App Analytics
Mobile app acquisition measurement
Measure app acquisition across store interactions, downloads, first opens and useful actions without mixing incompatible report definitions.
Mobile app acquisition measurement asks whether the routes bringing people to an app also bring eligible users who complete a useful task. Define that task first. Report store interactions, downloads or acquisitions, first opens, completed actions and later use separately: they occur at different points and may come from different systems.
Keep the counts distinct
| Observation | What it shows | What it cannot show alone |
|---|---|---|
| Store listing interaction | A visit or click under that store report’s rules | A completed download or useful app use |
| Download or acquisition | A completed store event under the selected metric’s rules | That the app was opened |
| First open | A launch recorded by an instrumented app | That the person completed a useful task |
| First useful action | An eligible user reached a defined completed state | That they will use the result later |
| Later useful action | A cohort member completed a relevant task in a later window | Why they returned |
Apple’s App Store Connect reports downloads separately from app usage, using distinct measures and definitions. Record the platform, report, metric, market, dates and denominator beside each count; these measures cannot be treated as counts of the same people.
Five acquisition counts, and what each one cannot tell you
- Store listing interactionA visit or click under that store report's rules: it does not show a completed download or useful app use
- Download or acquisitionA completed store event under the selected metric's rules: it does not show the app was opened
- First openA launch recorded by an instrumented app: it does not show the person completed a useful task
- First useful actionAn eligible user reached a defined completed state: it does not show they will use the result later
- Later useful actionA cohort member completed a relevant task in a later window: it does not show why they returned
Define a useful outcome
Choose a completed action that reflects what the app promises. For a hypothetical appointment app, that might be an eligible booking confirmed successfully; a tap on “Book” followed by an error would not qualify. State who can complete the task, which result counts, the time allowed and how repeat actions are handled.
Check that the recorded event marks that result. A report label alone does not establish that the action succeeded. Keep sign-up and task-start events as diagnostic steps, while judging acquisition against the completed outcome.
Choose a source definition
Decide whether the question concerns a person’s initial acquisition source, a store-reported download source or credit assigned by an ad system. Keep each system’s source definition and attribution rules visible.
Store categories have limits too. Read each platform’s category definitions before interpreting source labels; keep mixed and unknown categories labelled as such.
Do not force store, ad and app analytics totals to match. Investigate a specific difference using each system’s event, population and attribution rules.
Read the store funnel on its own terms
In App Store Connect, unique impressions and unique product page views count unique devices, while total downloads counts downloads and can be split into first-time downloads and redownloads. iCloud restores and automatic downloads to other linked devices are excluded from total downloads. Pre-orders count towards the conversion rate instead of the later download.
Apple attributes sales, usage and subscription data to the source recorded when someone taps to download or redownload the app. A manual redownload resets that source, so later activity can be credited to a different source. Territory and device filters help separate performance across markets and contexts.
Reading the App Store funnel: what to pin down before you interpret it
- Confirm that unique impressions and unique product page views count unique devices
- Confirm that total downloads is split into first-time downloads and redownloads
- Check that iCloud restores and automatic downloads to linked devices sit outside total downloads
- Note that pre-orders count towards the conversion rate instead of the later download
- Check whether a manual redownload has reset the recorded source, so later activity credits a different source
- Apply territory and device filters to separate performance across markets and contexts
Compare comparable acquisition cohorts
Group eligible new users by a stated starting event and acquisition period. Compare the share completing the same later useful action after a fully observed interval. Keep platform, market, app version, identity and source rules visible. A recent cohort has had less time to act.
Choose a later action that fits the product. Managing a booking may matter more than merely reopening a booking app. Treat differences between source cohorts as associations: their audiences, promises and circumstances can differ.
Native reports may cover narrower populations. App Store Connect usage data comes from people who agreed to share it, and Apple’s retention denominator excludes installers who never opened the app. Check the population in a chart before describing it as all acquired users.
Interpret post-install metrics carefully
Apple App Store Connect collects app usage data only from people who agree to share diagnostics and usage information. Its opt-in rate history can help show how changes in participation affect the data available, rather than implying that the report represents every user.
Apple defines a session as app use lasting at least two seconds; returning to the app after it has been in the background counts as another session. Active Devices counts devices with at least one session in the selected period, while Active Last 30 Days counts devices with a session in the previous 30 days.
Apple’s retention table groups installers by installation day and shows the share opening on later day offsets. A late first open can add an installer to the denominator for earlier retention rates, so figures for a past installation date may change as more people open the app. The table’s percentages include only installers who have ever opened the app.
Apple's post-install measures, defined
- SessionApp use lasting at least two seconds; returning to the app after it has been in the background counts as another session
- Active DevicesDevices with at least one session in the selected period
- Active Last 30 DaysDevices with a session in the previous 30 days
- Usage data populationOnly people who agreed to share diagnostics and usage information
- Retention denominatorInstallers who have ever opened the app
- Retention groupingInstallers grouped by installation day, showing the share opening on later day offsets
Make a decision from the review
For each comparable source cohort, show available entry counts, first opens, eligible users, completed first actions, later useful actions and measurable cost. Leave unavailable or unmatched measures labelled rather than estimating a conversion rate from incompatible totals. For an Australian service with location limits, assess eligibility in the areas it actually serves; an English-language download does not establish it.
Use each gap to choose an investigation. More listing clicks without more acquisitions points to the store step and its definitions. More first opens without more completed actions calls for a check of eligibility, the acquisition promise, instrumentation and the task path.
These patterns identify where to look; they do not diagnose a cause on their own. Record the proposed change and reassess once the relevant outcome window has closed.
From a gap in the counts to the next investigation
- More listing clicks, no more acquisitionsExamine the store step and its report definitions
- More first opens, no more completed actionsCheck eligibility, the acquisition promise, instrumentation and the task path
- Unavailable or unmatched measuresLeave them labelled rather than estimating a conversion rate from incompatible totals
- A suspected causeRecord the proposed change and reassess once the relevant outcome window has closed
In this guide
- Installs versus activated usersSee how store entry, first opens and activation differ, and when the data supports a first-open or download-to-activation rate.
- Comparing acquisition cohorts by later app useBuild comparable acquisition cohorts, choose a useful later action, allow windows to close and read platform retention reports within their limits.
- Separating paid attribution from organic app discoveryRead mixed App Store and Google Play source labels accurately, keep ad credit separate and avoid counting unknown acquisitions as organic.
