How these pages work
Every page follows the same order. It first checks that the number describes what the team thinks it describes, because app metrics are unusually easy to misread: a rate counted only among the people who reached a screen, a figure that blends two operating systems with different permission models, an install count that one network, the store and the measurement partner each define differently. A diagnosis table then sets out the symptom as the team sees it, the check that confirms or rules out each cause, the cause and the fix. Each cause gets its own short section with the fix, a way to verify that it worked and the habit that stops it returning, followed by a routine that catches the whole family of causes before the next release or campaign.
The pages deliberately do not re-teach the mechanics. How the notification permission works and why it is a single chance on iOS is explained in the push notifications and in-app messaging guide; how the first session should be built and when each permission belongs is the onboarding and permission requests guide; the store listing is the app store optimisation guide; paid installs are the app install campaigns guide; and which action counts as activation is settled by the tutorial on finding the app's activation event. Each troubleshooting page links to the relevant section at the point a cause is confirmed.
Which page to start from
Pick the row that matches what you can see, not the cause you suspect. The diagnosis pages beneath this introduction are listed automatically, so a new one appears here once it is published; until then, the guide named in the row holds the relevant checks.
| What you are seeing | Start with | Background |
|---|---|---|
| Few new users allow notifications when asked | Push notification opt-in rate low | App onboarding and permission requests |
| The share of users with notifications switched on keeps falling after campaigns | Users turning off app notifications | Push notifications and in-app messaging |
| Installs arrive, but few new users reach the moment the app is useful | App installs not turning into active users | App install campaigns |
| The store page converts fewer of its visitors than it used to | App store conversion rate dropped | Custom store listings for campaigns |
| Fewer people find and install the app without an ad than they used to | Organic app installs falling | App store optimisation |
| Each install costs more month after month | App cost per install rising | App attribution: MMPs and SKAdNetwork |
Permission, product, store or measurement: decide which question you are asking
Most wasted effort in app troubleshooting comes from fixing the wrong layer. A permission problem lives in the moment and the framing of a request: the operating system owns the dialog, so the app can change only when it appears and what it says beforehand, and a rewritten notification does nothing for people who were never asked at a sensible moment. A product problem lives in the first session, where an account form, a tour or a run of prompts stands between the install and the first useful thing, and it shows up as installs that never activate however good the campaign. A store problem lives on the listing, which the app marketing hub treats as the landing page of every install campaign: the screenshots, the rating and the promise either match what brought the visitor or they do not.
Measurement is the fourth layer and the one to check first. The mobile measurement hub explains why app figures disagree by design: networks attribute installs under their own rules, iOS results arrive through privacy-preserving postbacks that are delayed and aggregated, and a rate means little until its denominator is written down. Read iOS and Android separately, read changes by install CohortA group of customers who share a starting event and period, tracked together over time. rather than across the whole user base, and confirm that the event behind a falling number still fires before treating the fall as real. The push opt-in rate entry shows how the same allowers produce two different rates depending on whether they are divided by the people asked or by every install, and the CPIAverage ad cost of each app install. is a price of a visitor, as the install campaigns guide puts it, not of a user.
When to review instead
A troubleshooting page answers one question about one symptom. When several symptoms arrive together, or before a launch that depends on all of them, a structured pass is the better tool: the app store listing checklist checks the listing before each update, the app install campaign launch checklist checks a campaign before it spends, and the tutorial on testing app deep links before a campaign finds links that open the browser or the wrong screen, a fault that the tutorial's own test matrix diagnoses better than a separate page could. Install totals that never agree between the store, the measurement partner and analytics are a reconciliation rather than a fault, handled by the tutorial on reconciling app install numbers, and a rating that is stale or falling belongs to the tutorial on rating prompts and review replies. Apple's App Store Review Guidelines are the rules several of these pages rely on, as the messaging guides state them; read the current version before acting on any of them. The mobile and app marketing pillar places all of it in the wider growth system of store, first session, messaging and measurement.
