The App Store and Google Play rank apps using algorithms neither Apple nor Google publishes. Apple's developer documentation gestures at "user ratings, relevance, and engagement"; Google Play Console mentions "install velocity, retention, and ratings". Neither is a ranking contract you can optimize against directly.
What we do know comes from observation — watching apps rise and fall across hundreds of thousands of keyword-country-category combinations, correlating rank movement with measurable inputs. This guide synthesises what a decade of that empirical work tells us.
Which algorithm?
"The App Store algorithm" isn't one thing. At minimum, there are four distinct ranking systems to think about:
- Search rank — your position when a user searches a specific keyword. Per-keyword, per-country, per-device.
- Top Charts — Top Free, Top Paid, Top Grossing. Per-country, per-category, per-device.
- Category rank — your position within your primary App Store category, independent of the Free/Paid/Grossing split.
- Editorial surfacing — Today tab features, "App of the Day", curated collections. Human-curated with algorithmic recommendations behind the scenes.
These systems share some signals (downloads, reviews, retention) but weight them very differently. Search rank cares about keyword relevance in ways chart rank doesn't. Top Grossing cares about revenue; Top Free ignores it. Optimizing for one doesn't automatically help the others.
The signals both stores have confirmed
From developer documentation, WWDC and Google I/O talks, Console UI, and App Store Connect UI, we can confirm the following are explicitly used:
- Download volume (both stores). The absolute and relative count of installs in a recent window.
- Download velocity (both). Rate of change, not absolute volume. A newcomer doing 20k/day trending up outranks a veteran doing 50k/day trending flat.
- Star rating and review count (both). With heavy weight toward recent reviews — a 4.8-star app with 100 recent 3-star ratings can rank below a 4.3-star app with 100 recent 5-star ratings.
- Keyword match (iOS title / subtitle / keywords field; Google Play title / short description / full description).
- Uninstall rate (Google Play explicit; iOS implicit through retention telemetry).
- Crash rate (Google Play explicit).
- Update cadence (both — actively-maintained apps rank better than stale ones).
The signals MWM has observed empirically
The signals above are the ones the platforms acknowledge. From watching rank movement in our dataset, we believe the following are also materially weighted:
- Post-install engagement — D1, D7, D30 retention and session length appear to influence category and Overall rank, not just search relevance. Apps with strong retention curves outrank apps with better metadata but weaker retention on contested keywords.
- Review velocity (new reviews per week) separately from total count. An app with 5k total reviews and 200 new this week will outrank a 50k-review app with 5 new this week, if other signals are close.
- Cross-category behaviour — users installing your app shortly after installing other apps in a cluster appears to signal topical relevance. This is how "customers also installed" rails form.
- Velocity of paid vs. organic — the algorithm appears to distinguish paid spikes from organic ones, rewarding sustained organic growth more than comparable paid-driven spikes.
We observe that category rank is substantially more durable than Overall rank. An app that breaks into the top 10 of a category typically stays there for weeks; an app that breaks into Overall top 10 via a velocity spike often falls within 5-7 days unless sustained velocity continues.
How fast ranks move
Rank velocity is asymmetric: it's easier to rise fast on the Free chart than to rise fast on Paid or Grossing, and it's always easier to fall than to rise.
Typical patterns:
- A Super Bowl-grade marketing push can move an app 200-500 positions in a single day on the Free chart, reaching the top 10 Overall in major markets.
- A viral social-media moment (TikTok trend, celebrity mention) produces similar velocity spikes and often decays within 3-7 days.
- A category editorial feature typically adds 15-40% to baseline downloads for the duration of the feature, with a 30-60% sustained lift for 2-3 weeks after.
- Sustained paid UA at 10-20x baseline will raise chart rank, but the lift is smaller per dollar than an organic trigger. The algorithm appears to apply a "paid discount" multiplier.
- A broken app update (heavy crashes, bad reviews) can tank Overall rank 100-200 positions within 48 hours.
The common thread: what the algorithm actually scores is change, not state. State-based optimization (have great metadata, collect reviews slowly) underperforms velocity-based optimization (coordinate creative pushes with review prompts with update drops).
Search vs. chart: very different games
Search ranking is keyword-deterministic. If your app has "meditation" in its title, it will rank for "meditation" — the only questions are how high and in which countries. Metadata, localization, and review count dominate.
Chart ranking is velocity-deterministic. Metadata barely matters; what matters is whether your download curve is trending up. You can have terrible ASO and still top the Free chart with a good marketing push; you can have perfect ASO and never crack the chart if you can't generate velocity.
The combined view: great ASO compounds when you generate velocity. Great velocity without great ASO is a flash in the pan.
Category rank vs. Overall rank
For most publishers, category rank is the commercially relevant number. Overall top 10 is reserved for viral mega-apps and massive spenders; it's also brutally volatile. Category top 10, by contrast, is both more durable and more accessible.
Two observations from our rank archive:
- Category rank has lower velocity thresholds. Breaking into top 50 of Finance or Utilities in most countries requires a fraction of the daily downloads required to crack Overall top 100.
- Category rank has higher visibility-per-rank. Browsing users on the Games or Finance chart look deeper than browsing users on the Overall chart, where attention cliffs sharply after position 25.
If you're optimizing with limited UA budget or organic-only, category rank is where your effort goes furthest.
What you actually control
Despite the opacity, a great deal of ranking outcome is directly controllable:
- Keyword rank: drive via title, subtitle, keywords field (iOS), and description (Google Play).
- Conversion rate: drive via creatives (icon, screenshots, preview video).
- Review velocity: drive via well-timed native rating prompts, 10-25 days post-install, gated on successful action.
- Retention: drive via onboarding quality, notification strategy, product loop strength.
- Velocity triggers: drive via coordinated marketing, PR, seasonal events, In-App Events, editorial submissions.
Everything else — the exact weighting Apple chose this quarter, the tuning Google rolled out last week — is noise you can't act on. Build an ASO and growth programme around the levers you control, measure weekly, iterate monthly.
Seasonality and the editorial channel
Both stores have seasonal patterns worth tracking:
- Q4 (October-December) is peak download season for consumer apps. Competition for keyword ranks intensifies as UA spend spikes.
- January-February typically sees new-year installs for health, productivity, and finance apps.
- Back-to-school (August-September) drives education apps.
- Summer drives games, travel, and entertainment apps.
Editorial featuring on iOS (Today tab, App of the Day, collection cards) is algorithmically seeded but humanly curated. Submission is via App Store Connect; the success rate is low (single-digit percent) but the upside is very high — a Today feature typically delivers 30-100k incremental downloads in a day.
What's changing
Apple's ATT and SKAdNetwork transition (ongoing since iOS 14.5) has pushed both stores toward more algorithmic personalization and away from deterministic attribution-based re-ranking. Practically, this means:
- Individual user install decisions carry less signal; aggregate behavioural patterns carry more.
- Apps with high retention and engagement gain more relative weight vs. apps with high install counts and weak retention.
- Localized performance matters more — a US-only success story translates less well to category rank in Germany than it did five years ago.
The playbook: build for durable product quality, not just download volume. Apps that retain and monetize earn sustained rank; apps that don't fall out of the charts the moment the marketing budget cools.
Where to go next
- For the tactical playbook: How to Rank Your iOS App.
- For the metadata and creative mechanics: The Complete ASO Guide.
- To see rank movement empirically: browse MWM's live country rankings.