Keyword strategy (the cusp band)

What the cusp band is, and why it is cheap

The keywords you rank 11-30 for are the cheapest growth your app has. This guide is about finding that band, ordering it, and spending it one phrase at a time.

Ranking for a keyword needs two separate things, and their costs differ enormously. The first is relevance: Apple associating your app with the phrase. The second is rank: where it places you among the apps it associated.

Building relevance from nothing for a phrase you never appear on is expensive work. It means rewriting metadata, probably revisiting positioning, and waiting weeks for the index to settle. And the outcome is not guaranteed - Apple may never associate you with that phrase at all.

For a phrase where you already sit at 24, relevance is long since established. Apple has said "this app is about this phrase", put you on the list, and simply not placed you high. The only thing missing is a few places of rank, and rank is won by moving the phrase within your metadata.

The 11-30 band is exactly the band of that situation. Every phrase sitting there has passed Apple's approval and failed to reach the first page. The lower bound is 11 because the top 10 is already the first page; the upper bound is 30 because below it the gap usually stops closing with a single metadata move.

What really sets the band's value is what the first page means. Users do not scroll; they open one of the first few results. An app at 12 is effectively invisible for that phrase - but the invisibility is two places of distance, not irrelevance.

Ordering the band

There are usually more phrases in the band than you expect. When a list of thirty or forty appears the first reflex is to work all of them - which ends with none of them worked.

The first cut is popularity. Phrases whose popularity could not be measured leave the list - set aside, not deleted. Blank does not mean zero, it means unknown, and priority cannot be built on unknowns. Order what remains by popularity, largest first.

The second cut is difficulty. Look at the strength threshold for the phrases at the top. If the median rating count of the top 10 is several times yours, move that row to a separate list; you will return as the product grows. This cut usually takes a third of the list, and taking it is good.

The third cut is distance. Phrases at 11-15 are far cheaper than those at 26-30. Between two phrases of equal popularity, take the nearer one - winning a few places is easier than winning fifteen.

The fourth cut is room. Think about where the phrase will go in your metadata. If it does not fit the app name, there is no space in the subtitle and the keyword field is full, that phrase cannot be spent this week. Making room is its own job.

Work with the top three of what is left. Touching more than three in one week makes it impossible to measure what any of them did.

• Blank popularity leaves the list but is not deleted; it may be measurable later.
• Above the strength threshold: product work, not metadata work.
• Between equal phrases, take the one lower in the band.
• A phrase with no room in metadata cannot be spent this week.
• Three phrases a week at most; a fourth breaks the measurement.

What happens after the move

The metadata change has shipped. Now comes the hard part: waiting and measuring.

Apple's index does not update immediately. A metadata change usually takes a few days to reach search results, longer in some countries. A second change inside that window mixes two moves together and neither can be measured.

There is work to do while waiting: follow the control group. Watch how the phrases you did not touch moved. If they all fell two places, your phrase falling two places is not the move failing, it is the market. Without a control group the result cannot be interpreted at all.

Compare a week later. Three outcomes are possible. If you won, note the move and apply the same kind to the next phrase on the list - what you learned is that this field works for your app. If nothing changed, wait another week; the index settles late in some countries. If you lost, undo it, wait a week, and see whether the old rank returns; if it does the move was wrong, if it does not the cause was something else.

The most valuable output of this loop is not the ranks won, it is the record accumulating. Three months in you have a sentence like "adding a phrase to the subtitle was tried five times and won three" - and that sentence is in no general guide, because it belongs to your app.

When the band runs dry, and what comes after

Run the band regularly and at some point it empties: nothing left at 11-30 waiting to be spent. That is a good sign, and it raises a new question.

There are two versions of running dry. In the first, the phrases in the band reached the first page - the intended outcome. In the second, all of them were set aside on the strength threshold; the band is full but none of it can be touched with metadata. Those are very different situations and should not be confused.

In the first case the job is widening the pool. New phrases have to be found so new ones enter the band. The most productive source is rival reviews: users' own words are in no keyword tool's pool and their competition is low. The second source is new countries - the same phrases may sit inside the band in another storefront.

In the second case the job is not metadata, it is the product. If the strength threshold comes from rating counts, more ratings have to be gathered, and that is solved inside the app. ASO stops here and should say that it has; carrying on chasing keywords is time spent for nothing.

There is a third situation: the band never filling at all. If the app is new, or appears in the first 200 for nothing, the band comes out empty. Then relevance has to be built first - metadata written to describe what the product actually does. The band starts filling once relevance exists.

• Band emptied: widen the pool - rival reviews and new countries.
• Band full but all above the threshold: the work moved to the product.
• Band never fills: relevance is missing; the metadata does not describe the product.
• Keep the band's weekly photograph; its movement says more than individual phrases.