Meta App Campaigns: AEM Ranking, AEO and Value Optimisation

Meta gives an app eight event slots in Aggregated Event Measurement, ranked in priority order, and that ranking is the single most consequential configuration in your account. Choose it badly and every iOS optimisation decision downstream inherits the error, invisibly, for as long as it stands. Most accounts we take over have all eight filled with whatever the SDK emitted by default, in the order the SDK emitted them. We start there, then move the account from install optimisation to app event or value optimisation on a schedule the data supports. The sequence we work in: The event ranking first, argued from your cohort revenue and rewritten in Events Manager. Then app event optimisation on the deepest event that produces enough weekly volume for a Meta ad set to exit learning. Then value optimisation, but only once purchase or subscription values are passed with enough variance to be worth optimising - a value stream where every event carries the same number is a count wearing a costume, and VO will spend against it happily. Advantage+ and manual, decided rather than assumed: Advantage+ app campaigns win in some accounts and lose in others, and the honest way to know is a split test with real budget on both sides for at least two weeks, not a rollout because it is the newer product. We run that test, publish the result including when it goes against our prior, and then consolidate. We do not run both permanently side by side and call it diversification; that is two campaigns competing in the same auction with your money on both sides. Attribution, said plainly: Meta's modelled iOS numbers, your MMP's SKAdNetwork postbacks and your own in-app events will not agree, and the gap is usually twenty to forty percent. We use Meta's numbers for in-platform optimisation because that is what the delivery system responds to, your MMP for cross-channel budget allocation, and a quarterly geo lift test - or Meta's own conversion lift where the account is large enough to qualify - to decide what Meta is actually worth. The lift result is what we bring to your board slide, not the platform ROAS. Creative, and the part we will not sell: Concepts are built from what your best-performing organic and ASO assets already prove about the promise people respond to, then tested as concepts rather than as edits. Three or four distinct arguments a quarter, each with enough spend to reach a verdict. We do not deliver a monthly quantity of variations, and we say no to retainers written that way, because paying by the asset makes the agency's interest the opposite of yours the moment a concept starts working. What we refuse: Install optimisation as a standing state - it is defensible for two to four weeks on a new market entry and indefensible after that. Broad targeting sold as a strategy; it is a default. Running a marketplace's supply and demand sides in one campaign structure, since they are different products with different economics and the auction will feed the cheaper one. Reporting Meta-attributed revenue without the reconciliation to your MMP alongside it. Not for: Apps below the weekly event volume where an ad set can exit learning on anything deeper than an install - we will calculate that with you before quoting and tell you to wait. Regulated categories where Meta's app ads policy is the binding constraint rather than the media. Teams who want daily creative reporting; the reads are weekly here and asking for them daily makes the account worse. Setup and running: First three weeks are configuration - AEM ranking, event definitions, SKAN alignment, domain and app verification, dataset quality checks. Then weekly management with a written change note, monthly decisions, quarterly incrementality. You keep the Business Manager, the ad account and the billing.