Google has started zeroing out the Android Advertising ID (GAID) for any user who turns on “Opt out of Ads Personalization” in their device settings, mirroring what Apple did to IDFA with App Tracking Transparency. When that happens, every install, event, and conversion tied to that device suddenly carries an ID of all zeros instead of a unique identifier — and any attribution or analytics pipeline that assumes GAID is always populated will silently start misattributing or dropping that traffic rather than throwing an obvious error. Google has also shipped an alternative identifier for analytics and fraud detection use cases that doesn’t depend on ad personalization consent, but it’s opt-in for developers to adopt, not automatic.
Right now only a small share of Android users have flipped the opt-out setting, so the immediate volume impact looks minor if you check it in isolation. That’s exactly what makes this a tracking problem rather than a business problem yet: a low, growing percentage of zeroed GAIDs sitting inside your existing funnel and attribution reports doesn’t show up as a dashboard anomaly, it shows up as a slow, unexplained drift in install-to-purchase conversion rates and organic-vs-paid splits that nobody can pin down to a cause. Catching it now, while volume is low, is far cheaper than trying to reconstruct six months of quietly corrupted attribution data later.
Data Points to Track
- Zeroed-GAID event share, tracked as a percentage of total Android events per day, so you can see the opt-out rate rising before it distorts headline metrics
- Attribution source on zeroed-GAID installs, flagged separately from normal installs rather than folded into “organic” or “unknown” by default
- Adoption status of Google’s replacement analytics identifier, per app version, so you know exactly which release started reporting the new signal
- Fraud/bot detection scores split by GAID state, since zeroed-GAID traffic has historically included more automated and fraudulent installs mixed in with genuine opt-outs
- Revenue and LTV reported separately for zeroed-GAID cohorts, so a shrinking measurable audience doesn’t quietly deflate blended average metrics
Setup Steps
- Add a GAID-state field to every install and event record — populated, zeroed, or missing — rather than assuming the ID is always usable as a join key.
- Evaluate and integrate Google’s replacement analytics identifier in your SDK layer now, ahead of opt-out rates climbing further, rather than treating it as optional.
- Update attribution logic to route zeroed-GAID installs into an explicit “opted-out” bucket, not the default “organic” bucket, so paid spend doesn’t get misjudged as underperforming.
- Backfill a baseline measurement of your current zeroed-GAID share this week, so any future change is measured against a real number, not an assumption of zero.
- Set an alert on the zeroed-GAID share crossing a threshold (start with 5%), so a shift in opt-out behaviour surfaces as a metric, not as a support ticket about broken numbers.
Actionable Insights
Once zeroed-GAID traffic is broken out explicitly, you can answer the question that actually matters: is your paid acquisition genuinely getting less efficient, or is a growing slice of it just becoming unmeasurable? Those require completely different responses — one is a targeting or creative problem, the other is a measurement-infrastructure problem that needs the replacement identifier wired in faster. Tracking the opt-out share over time also gives you an early warning for budget conversations: if a channel’s zeroed-GAID rate is climbing faster than the platform average, that’s worth flagging before a quarterly review shows an unexplained ROAS drop with no obvious cause.
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