App Retargeting Built Around Low ATT Consent Rates

What paid retargeting can actually address: On iOS a paid retargeting campaign can only reach users who granted ATT. In the categories I work in that consent rate runs between 21 and 34 percent, and for a medtech app whose permission prompt sits next to a health data prompt it is usually at the bottom of that range. So roughly two thirds of your iOS base is unreachable by paid retargeting, permanently. Any plan that does not open with that number is selling you a channel at a size that does not exist. Where returning users actually come from: Push, in-app messaging, email and web to app, sequenced against a lifecycle definition, before a euro of paid retargeting. If you run Braze, CleverTap or OneSignal, the first weeks here go into segment definitions and link routing, because the usual cause of a broken re-engagement programme is not targeting. It is that the link drops the user on the home screen instead of the screen the message promised. Universal Links and App Links, deferred routing through your MMP's link product, tested per OS version rather than assumed working because it worked once. Where paid retargeting earns its place: Android at scale, and the consented iOS slice when the segment is large enough to clear a demand-side platform's minimum. Segments are built from behavioural audiences, not from recency alone. Opened thirty days ago is not a segment, it is a date filter, and bidding against it is how retargeting budgets get spent on people who were coming back regardless. Incrementality, or no campaign: This is the channel most likely to bill you for users you already had. Every programme here runs a holdout, a geo split where the market allows one and a randomised holdout inside the audience where it does not. The number I report is incremental returning users, not attributed ones. A mature programme usually measures somewhere between 30 and 60 percent incremental on a first clean read. If yours comes back under 20 percent after two cycles I will recommend cutting the budget and put it in writing, which has happened more than once. Privacy Sandbox on Android: The Attribution Reporting API and Protected Audience for Apps change how these audiences are assembled. I do not file this under future problems. The segment architecture I build is one that survives the loss of the advertising identifier rather than one that has to be rebuilt the week it goes. Regulated categories: A re-engagement message to the user of an MDR-regulated app is, in several member states, subject to the same claim rules as the listing. A push notification urging a clinical action is not a growth asset. Message trees are reviewed before they are scheduled, and the segments that cannot be messaged at all are documented at the start so nobody builds a campaign against them. Not included: New user acquisition, creative production, CRM tooling implementation, and web retargeting for the marketing site. Who should not buy this: Apps with fewer eligible returning users than a demand-side platform will accept as an audience. Teams who want this channel measured on last-touch attributed opens, because I will not report that as a result. And teams unwilling to hold out a control group. Without one, nothing said about retargeting is knowable, and I would rather decline the fee than produce a number I cannot defend.