App Store Algorithm Changes: What Changed and How to Adapt

Algorithm & Ranking Factors ASO

Jessica Abbadia 29 September 2026

Apple’s App Store algorithm is not static. It changes in observable ways that can move rankings within days, and Google Play moves on its own separate timeline. This page tracks the shifts that matter for ASO, what each one means in practice, and what to do about it.

One thing to settle first, because it reframes everything below. The biggest change of 2026 was not a single dated event. Both stores moved weight away from pre-install signals, meaning keyword match and raw download volume, and toward post-install ones: retention curves, session frequency, engagement depth, and semantic relevance. Keyword metadata still opens the door. Retention decides whether you stay in the room. Apple’s own published research points the same way, and adds a detail most coverage skipped: the biggest relevance gains landed on long-tail queries. Our App Store ranking factors guide covers the full signal set behind that shift.

If You Distribute Outside Google Play, Read This Section First

Android developer verification enforcement starts 30 September 2026 in Brazil, Indonesia, Singapore, and Thailand. From that date, certified devices running Android 7 and above in those four markets block normal installs of any app not registered to a verified developer.

Scope: certified devices running Android 7+. Unregistered apps remain installable via ADB or the advanced flow. Global expansion follows in 2027. This is an identity check, not a ranking factor.

Algorithm Changes Timeline

February 2026. Apple published its ranking signals. Apple researchers released Scaling Search Relevance: Augmenting App Store Ranking with LLM-Generated Judgments, the closest thing to primary-source documentation the App Store has produced. Much of the ASO coverage dates it to March and describes the result as “0.24% more downloads.” Both are slightly off, and the details matter.

Apple names two objectives: behavioral relevance, meaning results users tend to click or download, and textual relevance, meaning a result’s semantic fit to the query. The constraint they were solving is that expert-written textual relevance labels are scarce while behavioral labels are abundant, so they fine-tuned a model to generate millions of textual labels and fed them into the production ranker.

A worldwide A/B test on the App Store ranker produced a statistically significant +0.24% increase in conversion rate, not download volume. The line worth acting on is the next one: the largest gains landed in tail queries, where behavioral signals are too sparse to be reliable and the new textual labels carry the weight. If you compete on long-tail terms, semantic clarity in your metadata now does work that download velocity used to do.

June 2026. WWDC brought three App Store changes. At the keynote on 8 June, Apple introduced Personalized Collections, an on-device recommendation layer. App Notes gives a short editorial explanation of why an app surfaced. Creative Assets let developers place header images and video in the product page and in search results, managed in a new Asset Library and updatable without shipping an app release. We covered that layer in detail in iOS 27 ASO: what Apple’s App Store update changes. Personalized Collections rolled out in English in the US first, appearing across the Apps, Games and Search tabs.

Ongoing 2026. Engagement decay accelerated. Apps with strong install numbers but weak Day-7 and Day-30 retention now lose category position faster. Buying cheap installs has moved from ineffective to actively harmful.

Ongoing 2026. Ratings freshness weighting increased. A 4.5-star app with 100 recent reviews can outrank a 4.8-star app with 10 old ones.

Late July 2026. Mass keyword deindexing. The 26 July peak reached 94,092 U.S. apps losing ranked keywords, an outlier well above the 90-day baseline. Apple has not confirmed an update.

Apple Ads second search placement. Rolling out first in the UK and Japan.

July 2025, still underexploited. Custom Product Pages went organic. Only 31% of apps use CPPs at all.

The Change That Never Happened

Worth knowing, because it teaches you how to read volatility.

In June 2025, ranking anomalies across English-speaking markets led much of the ASO community to conclude that Apple had begun indexing screenshot caption text using OCR. The theory spread fast and was repeated widely. Apple later denied it, and AppTweak’s analysis agreed. Some sources still state it as fact today.

The lesson is not that the community is unreliable. It is that a correlated ranking shift is evidence something changed, not evidence of what changed. Verify the mechanism against the platform before you rebuild a strategy around it.

Moburst’s own screenshots never had to change when Apple denied the theory, which says something about how to build a metadata strategy in the first place. “We always put strong keywords into our screenshots, but not because we believed the algorithm was scanning them for rankings,” says Jessica Abbadia, VP of Organic at Moburst. “It’s been a Moburst best practice regardless of that theory, because your metadata, screenshots, and description need to tell one consistent story. If your keywords say one thing and your screenshots say another, you lose conversion. So we never had to unlearn this one. We were doing it for a different reason than the ranking theory, and that reason held up.”

Keyword Deindexing: What’s Changed and Why

Keyword stuffing in the subtitle or keyword field stopped producing a ranking lift years ago. By 2026, aggressive repetition can suppress an app rather than merely fail to help it. A keyword that drives taps without conversions is a liability. Our ASO guide covers metadata fundamentals if you need the baseline.

A large metadata overhaul in response to an unconfirmed shift can do more damage than the shift.

Recovery guidance from ASO monitoring platforms is deliberately unaggressive: monitor keyword coverage daily for a few weeks before changing anything. If freshness signals carry more weight during periods like this, pairing a meaningful app update with metadata and creative changes tends to work better than an isolated metadata edit.

That caution matches what Moburst does with clients in the room. “We’ve been through algorithm changes hundreds of times, it’s part of the job,” says Jessica Abbadia, VP of Organic at Moburst. “In the first few weeks after a change like July, you see a drop and you need to investigate. We check keyword volumes to see what lost traffic, we check typos, we look for patterns that tell us what the algorithm actually changed. Often that’s all we get: no explanation, just an event. So we remap the strategy from what we observe. We tell the client exactly that: an algorithm change happened, here’s the behavior we’re seeing, here’s what we’re changing and adding in response. Usually within four weeks we see performance recover, maybe not at the same rate, but we keep adjusting until the strategy fits the new pattern.”

Google Play behaves differently. Its natural-language processing reads semantic relationships rather than counting token frequency, so a repetitive phrase in a long description gets treated as low content quality and suppresses browse visibility, not just ranking for that term.

Custom Product Pages in Organic Search

Before July 2025, a Custom Product Page appeared only when someone arrived through a paid Apple Search Ads click or a shared link. Since keyword linking launched, any CPP with keywords assigned from your existing keyword field is eligible to surface organically for those terms, replacing your default product page for that search.

Apple allows up to 70 Custom Product Pages per app, doubled from 35 in October 2025. Each is independently indexable, which is enough to build a genuinely granular one-intent-to-one-page strategy instead of serving every search intent from one generic listing.

Adoption is the opportunity. Only 31% of apps use CPPs at all, so this remains one of the few organic levers where being early still counts for something.

To optimize a CPP for organic traffic, treat each one as a landing page for a specific search intent: distinct screenshots, promotional text and feature emphasis matched to that keyword. A CPP must be localized for a market to appear in that market’s search results.

The tradeoff is your default page. As more top keywords route to CPPs, the default listing increasingly serves branded search and direct navigation rather than the full spread of discovery traffic it used to catch.

AI-Powered Personalized Collections

Personalized Collections surfaces apps based on a user’s existing downloads and app activity, before they type anything. Processing happens on-device. App Notes explains why a specific app was suggested.

Apple has not published the selection logic and there is no ad product tied to placement. It runs on top of keyword search, top charts and editorial features rather than replacing any of them.

The structural difference matters. Search ranking responds to a query. Personalized Collections responds to a behavior pattern with no query at all. The lever is not a keyword field, it is the same clarity of positioning that helps AI chat assistants recommend an app correctly, and it is the discipline our ASO team applies across every discovery surface, since both reward an unambiguous, well-supported description of what the app does and who it is for.

Google Play Algorithm Shifts

Google Play’s mechanics are more documented than Apple’s, and more demanding.

Cross-platform teams should stop templating.

Google Play has also been surfacing more of the review corpus in search and listing context, letting a shopper see review text relevant to their search term rather than only an aggregate star rating.

Historical Changes for Reference

iOS 14’s App Tracking Transparency prompt cut the attribution data available for paid campaigns, which raised the relative value of organic ASO because paid measurement got noisier. The iOS 11 redesign brought a substantial search algorithm overhaul alongside the new visual layout. Google Play’s 2021 policy changes tightened permissions and data-safety disclosure in ways that affected how some apps could position themselves.

The throughline: keyword weighting shifts gradually and rarely gets documented by either platform, which is why your own data matters more than any single announced change.

How to Monitor Algorithm Changes

Track these four together. A change that hits one usually shows up in the others within days.

1. Keyword coverage. How many target keywords are ranked at all, versus not ranked. This is what catches a deindexing event.
2. Impression share. Moves before installs do.
3. Store listing conversion rate. Separates a visibility problem from a persuasion problem.
4. D7 and D30 retention. Since post-install signals now carry more weight, a retention decline is an early warning for a ranking decline that has not happened yet.

Our guide to increasing app store rankings covers the levers behind each of these.

Use daily rank tracking rather than weekly snapshots. Community signals surface a widespread shift within a day or two, often before any tool vendor publishes. Treat them as a detection layer, not an explanation layer.

FAQs

Jessica Abbadia

Jessica is Moburst's VP of Organic. She specializes in enhancing organic performance for apps and games all over the world, while actively developing innovative methods for increasing app visibility and conversion, as well as offering her vast knowledge for the benefit of the mobile community. She graduated from law school and now serves as an animal rights activist who also loves reading books while sipping a strong coffee and holding one - or more - of her three cats.

Sign up to our newsletter

Δ

Email(Required)