Using Apple Search Ads to Find Organic Keywords

Every other keyword number in ASO is an estimate. Search terms are a record of what people actually typed and whether it led to an install, which makes even a small ad budget the cheapest keyword research available.

Last updated 20 August 2026 · by the ASOWeb team

Why search terms beat every keyword tool

A keyword tool tells you an estimated search volume derived from a model. Apple's search terms report tells you that on 14 August, someone searched qr code wedding photos, saw your ad, tapped it, and installed your app.

That difference matters in three ways:

• Demand is observed, not inferred. If a term produced impressions, people are searching it.
• Intent is proven. A term that converts to installs at 40 percent is a term where your app is what people were looking for.
• Discovery is free. Apple's broad match will show you terms you never thought of, and those are frequently the best organic targets.

Getting the data out of Apple

You need an Apple Search Ads account with at least one campaign that has run. The search terms report is available in the Search Ads UI and through the Campaign Management API.

One detail worth knowing if you build this yourself: Apple runs two APIs. Search popularity lives on the Ads Platform API, while campaigns and their reports live on the Campaign Management API at a different host. Asking the wrong one for campaigns returns a 404 that reads like "you have no campaigns" rather than "wrong service", which is a very easy afternoon to lose.

The report also distinguishes two kinds of row. A targeted term is a keyword you bid on. An auto term is one Apple's discovery matched you to. The auto rows are where the value is: those are terms with proven demand that you did not know to target.

Joining paid terms to organic rank

The report on its own tells you what converts. It becomes actionable when you put it beside where you rank organically for the same term. Match on an exact normalised string, “habit tracker” and “habit tracker app” are different queries with different result sets, and treating them as one would let you claim a ranking you do not have.

One subtlety: a keyword is tracked per storefront, but Apple reports search terms merged across every storefront a campaign targets. So one term maps to several rank rows and you have to choose between them. Take the best rank: it answers "do I already own this query anywhere", which is the question that decides whether to keep paying.

The four findings and what to do with each

1. Converts in ads, not tracked organically. The strongest keyword suggestion available: proven demand, proven intent, and you are not even watching it. Add it to your tracked keywords and work it into your metadata.
2. You rank in the top three and still pay for it. Question the spend. You may be buying clicks you would have received anyway. The nuance is defensive: if a competitor bids on the term, ceding the ad slot above your organic result can cost more than the ad does.
3. You pay for it and rank nowhere. A clear signal about your listing. Apple thinks you are relevant enough to serve an ad, but not enough to rank you, so the term is missing from your metadata, or your relevance signals are too weak for that shelf.
4. High impressions, no installs. A relevance or creative problem rather than a keyword one. People saw you for that search and were not interested. Either the term is wrong for you, or your first screenshot is.

Doing this on a small budget

You do not need a real advertising programme to get useful data.

• Run a discovery campaign with a low daily cap for two to four weeks. Broad match on a handful of core terms, with search match on, is enough to surface dozens of real queries.
• Treat it as research spend, not acquisition spend. Judge it on the keyword list it produces, not on its return on ad spend.
• Pause it once you have the terms. The report stays available and the keyword list does not expire.

A few hundred pounds spent this way typically produces a better keyword list than any subscription tool, because it is measurement rather than modelling.

Frequently asked questions