# Keyword difficulty and search volume for apps: how to read them

> What app keyword volume, difficulty and Apple Ads or Google Ads popularity measure, where each one misleads, and how to combine them into decisions.

- URL: https://asomaster.app/blog/keyword-difficulty-and-volume/
- Published: 2026-10-27
- Updated: 2026-10-27
- Category: Metrics
- Publisher: ASOMaster (Cressu OÜ), https://asomaster.app/

## Key takeaways

- Rank is measured; estimated volume and difficulty are models; Apple Ads and Google Ads popularity are relative scores, not search counts.
- Apple Ads shows popularity as a 1–5 level; the finer 1–100 index from the API is still relative, and many long-tail phrases sit at the floor.
- Google Ads Keyword Planner measures Google web search, not Play Store search.
- Never average an estimate with real popularity: keep them in separate columns.
- Read rank first, then demand, then difficulty.

Every keyword tool for apps shows two or three numbers next to each phrase, and most people read them the wrong way: as if "volume 62" were a count of searches and "difficulty 40" were a promise. Neither store publishes search counts per phrase, so almost every number you see is either a model built from public data or a relative score from an ads platform. Both are useful. They answer different questions, and mixing them up is how teams end up rewriting a subtitle around a phrase nobody types.

This guide explains what each number measures, where it is reliable, where it misleads, and a simple way to turn a board full of numbers into a short list of decisions.

## What do the stores tell you for certain?

Only three things are measured rather than modelled: your rank in a given storefront, the results page itself (which apps rank and how many ratings they have), and the popularity Apple and Google report to advertisers. Apple Ads covers App Store search; Google Ads Keyword Planner covers Google web search, not Play search.

Start with these, because everything else is built on them.

- **Rank** is observable. Search a phrase in a given storefront and your app is at position 7, or it is not in the results you checked. Rank is the one number on your board that is a measurement rather than an estimate, which is why it should carry the most weight.
- **The results page itself** is observable: which apps rank, how many ratings each one has, what their name and subtitle say, how many installs a Play listing shows.
- **Apple Ads search popularity** is published by Apple to advertisers and reflects how often a keyword is searched on the App Store ([Apple Ads best practices](https://ads.apple.com/app-store/best-practices/recommendations)). Apple Ads shows search popularity as a 1–5 level in its interface; the finer score available through the API is a relative 1–100 index (ASOMaster shows it on its 0–100 scale), not a search count. Many long-tail phrases sit at the floor or have no score at all.
- **Google Ads Keyword Planner** gives average monthly search ranges for Google web search, by country and language ([Use Keyword Planner](https://support.google.com/google-ads/answer/7337243)). It does not cover Play Store search. It is a proxy for interest in a topic, not a measure of Play search.

That is the full list of real inputs. Store analytics in App Store Connect and Play Console tell you how many people found you through search overall, but not which phrase they typed. Anything else labelled "volume" is a model.

## What does estimated search volume measure?

Estimated volume is a relative 0–100 model of the demand behind a phrase in one market, built from the apps that match it in search results and how popular they are. It isn't a search count, it is only comparable within one market, and it lags new trends, but it exists for every phrase you check.

The usual way to build it without access to search logs, and the way ASOMaster's estimate works in broad terms, is to look at the apps that match the phrase in that market's results and use their popularity as evidence. If the top results for "brown noise" in the US are apps with hundreds of thousands of ratings that mention brown noise in their name or subtitle, people must be finding and installing them, so the phrase carries demand. If the only matches are a handful of apps with a few dozen ratings each, demand is thin.

A few properties follow from that design, and they matter when you read the number:

- **It is relative, on a 0 to 100 scale.** A 60 is more than a 30. It is not twice as many searches, and it is not a count.
- **It is per market.** "white noise" can be 66 in the US and 12 in Japan, where people search ホワイトノイズ instead. Compare a phrase across markets only with that in mind, and read our [guide to localizing keywords](https://asomaster.app/blog/localizing-app-keywords/) before you conclude a market is small.
- **It follows relevance.** Apps that rank for a phrase without matching it (because the store's search engine matched on their category or on install behaviour) count for less. That keeps one giant generalist app from inflating every phrase it appears under.
- **It lags new trends.** A phrase that started trending last week has few matching apps with many ratings yet, so the model underestimates it. Popularity from an ads platform reacts faster.
- **It can be low for good phrases in small stores.** Ratings counts differ a lot by country. A good model calibrates for that, but in small storefronts the evidence is thin and the estimate is noisier.

The estimate exists for every phrase you check, in every market, at no cost. That is its main strength: you can triage two hundred candidates across eighteen markets before spending anything on ads data.

## What does keyword difficulty measure?

Keyword difficulty is a 0–100 measure of how hard it is to reach the top ten for a phrase, judged from how established the apps already there are, mostly through rating counts adjusted for each country. It describes the competition, not your chance of ranking, and it says nothing about demand.

The common approach is to look at the apps already holding the top positions and how established they are, mostly through rating counts, adjusted for how many ratings apps typically collect in that country. A top ten made of apps with millions of ratings is hard to break into. A top ten where position 6 has 140 ratings is open.

What difficulty does not tell you:

- **It is not your chance of ranking.** Two apps can face the same difficulty and get very different results, because relevance, conversion and install velocity matter to the stores' ranking. Difficulty describes the competition, not you.
- **It says nothing about demand.** Some of the easiest phrases are easy because nobody searches them. Always read difficulty together with volume or popularity.
- **It needs calibration per market.** Ratings counts in Germany or Japan are not comparable to the US. ASOMaster marks a difficulty it could not yet calibrate for a market with a small `≈` sign, so you know to trust it less.

If you already rank in the top ten, difficulty matters less: you are past the wall it describes.

## Real popularity: what Apple and Google give you

Real popularity is the closest thing to search demand from the stores' side. It also has quirks you need to know before trusting it more than the estimate.

### Apple Ads popularity

- **It is relative and compressed.** Apple Ads shows search popularity as a 1–5 level in its interface; the finer score available through the API is a relative 1–100 index (ASOMaster shows it on its 0–100 scale), not a search count.
- **Most phrases sit at the floor.** Long-tail phrases, including many that convert well, all share the lowest value. Popularity separates head terms from the rest. It does not rank the long tail.
- **It is per storefront** and needs an Apple Ads account with access to the API. You do not have to run campaigns to read it, but you do need the account.

### Google Ads Keyword Planner

- **It measures Google web search, not Play search.** Someone searching "budget planner app" on Google is a reasonable signal that people want budget planners. It is not proof that the same phrase is typed into the Play Store at the same rate.
- **It is given as ranges** for most accounts, and it is grouped by close variants.
- **It is still useful for Play**, especially for comparing two phrasings of the same idea in a market where the Play estimate is thin.

### Imported popularity

If you already have popularity numbers from another tool, importing them is fine. Keep them labelled as imported, because each source uses its own scale.

## Can you combine estimated volume and real popularity?

No. Keep them in separate columns. They use different scales and different evidence, and they are wrong in different ways, so an average of the two is wrong on both. Read rank first, then let demand decide, using real popularity where you have it and the estimate everywhere else.

The single most important habit is never to put an estimate and a real popularity score in the same column. They are on different scales, built from different evidence, and wrong in different ways. A phrase with estimated volume 45 and Apple popularity at the floor is not "about 25". It is a phrase where matching apps are popular but few people search for this exact wording, which tells you to look for the wording they do use.

On the ASOMaster board, each cell shows rank with its movement, then difficulty (a peak glyph), estimated volume (a wave glyph) and real popularity (a target glyph). The estimate and the real number never share a glyph, and a cell with no popularity uses the estimate and marks it with a corner notch. The [metrics documentation](https://asomaster.app/docs/metrics/) explains each mark, and [real popularity](https://asomaster.app/docs/popularity/) covers connecting Apple Ads, Google Ads or a file.

![Keyword detail for one market showing the top ten apps, your position, rank history and the local search phrase](https://asomaster.app/shots/02-rankings-keyword-detail-light-1600.webp)
## Reading a board: a worked example

Here is a slice of a US board for Tidepool, a fictional sleep sounds app, from a snapshot taken months before the listing round in the [keyword research guide](https://asomaster.app/blog/app-store-keyword-research/) (which is why the ranks there are higher). Rank is what the store showed that day. Volume and difficulty are estimates on a 0 to 100 scale. Popularity is from Apple Ads.

| Phrase | Rank | Est. volume | Difficulty | Apple popularity |
|---|---|---|---|---|
| sleep sounds | 9 | 71 | 62 | high |
| white noise | 23 | 66 | 37 | high |
| brown noise | 41 | 41 | 39 | mid |
| rain sounds for sleep | 6 | 22 | 35 | floor |
| fan noise | not ranked | 31 | 24 | low |
| sleep sounds for adults | 14 | 8 | 19 | floor |

How to read each row:

1. **sleep sounds.** Rank 9 means you are already in the top ten. Difficulty barely matters now; the question is whether you can climb into the top three, which usually takes conversion and ratings, not metadata. Keep it in the name or subtitle and watch the trend.
2. **white noise.** High demand on both signals, rank 23 and a moderate difficulty of 37. This is the best opportunity on the slice: Tidepool is close, the demand is real and the top ten is beatable. Put both words in high-weight fields and check again after the next release.
3. **brown noise.** Moderate demand, difficulty 39, rank 41. The estimate and popularity agree there is demand and the top ten is beatable, so it is the second push phrase. Move it into the subtitle or keyword field.
4. **rain sounds for sleep.** Rank 6 but popularity at the floor and a modest estimate. You win a phrase few people type. Keep it if it costs nothing; do not spend subtitle space on it.
5. **fan noise.** Not ranked, low difficulty, some demand. A cheap test in the keyword field.
6. **sleep sounds for adults.** Low on every demand signal. Rank 14 is irrelevant if nobody searches it. Drop it and free the characters.

ASOMaster's verdict colours follow the same order of reasoning: rank comes first (a phrase you already rank for is never painted as hopeless), a phrase with too little demand is never green, and only then does difficulty decide between "opportunity" and "hard to win".

## Five mistakes that waste a release

1. **Chasing the highest volume.** The biggest phrases in a category are usually owned by apps with years of ratings. A rank of 40 on a huge phrase sends you almost no traffic, while a rank of 4 on a moderate one sends real installs.
2. **Treating difficulty as a forecast.** Low difficulty means the competition is weak, not that you will rank. Relevance still decides whether the store shows you at all.
3. **Comparing estimates across markets without localizing.** English phrases in non-English stores look tiny because locals search in their own language. Measure the local phrase.
4. **Reading one day's rank.** Ranks move daily. Look at the rank history over at least a week before you call a change a win or a loss, and mark the day you published so you can read the before and after.
5. **Averaging popularity and estimate.** Each signal is useful on its own scale. An average of the two is wrong on both.

## A routine that uses all three numbers

A workable weekly routine for one app:

1. Check ranks across your markets on a schedule (daily or weekly), so the rank column is current.
2. Fetch popularity after a check if you have an Apple Ads or Google Ads source connected, so both demand signals are fresh.
3. Sort by opportunity: phrases with real demand, a beatable top ten and a rank outside the top three.
4. For each candidate, open the keyword detail and look at the actual top ten. Are those apps relevant to the phrase? Could your app be the best result?
5. Pick at most three phrases to move into higher-weight fields in the next release. Note the date.
6. After the release, compare rank history before and after the publish marker, per market.

The same routine scales to many apps through an agent. Our guide to [AI agents and MCP for ASO](https://asomaster.app/blog/ai-agents-mcp-aso/) shows how to run steps 1 to 3 from Claude Code or Codex.

## Summary

- Rank is measured. Everything else is a model or a relative score.
- Estimated volume shows demand from matching apps; it exists everywhere and is free, but lags trends and is relative.
- Difficulty describes the top ten, not your chances, and needs per-market calibration.
- Apple Ads popularity separates head terms from the rest; most long-tail phrases share its floor. Google Ads covers web search, a proxy for Play.
- Keep estimates and real popularity in separate columns, read rank first, and let demand decide before difficulty does.

If you want to see these numbers side by side for your own app, the [Rankings board](https://asomaster.app/#rankings) shows all four in every cell. Plans are on the [pricing page](https://asomaster.app/pricing/).