How to Choose App Store Keywords That Actually Convert

The usual way teams pick app store keywords is to sort by search volume, take the top of the list, and put those terms in the metadata. It feels rigorous because there's a spreadsheet. Then a month passes, rankings barely move on the big terms, and the conclusion is that ASO doesn't work for this category.

The short version

Volume is the easiest number to get and the least useful one to lead with. Every keyword sits on three axes: how many people search it, how strong the apps already ranking for it are, and whether the person searching actually wants what you built.

In our experience, a modest-volume term with low difficulty and exact intent beats a high-volume term with weak intent almost every time. The high-volume term looks better in a report and performs worse in the store.

Filter on intent before you look at anything else

Here's the thing about intent: it's the only one of the three you can't buy your way out of. If a query describes a different category than yours, ranking for it sends you traffic that never installs. Worse than no traffic, actually — it drags your conversion rate, and conversion feeds back into ranking.

So the first pass through any candidate list is a delete pass. Read each term and ask: if someone typed this and landed on our page, would they install? Anything where the honest answer is "maybe, if they read far enough" goes in the bin.

Only after that pass do we look at volume and difficulty. Sorting by volume first is how weak-intent terms end up dominating a list.

Where the candidates should come from

• **Store autocomplete.** Both stores suggest queries based on real searches. It's free, current, and already ordered by popularity. We start here every time.
• **Competitor metadata.** Read the titles and subtitles of whoever currently outranks you. Note which terms keep appearing across several of them.
• **Your own reviews and support inbox.** Users describe what they wanted in their own words, and that vocabulary is frequently not the vocabulary on your listing.
• **Category browse labels.** Useful for spotting how people navigate when they aren't searching a specific phrase.

Notice that none of those is a paid tool. The paid tools are convenient, not magic.

Traps we see repeatedly

**Brand terms you don't own.** Ranking for a competitor's name can generate installs, and it will also generate complaints. It's also not a position you can defend — a policy change or one complaint can remove it overnight. We don't build plans on it.

**Terms your first session can't satisfy.** If the listing promises something the app doesn't deliver in the first two minutes, conversion drops and the reviews arrive quickly. The metadata isn't the problem there; the product page is telling the truth and the app isn't.

**Assuming one locale behaves like another.** A term that works in one market may be unused, or mean something different, in another. Research per locale. This is the single most common gap we find in otherwise well-run listings.

Where to place the terms once you've picked them

On iOS the title and subtitle carry more ranking weight than the keyword field, so the strongest terms go there. The keyword field is for the terms that didn't fit — not the leftovers you couldn't be bothered to prioritise, but genuinely the second tier.

On Google Play there's no keyword field to hide in. Distribute terms through the short and full description so they read naturally to a human, because that human is the one who decides whether to install. Repetition that reads badly to users doesn't earn you extra reach.

Proving it worked without fooling yourself

Record rank for every target keyword before you touch anything. Then check the same set on a fixed schedule — weekly is enough for most teams.

Two cautions. First, change one thing at a time when you can. If you rewrite the title and swap the screenshots in the same week, you won't know which one moved the number. Second, don't judge on 24 hours. Store rankings move for reasons unrelated to your metadata, and reading noise as signal is how teams end up reverting good changes.

What to do next

• Take your current keyword list and mark each term high, medium or low on intent. Delete the low ones regardless of their volume.
• Pull store autocomplete for your top three seed terms and add the relevant suggestions to your candidate pool.
• Record today's rank for every shortlisted keyword, so you have something to compare against after the next update.