Google retired most of the Android Privacy Sandbox in October 2025 — Topics, Protected Audience, and Attribution Reporting are being wound down after low adoption — and the Advertising ID (GAID) that everyone built attribution around is, for now, still working exactly as before. Google’s own documentation commits to keeping GAID functional for at least two years from the announcement, with substantial notice ahead of any change. That’s good news in the short term and a trap in the medium term: “status quo for now” is quietly becoming the excuse teams use to not do the attribution resilience work they were already supposed to do for Privacy Sandbox, just under a new deadline nobody has written down.
The teams who got burned by iOS’s ATT rollout learned the expensive way that “we have time” turns into “we didn’t build a fallback” almost every time. GAID’s two-year runway is exactly the kind of long, comfortable timeline that produces the same outcome on Android if nobody’s tracking how dependent the business actually is on it right now. Without that visibility, the day Google does start winding GAID down, attribution reporting breaks all at once instead of degrading gracefully into a system that was already built to handle it.
Data Points to Track
- GAID dependency ratio — the share of installs and conversions currently attributed via GAID versus via SKAdNetwork-equivalent, probabilistic, or first-party methods
- Attribution method per campaign and network, logged explicitly rather than assumed, so you know exactly which channels have no fallback today
- Consent and opt-out rates for advertising ID access, tracked over time as a leading indicator — rising opt-outs erode GAID’s usefulness well before Google changes anything
- Match rate deltas between GAID-based attribution and any parallel first-party or modeled attribution you’re already running, to quantify what you’d lose
- Regulatory signal tracking — jurisdiction-specific rules affecting advertising identifiers, since regional restrictions will likely arrive before a global GAID change does
Setup Steps
- Audit every attribution and measurement integration for GAID dependency today, not as a future project — treat the two-year clock as already running.
- Stand up a parallel first-party attribution path (deterministic login-based matching, server-side conversion events, or probabilistic modelling) for at least your highest-spend channels, so there’s a working comparison baseline before you need one.
- Track consent and advertising-ID-availability rates as a first-class metric, not a compliance afterthought, since declining availability degrades GAID’s accuracy gradually and predictably.
- Set an internal review cadence (quarterly is reasonable) to reassess GAID dependency ratio and fallback readiness, rather than waiting for a Google announcement to trigger the work.
- Document a rollback plan per channel — which attribution method takes over if GAID access for that channel drops significantly — so the plan exists before it’s needed.
Actionable Insights
A high GAID dependency ratio on your top-spend channels combined with no working fallback is the single clearest priority signal this data can give you — it’s the channel where a future change costs you the most and where you’re currently least prepared. Rising advertising-ID opt-out rates are worth watching even if GAID itself doesn’t change, because they erode attribution accuracy on exactly the same timeline whether or not Google acts. And having a parallel first-party attribution path already running means that whenever GAID’s two-year clock does run out, you’re switching a default rather than building a system under deadline pressure.
Related Resources
Need help tracking this in your app?
Our team sets up analytics pipelines for mobile and web teams every day. Talk to us and get your first events flowing in under an hour.
Talk to an expert