Can you do App Store keyword research without paying for a tool?

Yes, for the two numbers that matter. Apple's search-hints endpoint ranks the terms its own search box suggests, which is as close to free keyword volume as Apple offers, and the search endpoint returns the apps actually ranking for a term, from which difficulty can be computed. Neither needs an account.

Why is a low-difficulty keyword sometimes worthless?

Because it is a brand name. `notion` scores easy purely because Notion's own app has relatively few ratings, but nobody typing it will accept a different note app. Difficulty measures how strong the ranking apps are; it says nothing about whether the search is winnable by someone else.

How do you tell a brand from a category automatically?

Look at the sellers, not the ratings. If two or more of the top ten results come from a seller whose name contains the whole phrase, the term belongs to that company. If three or more unrelated sellers have named their app after the phrase, several independent parties think it describes a job, and it is a category.

How to do App Store keyword research for free

App Store keyword research comes down to two numbers and one judgement. The numbers are how many people search a term and how hard the apps ranking for it are to displace. The judgement is whether the term names a job anyone can compete for, or a specific company. Get the judgement wrong and the numbers are worse than useless.

Popularity: read Apple's own suggestions

Apple does not publish search volume. It does power the suggestions in its own search box, and those are ranked. A term that Apple surfaces early under a short prefix is a term people type, and that ranking is the closest thing to free volume data that exists.

Two things follow. Prefixes are the unit, so seed your research with the letters and category names people actually start typing rather than with finished phrases. And the suggestions are per storefront: the endpoint needs a storefront identifier, and a term that is popular across US, UK, Turkey, Germany, Japan, South Korea and China is a different term in each one.

Difficulty: read the apps that already rank

Search the term and look at the top ten results. Difficulty is a function of how strong those apps are, which in public data means their rating counts and their average ratings. Ten apps with a million ratings each is a wall; ten apps with a few hundred is an opening.

Weight the count far more heavily than the star average. A 4.8 average on 200 ratings tells you almost nothing, while 200,000 ratings at 4.2 tells you an app has an installed base you are not going to outrank on relevance alone.

The judgement: brand or category?

This is the step that separates useful research from a list that wastes a month. Somebody typing meditation timer will take any good one. Somebody typing notion will not take yours. Both can score as low difficulty, and only one is winnable.

The tempting test is ratings dominance: if one app in the top ten towers over the rest, it must be a brand. It does not work. Meditation timer is 96% one app by rating share and is a genuine category, while shopping calculator is 100% one app, that app holds about 2,267 ratings, and the term is wide open.

The signal is the seller. Two or more of the top ten coming from a seller whose own name contains the phrase means the term belongs to that company, which is what catches quest diagnostics, notion and every local YMCA app at once. Otherwise, three or more unrelated sellers who each named their app after the phrase means several independent parties believe it describes a job, and that is a category.

Do not simplify that to matching app names alone. A brand's satellites and its third-party clients all carry the brand in their names, which is exactly why name matching classifies notion as a category.

Then check whether the category is enterable at all

A winnable keyword inside an unenterable category is still a dead end. Once a term survives the tests above, the next question is the ratings wall on the category it sits in, which is a different guide.

Our own keyword difficulty checker runs the popularity and difficulty passes on any term with no account, and every scored term ends up on a public page with the date it was last checked.

Be polite with the endpoints

Apple's search, autocomplete and review feeds are free, undocumented, unauthenticated and rate-limited. A throttled feed answers 200 with nothing, which looks exactly like a term with no results, so pace your requests, back off when they thin out, and never treat an empty response as an answer.