked for; gathered in clusters they produce a real total. Four and above is usually a fragment of a sentence and rarely searched.

A practical note: because words in the keyword field combine freely, writing the parts of a two-word phrase separately gets more phrases out of the same field than writing it whole.

Once the pool is wide, three measurements run, and the order matters because each one has a price.

Popularity first: a relative value derived from search suggestions. It is a cheap measurement and it eliminates most of the pool immediately. Where no suggestion returns, the field stays blank - not zero, because zero is a reading and people decide on readings.

Rank second: each remaining phrase is searched in that country and the first 200 are read. This one is expensive, because the requests per second to Apple are capped. So measurement runs per phrase rather than per row: every app tracking the same phrase is read out of one search. Adding a rival costs nothing.

Difficulty third: computed from the same search. How many apps carry the phrase in their title, what the rating counts at the top are, how many words the phrase has. It requires no extra request, so it comes free.

After those three every row in the pool carries four facts: the phrase, its popularity, your current rank and its difficulty. The list has become a table, and a table can be decided on.