Apple Now Publishes App Store Search Data: Inside the New Ads API, With 65 Weeks of It Analyzed

TL;DR: With the new Apple Ads Platform API, Apple publishes real App Store search demand for the first time: a weekly report of the top 500 search terms per category, scored 1–100, for any country you ask. We've been ingesting it since it appeared and backfilled everything Apple retains, 487,334 term-week records across 65 weeks of US data. This is what's inside: a top-100 that is almost entirely brand names, a "1–100" score that never goes below 39, misspellings ("tik tok", "door dash") outranking the correct spellings, 8% of US searches not in English, and launch spikes you can date to the week. Gemini went from the rank-40s to #1 in Productivity in four weeks; Sora entered at rank 6 the week it launched.

For fifteen years, App Store search volume was guesswork. Apple showed advertisers a 1–5 "popularity" bar in Search Ads, tools scraped it, and everyone rescaled it into fake-precise 0–100 scores. That era ended twice in six months: in September 2025 the old metric broke (readings for most keywords dropped to the floor value and stayed there, blinding every tool built on it), and then Apple shipped its replacement, a real insights endpoint in the rearchitected Ads Platform API:

POST https://api.ads.apple.com/v1/insights/apps/search-term-popularity/query

The report it returns is the closest thing to ground truth that has ever existed for App Store search, and you don't need to run a single ad campaign to read it.

What the endpoint returns

Every week (published Mondays 07:00 UTC for the preceding Sunday–Saturday), for each country and each of 15 category buckets (GAMES, SOCIAL_NETWORKING, PRODUCTIVITY_UTILITIES, …), Apple lists the top 500 search terms by real search volume. Only terms with at least 500 searches that week qualify. Each row carries four scores:

That's 15 × 500 = 7,500 term-rows per country per week. Apple retains 65 weeks of weekly history (a monthly variant retains 15 months, which adds nothing). Auth is OAuth 2 client-credentials through an API user an Apple Ads account admin creates. The account must be in good standing, with billing details set up, but the data is store-wide, not scoped to your campaigns.

If you'd rather read it in a dashboard, App Store Connect gained a beta "Search Term Rank" report with the same data. But the API is the programmatic source, and it's the one you can archive. That matters, because Apple deletes everything older than 65 weeks.

The parts the documentation gets wrong

We implemented the ingest against the live API, and four things work differently than Apple's docs say. If you're building on this endpoint:

1. Your orgId comes from GET /v1/acls (under result.acls[].adAccount.orgId), and every data call must carry it in an X-Ap-Context: orgId=... header or you get a 403.
2. The documented sorting[].order property is rejected (REQUEST_UNRECOGNIZED_PROPERTY). Omit sorting entirely; the default is genre ASC, rankInGenre ASC.
3. Rows live under result.rows, and pagination has no total count: page (cap 5,000) until you get a short page.
4. The rate limit is tight, 5 requests per ~1-second window on the insights endpoint. A full country-week is 2 requests, so even backfilling 65 weeks × 10 countries is only ~1,300 calls.

What 65 weeks of the data show

We loaded Apple's full retention window for the US storefront, 487,334 rows spanning May 2025 through the week of August 2, 2026, and joined it to our keyword corpus. The findings below are from that dataset.

The top of the App Store is brands, one exception aside

The 30 most-searched terms in the US store (week of August 2, 2026, cross-category score):

Of the top 60 terms, 59 are brand names. The lone generic is "vpn" at #12. The next generics sit far down: "games" (77), "ai" (76), "music" (75), "bible" (73), "photo editor" (70), "solitaire" (70). People overwhelmingly open App Store search to fetch an app they already know by name; generic category discovery is real but a full tier below. Eight of the top 15 are social apps, and the category's #1, Instagram, is the single most-searched thing on the US App Store, ahead of TikTok, YouTube, and ChatGPT.

The 1–100 scale actually starts at 39

Across the full report for the latest week, the lowest searchPopularity1to100 is 39. The distribution is a narrow pyramid:

Only 54 terms in the entire country score 80 or above, while 3,968 — more than half the report — sit in the 50s. The scale compresses hard: remember the entry ticket is already 500 real searches, so a "50" is a genuinely searched term, and the space between 84 ("cash app") and 100 ("instagram") is enormous. Treat differences of 2–3 points as meaningful at the top and noise in the middle.

The classic 1–5 bar, it turns out, is just this score quantized: 2 maps to 39–40, 3 to 40–60, 4 to 60–80, 5 to 80–100. In the latest week, 76% of all qualifying terms score "3". That is exactly why the old five-step metric was useless for comparing keywords, and why its replacement matters.

Apple files each term under exactly one category

A surprise with product consequences: across every week we ingested, no search term ever appears in two category buckets. "minecraft" is a GAMES term, "capcut" a PHOTO_VIDEO term; Apple assigns each query one home category and ranks it only there. The buckets are coarser than App Store genres (15 buckets, with things like Productivity and Utilities merged), and depth varies hugely: the 500th GAMES term still scores 57 ("star wars", "asmr", "assassins creed" sit at ranks 496–500), while the 500th SPORTS term scores 39. Rank 200 in GAMES represents far more searches than rank 200 in SPORTS.

Misspellings are separate terms, sometimes bigger than the brand

Apple does no spelling normalization. Every variant is its own row with its own real volume:

For TikTok, DoorDash, and Cash App, the spaced spelling beats the official one. "tick tock" (68), "face book" (56), and "you tube" (60) all clear 500 weekly searches on their own. If your keyword field targets only the canonical spelling of the brands around you, Apple's own data says you're skipping measurable traffic.

8% of US App Store searches aren't in English

599 of the week's 7,500 US terms contain non-Latin characters, and they're not fringe: Arabic "تيك توك" (TikTok) scores 79, rank 8 in all of Entertainment; Arabic shorthand for Instagram ("انستا") also scores 79; Russian "впн" (vpn) 75; Arabic "ببجي" (PUBG) 71; Russian "яндекс" (Yandex) 71. The US storefront serves a very multilingual user base, plus users in countries without their own App Store. Localized metadata competes for these searches. Almost nobody targets them.

The report churns, and the churn is the signal

Week over week, an average of 542 terms (7.2%) are new to the report. Over 65 weeks we saw 19,961 distinct terms, of which only 4,093 appeared every single week (the evergreen core), while 7,932 appeared four weeks or fewer. Those short-lived entries are the interesting ones. Three patterns:

Seasonality you can set a calendar by. "fantasy football" enters the GAMES report in June at rank ~488, hits rank 6 in late August (draft season), fades through fall, and exits entirely by February. "turbotax" appears in December, peaks January through April, and vanishes in May. The 65-week window covers a full cycle of every seasonal keyword in the store.

Launches, dated to the week. "sora by openai" debuted at rank 9 in PHOTO_VIDEO the week of September 28, 2025, the week Sora 2 launched, and generic "sora" jumped from rank 392 to 3, decaying steadily since. "gemini" climbed from the rank-40s in mid-August 2025 to #1 in Productivity four weeks later, the month its image model went viral, and has held the top 10 since. ICEBlock debuted at rank 5 in Productivity the week of June 29, 2025, straight from the news cycle.

Weekly momentum. In the latest week alone, "madden" rose 356 places in GAMES and "packers" 273 in SPORTS (a new Madden and the NFL preseason), while "canvas parent" and "5-star students" climbed ~300 places each in EDUCATION: back-to-school, visible in search demand before it shows in any chart.

What it still can't tell you

The report is the head of the market, not the tail. Five hundred terms per category, 500-search minimum; everything below that line (which is most of the keyword universe) remains estimation territory. The scores are relative indices, not search counts. It's weekly, so it can't catch an intraday trend. And Apple's own dashboard twin of this report is labeled beta; the September 2025 shake-up is a reminder that these metrics can change under you. Archive what you pull.

How we use it

AppSigma's keyword tracker now serves Apple's own popularity number as the traffic score for any keyword in the current report: first-party truth, no estimation. For the rest of the corpus, the report became our calibration set. Our demand model (built from download attribution, advertiser counts, and autocomplete data) is refitted against Apple's anchors every day, so the estimated scores live on the same scale as the real ones. And because every reported term had ≥500 verified searches, the weekly file doubles as a keyword-discovery feed — each new term gets a ranking crawl within hours of Apple publishing it.

Method notes: all figures are from the US storefront, weeks of 2025-05-11 through 2026-08-02 (Apple's full 65-week retention window at the time of writing), 487,334 rows pulled from the Apple Ads Platform API v1 search-term-popularity endpoint. "Latest week" means the Sunday–Saturday week starting August 2, 2026. Cross-category term lists deduplicate each term to its best-ranked category row. Churn counts a term as new when it was absent from the same category's report the prior week.

Frequently asked questions

What is the Apple Ads Platform API search term popularity endpoint?

A first-party Apple API (POST /v1/insights/apps/search-term-popularity/query on api.ads.apple.com) that returns the top 500 App Store search terms per country and category each week, with 1–100 popularity scores. It's the first time Apple has published real App Store search demand programmatically and in bulk.

Do you need to run ad campaigns to get Apple's search popularity data?

No. You need an Apple Ads account in good standing (tax and payment details on file) and an API user created by the account admin, but the search term popularity report covers the whole store: it is not limited to keywords in your campaigns, and reading it costs nothing.

What replaced the old Apple Search Ads popularity score?

In September 2025 the old Campaign Management API popularity metric collapsed and most keywords went flat at the minimum value, breaking the third-party tools that scraped it. The sanctioned replacement is the Ads Platform API's search term popularity endpoint, which scores terms 1–100 instead of the old five-step bar; the 1–5 value is still returned, and it's simply the 1–100 score quantized into four bands.

What are the most searched terms on the App Store?

In the US, as of August 2026: instagram, x, snapchat, tik tok (as two words), facebook, reddit, whatsapp, telegram, youtube, and tiktok. The top of the store is almost entirely brand names; the highest-ranked generic term is "vpn" at #12.