Google Just Made AI Overview Traffic Untrackable. Here's What You Can Still Measure

Published June 2026. This covers a developing situation; dates and details are current as of writing.

In early May 2026, Google quietly removed the signal that let websites identify clicks coming from AI Overviews, and announced no replacement. If you run a site, here is what that means in plain terms: the AI-generated answer box that now appears on roughly 48% of searches, and that has cut organic click-through rates by around 61%, just became a black box you cannot measure directly. Google is reshaping how half of all searches behave, and it took away the one instrument you had for seeing it.

Most coverage of this is, understandably, a lament. This is the opposite: a practical guide to what you can still measure, because the situation is less hopeless than the headlines suggest, and the teams that adapt now will have a measurement edge while everyone else stares at a gap.

What exactly went dark

To be precise about the problem, because precision is what is missing from the panic: Google removed the method analytics tools used to attribute visits to AI Overview clicks. The visits still happen; you simply can no longer cleanly label them as "came from an AI Overview" in your traffic data. For now, AI Overview presence is trackable only through rank-tracking tools (which tell you an Overview appeared for a keyword), not through your analytics (which would tell you what it did to your actual traffic). That split, presence without traffic attribution, is the gap.

Pair it with the scale (48% of searches, 61% CTR drop where Overviews appear) and you have a channel that is simultaneously your biggest organic threat and your least visible one. That is a bad combination, and it is why this is worth acting on rather than mourning.

What you can still measure (the actual workarounds)

The signal Google removed is one specific convenience. The underlying reality leaves several fingerprints you can still read.

1. The organic CTR anomaly. You cannot see the AI Overview click directly, but you can see its shadow in Search Console: keywords where you hold a top-three rank but your CTR has fallen off a cliff are almost certainly losing clicks to an Overview. Cross-reference your rank-tracker's "Overview present" flags with Search Console CTR by query, and the AI Overview impact resolves into a list of specific keywords, even without the removed signal. It is reconstruction instead of direct measurement, but it works.

2. The "Direct" and brand-search bump. When an AI Overview answers someone's question and they later visit you by name, that visit lands in Direct or branded organic search, not in the Overview's column (which no longer exists anyway). A rising Direct and branded-search baseline, correlated with high-Overview keywords, is the indirect influence showing up where it always did. Watch the trend, not the absent label.

3. Server logs for the crawl side. AI Overviews are built on what Google's crawlers ingested. You cannot see the Overview click in logs, but you can see Googlebot's behavior, and combined with the AI bot picture ↗, it tells you which content is feeding the answers that may be replacing your clicks.

4. The channels Google does not control. Here is the strategic pivot: while Google makes its own surface less measurable, the AI assistants that send real referral traffic (ChatGPT, Perplexity, Claude) still pass identifiable referrers, and that traffic is growing fast ↗. The rational response to one channel going dark is to measure the adjacent channels that are still lit, and that are increasingly where the clicks are going anyway.

The bigger lesson this gap teaches

This is the second time in two months that a platform has made measurement harder by fiat (Google here, and the broader shift to ad-supported AI answers that muddies AI referral data elsewhere). The pattern is the lesson: the more your measurement depends on a platform's goodwill, the more exposed you are when the platform changes its mind. Google can remove a tracking signal overnight because it owns the surface. It cannot remove the data you collect first-party, on your own domain, about visits that reach your own site.

That is the durable position. You will never perfectly attribute an AI Overview click that Google has decided to hide. You can absolutely build a measurement setup that does not break every time a platform reshuffles, by owning the parts that are yours: first-party, cookieless collection of what actually arrives, tied to what it is worth ↗.

Stop depending on platforms that can go dark overnight. Datalenk measures the traffic that reaches your site first-party (including the AI assistant referrals that are still trackable), tied to revenue, so a platform's tracking change does not blind you. Try it free.

What to do this week

1. Build the CTR-anomaly report: rank-tracker "Overview present" keywords cross-referenced with Search Console CTR. That is your best AI Overview impact map now.
2. Watch Direct and branded search as proxies for indirect AI influence, by trend not by absolute.
3. Instrument the still-lit channels: make sure AI assistant referrals (ChatGPT, Perplexity, Claude) are classified and measured, because that is where the trackable AI traffic is ↗.
4. Move your source of truth first-party, so the next platform tracking change is an inconvenience, not a blackout.

The signal Google removed was a convenience built on Google's surface. The measurement that survives is the one you own.

FAQ

What did Google change about AI Overview tracking? In early May 2026 Google removed the method that let analytics tools identify clicks coming from AI Overviews, with no replacement announced. AI Overview presence is now visible only through rank-tracking tools, not through traffic attribution in your analytics.

Can I still track AI Overview traffic? Not directly. You can reconstruct its impact by cross-referencing rank-tracker "Overview present" flags with Search Console CTR drops on top-ranked keywords, and by watching Direct and branded-search trends as proxies for indirect influence.

How big is the AI Overview impact? AI Overviews appear on roughly 48% of searches as of early 2026, and reduce organic click-through rates by about 61% where they appear, which makes the loss of direct tracking especially costly.

What is the long-term fix? Reduce dependence on platform-provided tracking signals by collecting first-party data on what reaches your own site, and by measuring the AI assistant channels (ChatGPT, Perplexity, Claude) that still pass identifiable referrers and are growing.