Reference
Choosing keywords you can rank for
Three numbers decide whether a keyword is worth a slot. Here is what each one is.
Every keyword decision comes down to three numbers: how many people search it, how hard the apps above you are to displace, and what is left of the first once you account for the second. Everything else is detail.
Popularity — how much is searched
The demand behind a term. Apple publishes a search-popularity figure through its Search Ads API, which is the only first-party number either store gives out; Google publishes nothing comparable, because Keyword Planner measures web search and not the Play store.
Where a store does not publish it, popularity has to be estimated — and the honest estimate comes from the store's own autocomplete. A term the store suggests after two typed characters is in more demand than one it only offers once you have typed the whole phrase. It is an inference, and any tool showing you a popularity number for Google Play is showing you an inference. The useful question to ask of any ASO tool is whether it tells you which of its numbers are measured and which are modelled.
Difficulty — how hard the top ten are
Popularity on its own is a trap. "photo editor" is searched constantly and held by apps with millions of ratings; you will not be on that page, and being indexed for it changes nothing.
Difficulty is a property of the apps currently ranking, so it can only be computed by looking at them: how many ratings they carry, how well their own metadata matches the term, how entrenched the leaders are. A term where the top ten are ten strangers with four hundred ratings each is a different proposition from one where the top three are Calm, Headspace and BetterSleep, even if both are searched the same amount.
This is also why difficulty is not a fixed property of a keyword. It changes when the apps above you change.
Opportunity — what is left for you
Popularity discounted by difficulty: the share of the demand you could realistically win. It is the number to sort by, because it is the only one that answers the question you are actually asking — is this term worth one of my characters.
A term with popularity 70 and difficulty 80 can be worth less than one with popularity 45 and difficulty 20. The first is a bigger room you will not get into; the second is a smaller room with an open door.
What a popularity number is actually worth
Apple's figure runs 1 to 100, which invites you to read 50 as average. It is not. The terms Apple publishes sit in a narrow band near the top of its own scale, because it only charts the ones that are searched enough to chart.
Across the 7,500 US terms Apple charted in August 2026, the lowest was 46 and the median was 58. The 90th percentile was 68. The 99th was 80. Only 76 terms in the entire US chart reached 80 or above.
| Apple popularity | Where that sits |
|---|---|
| 46–53 | The bottom quarter of what Apple charts at all. |
| 54–62 | The middle half. Most real keywords live here. |
| 63–68 | Top quarter. |
| 69–79 | Top ten per cent. A head term in its category. |
| 80+ | Top one per cent. Roughly seventy terms in the country. |
Two things follow. A term at 58 is not mediocre, it is typical; and the difference between 55 and 65 is far larger than ten points of anything linear. Below 46 Apple publishes nothing at all, so every number you see under that is somebody's estimate — including ours.
How often any of this changes
Less than the dashboards imply. Comparing Apple's July and August 2026 charts, of the 6,714 terms present in both, a little over half moved at all, the median move was one point, and nine in ten moved by two points or less. Popularity is close to stable month to month, and a one-point change is noise rather than a trend.
The movement is at the edge instead. Around 786 terms — roughly a tenth of the chart — dropped out between those two months, and about the same number appeared. Terms do not usually slide down the scale; they fall off it.
Difficulty is the number that moves, because it is a property of the apps above you rather than of the term. A competitor shipping a new title, or three of them buying installs, changes it without the demand changing at all. Checking popularity monthly is enough. Checking difficulty matters most right before you commit characters to something.
How rare a winnable keyword is
Worth knowing before you start, because it sets what success looks like. Of the 4,615 keywords scored in our own database — everything this tool has been asked to look at — the median difficulty is 60, and only three per cent sit at 40 or below. Roughly one in seven is somebody's brand name.
Terms that combine real demand with a beatable top ten — popularity at 45 or better and difficulty at 45 or worse — are about two per cent of the total.
That is the honest shape of the work: you are looking through a hundred keywords to find two. It is also why every tool will happily hand you five hundred candidates. The candidates are not the scarce part.
Reading a keyword in practice
| Shape | What it usually means |
|---|---|
| High popularity, high difficulty | A head term. Worth tracking to see whether you ever move, rarely worth a title slot early on. |
| Moderate popularity, low difficulty | The keywords worth having. Not glamorous, and where most real ranking comes from. |
| Low popularity, low difficulty | Long tail. Individually small, collectively significant, and cheap to cover in body copy on Play. |
| Any popularity, and it is a brand | Somebody else's name. Hard to rank, the intent is theirs, and Apple rejects competitor names outright. |
| You already rank 1–10 | Defend it. Terms you hold are worth more than terms you might get, and they are the ones a listing rewrite can quietly lose. |
Where to find candidates
- Your own listing. Your title, subtitle and category are what the store already associates with you. Expanding those through autocomplete finds the field you are nearest to ranking in.
- Store autocomplete. Real demand, revealed by the store, free. Typing your seed term plus each letter of the alphabet turns one term into most of a category.
- Competitor listings. Not their brand — the generic terms they rank for that you have not written down. An app you compete with is a list of keywords somebody has already validated.
- What you already rank for. The cheapest keywords to improve are ones you are on page two of. The store already thinks you are relevant; you are asking it to think so slightly harder.
How many to keep
Fewer than the tools will offer you. The App Store gives you 160 indexed characters in total, which is perhaps 25 to 35 useful words; Play gives you more room in the body but every term you name is a term the copy has to carry naturally. A list of 500 keywords is a research artifact, not a plan.
Keep the ones you would be willing to write into the listing, plus the ones you already rank for and want to watch. That is usually somewhere between twenty and fifty.
A keyword list is per storefront
There is no such thing as a keyword's popularity in general. Each storefront is its own index with its own demand and its own top ten, and Apple publishes its popularity figure per storefront for that reason. A term that is competitive in the United States can be almost unattended in Spain or Brazil, and the app holding the top spot is frequently not the same app.
The practical version: research the storefronts you actually get downloads from, not the ones you would like to. If four fifths of your installs are from one country, that country's numbers are the only ones that should decide your title. The rest are a second exercise, worth doing when the first is finished.
Mistakes that cost characters
On the App Store you are given 30 characters of name, 30 of subtitle and 100 in the keyword field — 160 in total that get indexed, and no more. The description is not searched at all. Most of the waste comes from the same handful of habits:
- Repeating a word across fields. Apple indexes the union of them, so a word already in your name earns nothing by appearing in the keyword field as well. That is one of your 160 characters spent on nothing.
- Writing phrases where words would do. Apple combines your indexed words into phrases itself. "sleep" and "sounds" already covers "sleep sounds"; writing the phrase out pays for the space twice.
- Spaces after the commas. The keyword field is comma-separated, and every space is a character you have bought and cannot search on.
- "app", "free", "best". They describe every app in the store, so they distinguish nothing and cost the same as a word that would.
- Competitor names. Apple rejects them, and the intent behind them belongs to someone else anyway. Roughly one in seven of the keywords in our database is a brand, so they take up a real share of any list you generate — and none of that share is yours.
Google Play works the other way and is covered separately in Google Play metadata: the long description is indexed there, so the constraint is not characters but writing naturally enough that the copy still reads.
Then measure
A keyword list is only worth having if you know where you stand on it. Rank is a reading taken on a day: run the search, find your app, write down the position and the date. Do that regularly and a listing change becomes a comparison instead of a hope — and you find out which of your metadata decisions actually moved anything.
The fields those keywords have to go into are covered in App Store metadata and Google Play metadata. The wider picture is in App Store Optimization.
Keywords with the most room left
21 Questions - Card Games
Bubble Level for iPhone
Apple Health
StressWatch: AI Stress Monitor
Roost Bedtime Routines
Photo Cleaner ®
Every one of them is on the keyword index.