iOS Measurement Rebuild: SKAdNetwork 4.0 and Aggregated Events

SKAdNetwork 4.0 gives you 64 fine values in a single 6-bit field, three postback windows, and coarse low, medium or high when the install cohort is too small to clear the crowd anonymity threshold. Most schemas we inherit spend all 64 values on day-zero revenue buckets, which means the second and third postbacks carry almost no information and the account has no way to learn which users came back. Roughly 30 percent of your iOS users will grant ATT consent. The other 70 percent are the reason this schema is the most consequential decision in your app account. What we do differently: We design the conversion value schema against the decision you actually need to make, then work backwards to the bits. If your payback question is day-7 subscription retention, the first postback carries an early activation signal and the later windows carry the retention state, and we accept a coarser revenue read as the price. Agencies default to revenue buckets because it demos well in a dashboard. It also produces a bidder that cannot tell a trial from a renewal. What the work covers: On app: conversion value mapping across all three postback windows, lockWindow decisions per window, coarse value fallbacks written explicitly rather than left to the MMP default, and the crowd anonymity math for your actual install volume by campaign. Implementation in AppsFlyer SKAN Conversion Studio, Adjust, or Singular, against the SDK version you are actually shipping. We check that your postback endpoint is receiving from Apple directly as well as through the MMP, because the two disagree more often than people expect. On web: Aggregated Event Measurement. You get eight events per verified domain and their priority order is the whole configuration. We set it against margin, not against funnel depth, which is why Purchase is not automatically slot one for a business with a subscription upgrade path. Domain verification, subdomain handling for separate storefronts, and the 72-hour aggregation delay written into how your team reads the dashboard so nobody panics on a Tuesday. On TikTok: SKAN campaign configuration and the parallel non-SKAN measurement for Android and web, kept separate in reporting so the two are never accidentally summed. What we refuse: We will not write an ATT pre-prompt that misleads users into opting in, and we will not test one. We have written pre-prompts that raised opt-in by explaining the trade honestly, and that is the only version of this work we do. We will not present SKAN postbacks and MMP-attributed installs side by side as if they were the same measurement. We do not offer probabilistic fingerprinting for iOS attribution; it is against Apple's terms and it will eventually cost you the app, not the campaign. What you receive: The conversion value schema as a documented mapping, with the reasoning for each bit allocation written down so a future team can argue with it. MMP configuration applied and screenshotted. AEM priority order with the margin logic behind it. A reading guide for your analytics team covering what each postback can and cannot answer, including the questions your existing dashboard will now correctly refuse to answer. Not included: SDK integration from scratch, media buying, creative, and web-to-app funnel development. Server-side app event forwarding, which is separate work. Who this is not for: Apps below roughly 3,000 iOS installs a month, where crowd anonymity suppresses fine values often enough that the schema barely matters and simpler measurement will serve you better. Teams who want the schema to make iOS look as measurable as Android. It will not, and the value of this engagement is knowing exactly where the edge of your knowledge sits.