Choose the return action
Choose a useful app action before building the audience. A return visit may be necessary, but decide what the visit should enable: completing unfinished setup, using a relevant feature, or making an eligible purchase. Define the action and its validation record so the campaign can be judged against product value rather than app opens alone.
Write the reason someone should return now. Confirm that the destination contains the promised feature or offer and that it applies to the intended person. Avoid presenting a reminder as useful simply because an account has become inactive. Product context should determine whether the message is relevant.
For the proposed test, name one primary return action and any supporting signals. Keep a transaction, a session, and a subscription status change separate. Decide whether the useful outcome is counted per account or per action, and how repeated activity will be handled. Attach that definition to both the audience brief and the reporting plan.
Treat the following plan as work to validate for the app's current setup. It does not describe a historical reactivation case or imply that a particular source will accept the proposed audience.
Define eligibility and suppression
Translate the audience into rules that someone can inspect and reproduce. Define the qualifying behavior, relevant time window, market, platform, and required permissions. Specify when membership is evaluated and refreshed. Use the audience logic framework to identify the inclusion and exclusion decisions before exporting a list.
Suppression rules should remove people for whom the action or message no longer applies. That might include users who already completed the action, hold the entitlement being promoted, opted out of the relevant use, or have an unresolved account restriction. Have the appropriate owner confirm the rules for the selected source and data use.
Audience planning table
These rows are proposed segment patterns to adapt, not existing audiences or performance claims.
| Segment | Useful return action | Eligibility | Suppressions | Destination | Measurement |
|---|---|---|---|---|---|
| Unfinished setup | Complete the missing setup step | Started setup and still needs the step | Completed setup or ineligible account | Relevant setup screen | Validated completion per eligible account |
| Lapsed feature users | Use an available feature again | Prior feature use and agreed inactivity rule | Recent users or unavailable entitlement | Relevant feature screen | Qualified feature action within the agreed observation |
| Eligible prior purchasers | Make a relevant repeat purchase | Prior purchase and current product eligibility | Recent completion, unsuitable offer, or account restriction | Valid product or offer screen | Validated transaction and agreed revenue basis |
Record which system supplies each rule and how conflicts are resolved. Test whether a user completing the action leaves the audience on the intended refresh cycle. If membership cannot be updated reliably, narrow the proposed use or postpone delivery until the suppression dependency is resolved.
Match the audience to a reachable source
Separate the app's eligible population from the audience a source can actually reach. Confirm supported identifiers, permissions, list processing, campaign eligibility, and exclusions in the working account. Keep eligibility, matched audience availability, and delivered exposure as separate quantities in the review record.
Google's App campaign for engagement setup guide specifies requirements including a minimum of 50,000 app installs, an app-user audience list, deep links, and supported conversion tracking. That install-base requirement is not a forecast of how many eligible users your audience will contain or how many people the campaign can reach.
Use the source eligibility review to document which available sources support the proposed test. Ask the buyer to verify account controls and the measurement owner to verify the reporting path. Do not commit to a source because it supports new-user acquisition; check the return journey and intended audience explicitly.
Record overlap with other paid campaigns and owned messages. Decide which overlaps the test will prevent and which it will leave in place. Keep the message policy consistent with the audience rules, including any limit on repeated exposure. Document controls that the chosen setup cannot enforce.
Test the deep link and fallback
Test from the actual ad route to the useful action. Cover an installed app, a logged-out account, an outdated app version, unavailable content, and an absent app. Record what should happen in each state and who owns a failure. A successful open of the app's home screen is insufficient if the promise requires a specific feature.
Google's engagement deep-link guide recommends Android App Links and iOS Universal Links. It states that redirecting attribution-partner links are unsupported as final URLs for App campaigns for engagement. Check that channel's final-URL rules rather than assuming a link used elsewhere is suitable.
Preserve the intended destination through login where the app supports it. Verify that an expired or ineligible offer leads to an understandable alternative. Treat fallback and any post-install continuation as separate journeys to validate; do not assume that opening a store page preserves the original context after installation.
Audience and link QA checklist
- Confirm inclusion windows, eligibility, permissions, and suppression refresh.
- Test removal after the useful return action is completed.
- Verify source eligibility and final-URL requirements in the account.
- Open the intended screen through the actual ad route on each platform.
- Check login, entitlement, unavailable content, and outdated app states.
- Verify absent-app fallback and any supported post-install continuation.
- Confirm campaign parameters and the return event reach the intended reports.
Keep these checks in the return campaign build record, with the app version and test evidence. Recheck the affected routes when a product or link configuration change alters the journey.
Separate re-engagement from re-attribution
Use the terminology of the selected measurement provider. AppsFlyer's retargeting attribution guide distinguishes re-engagement with an installed app from re-attribution following a qualifying reinstall. It also describes overlapping attribution of in-app events to retargeting and original acquisition sources during the re-engagement window. Do not add those reporting views as independent business outcomes.
Adjust's reattribution documentation uses reattribution for an inactive user's qualifying return after marketing engagement; a reinstall is not always required. This is why the dictionary must name the provider and report. A label alone does not establish the same population across systems.
Record the inactivity rule, click or view lookback, and period for crediting later actions separately. Confirm how the provider treats redownloads and repeated returns. Keep the original acquisition record and the return campaign's attribution identifiable. When a report changes credit, reconcile it against the same validated product action before describing additional business value.
Plan the holdout and decision
A credited return answers which campaign received attribution under configured rules. To assess incremental value, plan a comparison that estimates what would have happened without the tested advertising. Google's Conversion Lift overview describes controlled treatment and control groups and notes that the tool is unavailable to some accounts. Verify eligibility before making it a dependency.
For a proposed holdout, agree assignment before delivery and retain it through observation. Choose an assignment unit that fits the app's identity and suppression setup. Compare the useful action across the assigned groups, including returns without an attributed ad interaction. Avoid restricting the readout to people who clicked or returned, because that would discard the original comparison population.
Holdout planning checklist
- Define the eligible population and assignment unit before exposure.
- Specify the allocation method, holdout size, and evidence needed for a useful read.
- Confirm the control group is excluded from the tested advertising.
- Record campaign overlap and the treatment of owned messages in both groups.
- Set the primary event, observation window, reporting cutoff, and revenue basis.
- Preserve assignment and compare outcomes on the agreed population basis.
- Plan how to report uncertainty, contamination, and an inconclusive result.
- Assign the decision owner and conditions for continuing or stopping.
Use the incrementality planning scope to resolve design dependencies before launch. Review useful actions, value, spending, and uncertainty together. If the available audience or controls cannot support an incremental estimate, retain a clearly labeled attribution read and document the experiment needed to answer the remaining question.


