8. Privacy, Attribution, and the Measurement Reset ### 8.1 The Post-ATT Measurement Stack The era of deterministic, user-level attribution is over. ATT opt-in rates have stabilized at **15–30%** (Zoomd), meaning the majority of iOS users are invisible to traditional attribution methods. The measurement stack has been rebuilt around: - **SKAdNetwork (Apple)**: Apple's privacy-preserving attribution framework, which provides aggregate, delayed conversion data rather than user-level tracking. It has matured significantly through 2025, with conversion value optimization becoming the primary optimization lever. - **Privacy Sandbox (Google)**: Google's equivalent framework for Android, providing privacy-preserving APIs for attribution and targeting. It gained significant traction in 2025. - **Media Mix Models (MMM)**: Top-down statistical models that estimate the incremental impact of marketing spend across channels, without requiring user-level data. - **Incrementality testing**: Controlled experiments (geo-holdouts, conversion lift studies) that measure the true causal impact of advertising. ### 8.2 First-Party Data as the New Competitive Moat As third-party data has degraded, **first-party data strategies have become essential.** Apps that can collect, organize, and activate their own user data have a significant advantage in: - Building lookalike and predictive models from their own high-value users - Retargeting existing users with relevant messaging (remarketing share rose from 25% to 29% of total spend) - Feeding AI optimization systems with higher-quality signals than what's available through privacy-constrained ad networks **Implication for founders**: The data infrastructure you build is not just a retention tool — it is an acquisition asset. Apps with rich first-party data can train better predictive models, optimize bids more effectively, and reduce their dependence on expensive third-party targeting. ### 8.3 Web-to-App Routing A growing trend in 2026: **routing traffic through web landing pages before driving users to install.** This approach (Admiral Media): - Provides richer attribution data than direct app store installs - Allows pre-qualification of users before the install event - Creates an additional surface for creative testing and messaging optimization - Reduces dependence on app store attribution limitations This is particularly relevant in a world where AI assistants are surfacing web content before app store listings — a well-optimized web landing page becomes both an acquisition surface and an attribution bridge. ---