Google has officially wound down what remained of Privacy Sandbox on both Chrome and Android — Topics, Protected Audience, and Attribution Reporting are all being retired, closing out an initiative that was supposed to replace cross-app identifiers with privacy-preserving on-device measurement. The one piece of that story that survived is the Google Advertising ID itself: Google has committed to preserving GAID for at least two more years, with substantial notice before any future change. For teams that spent the last year building or piloting Privacy-Sandbox-based attribution, the practical result is a return to GAID-based, deterministic attribution as the default — but on a runway with a stated end date, not a guarantee.
That combination is exactly what makes this worth instrumenting rather than just reading about. A team that quietly reverts to “GAID works again” without tracking its availability rate has no early warning if match rates start slipping again, and a team that left half-finished Privacy Sandbox integration code in its attribution SDK risks that code silently no-op’ing or throwing on retired endpoints without anyone noticing until an install cohort’s attribution data comes back empty. The API being gone doesn’t mean every call your app used to make to it disappeared with it.
Data Points to Track
- GAID availability rate per install, tracked as a trend line over time rather than assumed to be a stable constant now that it’s “back”
- Attribution match rate for GAID-based installs, split out from any residual probabilistic or Privacy-Sandbox-based fallback still wired into your attribution SDK
- Any remaining calls to Topics, Protected Audience, or Attribution Reporting APIs left in your codebase from a migration attempt, and whether they now error, no-op, or fail silently
- Limit ad tracking / opt-out rate trend for Android users, to catch whether the shutdown itself changes user opt-out behaviour rather than assuming it’s unaffected
- Attribution source breakdown — SDK network attribution, GAID-based, and referrer-based — logged per install so a future GAID change doesn’t blindside the acquisition funnel the way this one caught some teams off guard
Setup Steps
- Find and remove or safely guard any Privacy Sandbox API calls added during a migration attempt, so a retired endpoint fails loudly in staging rather than silently in production.
- Confirm your attribution SDK’s fallback path is instrumented and tested for the case where GAID becomes unavailable on a given device, rather than assuming that path is dead code now.
- Stand up a dashboard tracking GAID availability rate as a leading indicator, so a future policy or platform change shows up as a trend before it shows up as a funnel drop.
- Re-point any reporting or dashboards built around Privacy-Sandbox-based attribution back to GAID-based sources, and archive rather than delete the old configuration in case it’s needed again.
- Document your current attribution architecture now, including exactly which identifier and fallback each channel relies on — the two-year GAID commitment is a countdown clock, not a permanent fix.
Actionable Insights
The direct lesson from this reversal is that attribution architecture built around a single identifier — Privacy Sandbox APIs, GAID, or whatever comes next — is only as reliable as that identifier’s own lifespan, and the last two years of Android attribution history make clear that lifespan is not guaranteed. Tracking GAID availability and match rate as ongoing metrics, rather than one-off migration checkpoints, turns the next identifier change into a gradual trend your team sees coming instead of a step change in acquisition data that shows up unexplained in a monthly report.
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