Quick answer: At WWDC26 Apple shipped the biggest App Store Connect Analytics update since the product launched — more than 100 new metrics and support for up to 7 simultaneous filters — and the part that matters most for ASO went live on July 1, 2026: performance data per search keyword. For every search term that brings people to your app, you can now see impressions, tap-through rate, conversion rate, 1/7/30-day retention, and in-app purchase revenue — in the App Store Connect app, the web dashboard, and the Analytics Reports API. For a decade the industry inferred all of this from rank positions and third-party estimates. Now Apple hands over the actual numbers, with three real catches: the data covers only your own app, parts of it run 5–7 days behind, and unlocking the API requires a key role most developers don't have yet. This guide covers all of it.
What Apple now shows for every search keyword
App Store Connect always told you that Search drove some share of your impressions and downloads. It never told you which searches. The July 2026 update breaks that aggregate apart into five keyword-level views:
- Impressions by keyword — how many times your app appeared in search results for a given term.
- Tap-through rate by keyword — the share of those impressions that became product page taps, segmentable by device, geography, and iOS version.
- Conversion rate by keyword — installs divided by taps, per term.
- Retention cohorts by acquisition keyword — day-1, day-7, and day-30 retention for users grouped by the search term that brought them in.
- Revenue by acquisition keyword — in-app purchase revenue attributed back to the search term the buyer originally arrived through.
The first three describe the funnel above the install. The last two are the genuinely new capability: connecting a search term to what users did after installing. A keyword that produces installs who churn by day 7 and a keyword that produces subscribers used to look identical in every dashboard on the market. They no longer do.
Segmentation is the quiet second half of the update. Tap-through rate by keyword crossed with device and geography answers questions like "does this term convert on iPad?" or "is my German listing underperforming for this exact query?" — and with up to 7 filters applied at once, you can stack those cuts instead of choosing one.
Where the data lives
Three surfaces carry the same numbers:
- The App Store Connect app — fine for checking a headline number from your phone.
- The web dashboard — App Analytics in App Store Connect, for browsing, filtering, and one-off questions.
- The Analytics Reports API — for anything systematic.
If you expect to look at this more than once a month, use the API. Keyword-level analysis is a joins-and-time-series problem — you will want these rows next to your tracked ranks, your metadata change dates, and last month's numbers — and that means owning the raw data, not screenshotting a dashboard.
The access problem: why your existing API key gets 403s
This is the step that stops most people, so it comes before the API mechanics. The Analytics Reports API enforces role requirements that do not match the key most teams already have:
- Creating the analytics report request requires an App Store Connect API key with the Admin role.
- Downloading the reports requires Sales & Reports access.
The key you generated for CI publishing or metadata sync is almost certainly an App Manager key. App Manager is enough to ship builds and edit listings; it is not enough here, and the failure mode is a bare 403 that does not explain itself. If your first request bounces, suspect the role before you suspect your code.
The fix takes two minutes: App Store Connect → Users and Access → Integrations → generate a key with the role you need. A sane split is one Admin-role key that exists only to create the report request — a one-time act per app — and a Sales & Reports key for the recurring downloads. Don't promote your everyday publishing key to Admin just for this; scoped keys are cheap and over-scoped keys are how incidents happen. If you have never issued an App Store Connect key at all, our App Store Connect setup guide walks through the issuer ID, key ID, and .p8 file once, and the same key then serves publishing and reporting.
Pulling keyword reports through the API
The keyword data rides the standard Analytics Reports API flow — the same delivery pipeline Apple uses for its other analytics report categories[1]. It is a warehouse delivery, not a query endpoint:
- Create one report request per app with access type ONGOING. You do this once; Apple keeps producing data from then on.
- List the reports available under that request. The keyword data arrives under two new categories, reported as MetricCategoryKeywordConversion (impressions, taps, tap-through rate, conversion) and MetricCategoryKeywordRetention (the 1/7/30-day cohorts).
- Enumerate the dated instances of each report — one per delivery date.
- Download the segments: gzipped TSV files. Decompress, parse, and load them into whatever you analyze with.
Two operational notes. First delivery is not instant: expect 24–48 hours between creating the ONGOING request and the first instances appearing, so create the request today even if the pipeline is next sprint's work. Second, this is append-style delivery — your job is to keep pulling new instances, dedupe, and maintain your own history. Apple gives you files, not a database.
The limitations, stated plainly
Four limits define what this data can and cannot answer. None of them appeared in the keynote.
1. Own-app data only. You see keyword performance for your app and nobody else's. There is no view of a competitor's impressions, conversion, retention, or revenue for any term — the competitive half of keyword research is exactly as blind as it was before. The market-wide Search Term Rank report below is the partial exception, and it contains no per-app performance numbers.
2. Retention and revenue run 5–7 days behind. The acquisition-funnel metrics arrive on the normal analytics cadence; the retention cohorts and revenue attribution lag about a week. Don't read a fresh cohort's day-7 number early — it isn't finished being counted.
3. Attribution is last-click. A user who saw your app across five different searches and installed after the sixth is credited entirely to the sixth. In practice this inflates brand terms and re-engagement queries: people who already decided to install search your name to go get it, and the brand keyword collects credit for demand that generic terms created. Read brand-term numbers with that discount applied — more on this in play 7 below.
4. It is not real-time. Nothing in these reports supports same-day reactions. This is a weekly-rhythm dataset: measure, change one thing, wait for the cohort to mature, read again.
The other new report: Monthly Search Term Rank (market-wide)
Separate from keyword analytics — and easy to miss — is the Monthly Search Term Rank report, under Reports → Insights (Beta) in App Store Connect. Apple introduced it in October 2025 with data reaching back to July 2024[2], and it is the market-wide complement to the own-app analytics above. Each row is a search term with its rank within a genre and a Search Popularity in Genre score on a 1–100 scale, plus country-level popularity on both a 1–100 and a coarser 1–5 scale. The dimensions are country × month × genre, and the report refreshes monthly.
This is Apple publishing relative search demand across the whole store: which queries are big in your genre and country, and how that shifts month to month — whether your app ranks for them or not. Two cautions. The monthly refresh means it describes seasons and trends, not this week. And its popularity scales do not correlate with the old Apple Ads 5-point search popularity scale — do not build a mapping between them, and treat any historical spreadsheet keyed to the Apple Ads scale as a separate series. For how these relative scales relate to the absolute "searches per month" estimates third-party tools publish, see our guide to App Store keyword search volume.
The mental model: keyword analytics tells you how your app performs on terms it already surfaces for; Search Term Rank tells you what the market is searching. Prospecting lives in the second, performance tuning in the first.
What Apple still doesn't give you: difficulty
Apple did not ship a keyword difficulty metric, and the distinction matters. Popularity tells you how many people search a term; it says nothing about whether you can rank for it. A popularity-80 term whose top ten is held by apps with a million ratings each is worth nothing to a new entrant, while a popularity-30 term with a weak top ten can be a real install source. Difficulty is a property of the competitive field — who occupies the visible results and how strong their review mass, rating quality, and title match are — and that still has to be computed from analysis of the live results, which Apple does not publish and has no incentive to.
So the honest post-2026 stack is: Apple for your own funnel and for market demand, competitive analysis for whether a term is winnable. AppDrift's keyword rank tracking shows a 0–100 difficulty score and a popularity score beside every tracked keyword, computed from the live top results — which is precisely the piece the official reports leave out.
7 plays to run with the new data
Each of these is an analysis you can run this week with nothing beyond the reports above. They are suggestions for reading your own numbers, not promises about what those numbers will say.
1. Kill zero-impression keywords. Cross impressions-by-keyword against the terms you actively track and hold metadata characters for. A term Apple shows zero impressions for is not merely ranking poorly — it is generating no surface at all. Free the slot and the characters, and refill them from your keyword research pipeline.
2. High tap-through, low conversion: find the product page mismatch. A term where many searchers tap but few install is a promise your product page isn't keeping — the search result (icon, title, subtitle) sells one thing, the screenshots and description sell another. These terms are your product-page test queue: the demand is proven, the page is the leak.
3. High-conversion terms: double down in metadata. Terms that convert above your average have earned more prominent placement — promote them from the keyword field toward the subtitle or title, where the ranking weight is. Our guide on how to choose App Store keywords covers the field-weighting mechanics.
4. Read retention-by-keyword as intent quality. Two terms with equal installs are not equal keywords if one cohort is gone by day 7. Retention per acquisition keyword is the closest thing ASO has ever had to a measured intent-quality signal: terms whose cohorts stick describe what your app actually is; terms whose cohorts churn describe what your listing merely implied.
5. Follow the revenue attribution. IAP revenue per acquisition keyword shows where the money enters. Terms that produce buyers deserve your title characters, your metadata iterations, and — if you run Apple Search Ads — your bids. Terms that produce installs and no revenue are growth-chart decoration. Remember the 5–7 day lag before judging a recent change.
6. Reconcile Apple's install-credited terms with your tracked ranks. The single highest-yield query: which search terms does Apple credit with installs that you are not tracking at all? Every such term is a keyword you are winning by accident — invisible to your rank tracker, absent from your metadata plan, and one competitor update away from disappearing. Fold them into tracking and defend them on purpose.
7. Discount the brand-term story. Under last-click attribution your brand keyword will look like a monster: highest conversion, best retention, most revenue. Some of that is real; some is generic-term demand taking its last step through a brand search. Before reallocating effort toward brand, check whether brand-term installs actually rose or merely absorbed credit that generic terms generated.
Or let the pipeline run itself
If you would rather not build the TSV ingestion yourself: AppDrift now pulls these official keyword reports automatically, using the same App Store Connect key you already connect for publishing — it creates the report request, fetches each new instance, and joins the results with your tracked ranks and difficulty scores. The most useful output is play 6 done continuously: search terms Apple credits with installs that you aren't tracking yet, surfaced as weekly actions in ASO Autopilot rather than rows in a gzip file.
Frequently Asked Questions
What keyword data does App Store Connect show now?
Since July 1, 2026: impressions, tap-through rate (segmentable by device, geography, and iOS version), conversion rate (installs ÷ taps), 1/7/30-day retention cohorts by acquisition keyword, and in-app purchase revenue attributed to the acquisition keyword — in the App Store Connect app, the web dashboard, and the Analytics Reports API.
Why does my App Store Connect API key get a 403 on analytics reports?
Almost always the role. Creating the analytics report request requires an Admin-role key; downloads require Sales & Reports access. Most existing keys are App Manager keys created for publishing, which are not sufficient. Generate a new key under Users and Access → Integrations with the right role.
Can I see competitors' keyword performance in App Store Connect?
No. Keyword analytics cover your own app only. The one market-wide dataset is the Monthly Search Term Rank report — search terms, genre ranks, and popularity scores by country and month — and it contains no performance numbers for any specific app.
Is the new keyword data real-time?
No. Acquisition metrics arrive on the normal analytics cadence, retention and revenue run about 5–7 days behind, and the Search Term Rank report refreshes monthly. Treat it as a weekly-rhythm dataset.
Did Apple add a keyword difficulty score?
No. Apple publishes demand (popularity, search term ranks) and your own performance per keyword, but nothing about how hard a term is to rank for. Difficulty still has to be computed from competitive analysis of the live results.
