A competitor’s hidden App Store keyword field is not public. Unless you have access to that app’s App Store Connect account, you cannot know exactly what keywords they entered.
The better question is not “what did they type?” It is “which searches does Apple currently rank them for, and what evidence explains that rank?”
That distinction matters. A tool that says it found a competitor’s keyword should be clear about the source: live rank evidence, visible metadata, autocomplete expansion, Apple Ads evidence, or a model. Those are not the same thing.
Live rankings are the strongest public signal
If an app ranks in the top 10 for a keyword in a country, that is strong evidence that Apple sees some relevance and performance for that query. Tracking this over time is more useful than guessing the private keyword field.
The useful record is specific: competitor app, keyword, country, rank, date, and movement. “Competitor ranks number 4 for invoice maker in the US” is evidence. “Competitor uses invoice maker in hidden keywords” is usually a guess unless you own that account.
Country changes the evidence
Competitor keywords are not global. An app may rank for a phrase in the United States and fail to rank in Germany. Another app may dominate a translated query because local competitors are weaker. If you only check one storefront, you may copy a strategy that does not apply to your market.
Metadata mining generates candidates
Titles, subtitles, descriptions, screenshot captions, IAP names, and review language can surface keyword ideas. But visible metadata is not proof of rank. It is a candidate source.
A competitor may visibly target “habit tracker” but rank poorly. Another may rank for “routine planner” because Apple understands the semantic relationship, even if that phrase is not visible.
The right workflow is to mine metadata, then check rank. Metadata gives you guesses. Live search tells you which guesses currently matter.
Autocomplete shows possible demand
Store suggestions can reveal how users phrase a need. They help expand a keyword list, especially for long-tail searches. But autocomplete does not prove that a competitor ranks or that the keyword converts.
Autocomplete is strongest when it finds phrases you would not have written yourself. It is weakest when teams treat every suggestion as a priority. A suggested phrase still needs relevance, demand, reachable competition, and product-page support.
Apple Ads search terms are powerful but account-bound
Search Match and broad match can reveal real search terms from your own Apple Ads campaigns. That is strong discovery data because it comes from actual ad exposure. But it requires the user’s campaign data and only covers what the campaign reached.
Paid search terms can explain your own opportunity, not a competitor’s private metadata. They are useful when a term gets impressions, taps, installs, or downstream revenue and also makes sense organically. They are not useful when a broad-match campaign drifts into irrelevant traffic.
Reviews and screenshots can explain intent
Reviews can reveal the words users use after they have tried the product. Screenshot captions can reveal how the competitor wants to position the product. Neither source proves rank, but both can explain why a competitor might convert for a keyword.
For example, if a competitor ranks for “budget planner” and the reviews repeatedly mention envelopes, bills, and shared budgets, those words may become useful candidate terms. The next step is still to check search results.
Be careful with estimated traffic
Some tools estimate downloads or revenue tied to competitor keywords. These estimates can be directionally useful, but they are not Apple-provided facts. They usually depend on rank, category, country, ratings, observed store behavior, and the tool’s own model.
Treat estimates as prioritization hints. Do not build a strategy on a decimal point.
What a useful workflow should show
A clean competitor keyword workflow should show ranked evidence first: the competitor, keyword, country, position, movement, and visible reasons. Metadata and autocomplete should help explain or expand the idea, not pretend to expose a private field.
The best output is a decision: track this keyword, test this country, adjust this screenshot promise, add this unique word to metadata, or ignore the term because the evidence is weak.
A founder-safe rule
Use competitor keywords to find market language, not to clone a competitor. If a competitor ranks for a term but your app does not satisfy that search, skip it. If your app satisfies the search better and the result page has visible weaknesses, that is the opportunity.