When to run this checklist
Run it before an app install campaign goes live on any network, whether that is Google, Meta, TikTok or an app-specific network. Run it again whenever the app's measurement SDK is updated, the attribution partner changes, or the store listing changes, because each of those can break attribution or availability without any sign in the ad platform. A new app release can change event names, and a renamed event leaves the campaign optimising toward something that no longer fires.
The user acquisition manager runs the list, with the app developer or analytics owner for the measurement group: the test-device checks need someone who can install a build and read the debug output. Whoever owns the store listing should confirm the store and links group, and finance or growth should agree the cost target before anyone sets it.
Before you start
- The campaign brief, with the goal (installs or a named in-app action), countries, platforms and budget.
- The measurement plan: the MMPAn independent service that attributes app installs and in-app events across ad networks. or app analytics in use, the event list and the SKAdNetwork conversion value schema.
- A test device for each platform, reset or with the app uninstalled, and a way to see test installs in the measurement partner's debug view.
- The store listings for each platform, and the countries each is available in.
- The deep link map: which link opens which screen, and what happens when the app is not yet installed.
- The payback model, showing what an install or an in-app action is worth and over what period.
The checklist
The groups follow the order an app campaign depends on: measurement first, because the campaign learns from it; then the store and links, because the click has to land; then setup, creative, and the economics that set the target.
Measurement
Store and links
Campaign setup
Creative
Economics and sign-off
Where each check is explained
The widget keeps each line short enough to tick; the reasoning lives in the guides. Each check links to the page that explains it.
| Check | Explained in |
|---|---|
| Attribution SDK reports from a test device | App attribution: MMPs and SKAdNetwork |
| SKAdNetwork conversion value schema | App attribution: MMPs and SKAdNetwork |
| In-app events mapped | Deep links and app analytics events |
| Networks linked to the measurement partner | Google Ads for app installs |
| Store listing live in every country | App store optimisation |
| Deep links and deferred deep links | Deep links and app analytics events |
| Install or in-app action optimisation | App install campaigns |
| Platform and operating system targeting | App install campaigns |
| Each network's own settings | Meta Ads for app installs |
| Existing users excluded | App install campaigns |
| App shown in use | TikTok Ads for app installs |
| Asset sizes and orientations | App install campaigns |
| Cost target from the payback model | Retention and payback period |
| Approvals and launch-day checker | Paid campaign launch SOP |
If an item fails
The campaign waits. Measurement failures stop everything, because an app campaign that optimises toward an event learns only from what the attribution chain reports. If the test device install does not appear in the measurement partner, or appears without the in-app eventsAn action recorded inside an app, such as a sign-up, trial or purchase, reported through the measurement partner or SDK and used to optimise campaigns., the fault is usually in the SDK integration or the network link, and it needs the developer. Launching anyway produces a campaign that cannot be judged and, worse, bids on silence. An unset or undocumented SKAdNetwork schema means iOS results arrive as values nobody can decode; set it before launch, because changing it later breaks comparison with what came before.
A store listing missing in a targeted country is a hard stop for that country: remove it from targeting or wait. A broken Deep LinkA link that opens a specific screen inside a mobile app rather than its home screen. is softer, since installs still happen, but the user lands on the home screen instead of the promised product or offer, and the in-app action the campaign optimises toward becomes less likely. Deferred deep links fail more often than direct ones and need testing on a device without the app.
Setup failures are fixed per network, and the network-specific guides explain each platform's own requirements. A CPIAverage ad cost of each app install. target picked from a benchmark rather than the payback model is the failure most likely to survive review, because the campaign will run either way. It should not: if the model says an install pays back only after a long Payback PeriodMonths of customer margin needed to recover the cost of acquiring that customer., the target has to reflect it, or the campaign will buy installs that never earn their cost.
A language-learning app's iOS launch
A language-learning app called Parlotta launches install campaigns on two networks in the UK and Ireland. The
test-device pass shows Android installs arriving in the measurement partner with the trial-start event, but on
iOS the trial-start event is missing: the latest app build renamed it from trial_start to start_trial. The
developer updates the event mapping, the SKAdNetwork schema is redocumented to match, and the iOS campaign
switches its optimisation event. The payback model says a trial start is worth about £6 over the first 90 days, so
the team sets its target per trial at £4 rather than the £9 suggested in a vendor benchmark.
Common mistakes
- Testing on a device that already has the app. Installs are not attributed, deferred deep links cannot be tested, and the pass proves nothing.
- Optimising toward installs when the goal is paying users. Install volume rises while the users who matter stay flat.
- Changing the SKAdNetwork schema after launch. Earlier and later results can no longer be compared.
- One creative set for every network. Sizes, orientations and placements differ, and the network crops what it receives.
- Showing the logo instead of the app. People install what they can picture using.
- Targets borrowed from benchmarks. The campaign meets a cost target that the app's own economics never supported.
The app install campaigns guide explains the choices behind each group, and the checklists library holds the rest.
