GA4 Now Tracks AI Assistant Traffic, FAQ Results Disappear

This week’s Pulse focuses on critical shifts in how you measure AI-driven traffic, the fading value of certain structured data, and a stark warning from a publishing giant about the future of search. These changes demand immediate attention to your analytics setup and strategic planning.

Google Analytics now automatically categorizes traffic from recognized AI chatbots into a dedicated “AI Assistant” default channel group. This replaces the previous reliance on custom regex patterns to isolate these visits from general referrals. Sessions from recognized assistants like ChatGPT, Gemini, and Claude will now carry the medium “ai-assistant,” route to the new channel, and receive a reserved “(ai-assistant)” campaign label. While Google hasn’t published the full list of recognized referrers, all three changes occur automatically.

For anyone who built custom channel groups to track AI chatbot traffic, this update offers a direct comparison against Google’s native version. The custom regex patterns Google recommended last August remain useful for platforms outside the recognized referrer list, and both systems can run concurrently. The real opportunity lies in what you do with this data. AI assistant traffic now appears as a distinct line item in acquisition, user, and channel reports, making it straightforward to compare conversion behavior, session quality, and volume against organic search without manual workarounds. However, since Google hasn’t specified how quickly the referrer list will expand, keep your custom groups active for any AI assistants beyond the three named examples.

Industry voices weighed in on the change. Kevin Indig, Growth Advisor at Growth Memo, called it overdue, noting he had just complained about the lack of this feature on stage. Johan Strand, Senior Digital Analyst at Ctrl Digital, advised adapting existing custom channel groups to align with the new system.

In a related development, Google has completed the deprecation of FAQ rich results, a process that began a few years ago. The company added a notice to its FAQ structured data documentation without a formal blog post. FAQ rich results no longer appear in search, and Google will remove the FAQ search appearance filter in Search Console, the rich result report, and support for the Rich Results Test in June. API support ends in August. If your reporting pipelines rely on FAQ-specific data from the API, those calls must be updated before the August cutoff. Leaving the markup in place won’t cause problems, but it no longer generates visible results. Whether FAQ schema aids AI search remains an open question.

Meanwhile, a new Ahrefs report casts doubt on the causal link between schema markup and AI citations. The study tracked 1,885 pages that added JSON-LD schema and found no meaningful increase in citations across Google AI Overviews, AI Mode, or ChatGPT. Each treated page was matched against controls that never added schema, with changes measured over 30-day windows. AI Overviews actually showed a 4.6% decline relative to controls, while AI Mode (+2.4%) and ChatGPT (+2.2%) showed changes too small to distinguish from noise.

The report challenges the widely cited correlation between schema and AI visibility. Ahrefs tested whether the relationship appeared causal and found no evidence of a meaningful lift, at least for pages already being cited. Sites with schema tend to also invest in better content, stronger authority, and more links, factors that may explain the correlation better than the markup itself. The report can’t address whether schema helps pages that aren’t yet visible to AI systems, but for pages already earning citations, adding JSON-LD is unlikely to be the unlock. Chris Long, Co-founder of Nectiv, noted on LinkedIn that this data is changing his perspective on the effectiveness of schema for influencing AI citations.

Perhaps the most sobering story this week comes from Condé Nast CEO Roger Lynch, who told his teams to plan their businesses as if search traffic will be zero. In an interview on TBPN, a tech talk show OpenAI acquired in April, Lynch described three consecutive years where internal forecasts underestimated actual declines in search traffic. He expects search to settle at a single-digit percentage of total traffic. Lynch pointed to a “barbell effect” where large, authoritative brands and small, niche publications perform well, while brands in the middle are most exposed. Condé Nast’s digital subscriptions grew 29% in revenue last year.

Lynch’s comments align with third-party data showing a 60% decline in search referrals for small publishers over two years, according to Chartbeat, and a Reuters Institute finding that media leaders expect search traffic to fall by more than 40% over three years. The difference is that a CEO running Vogue, The New Yorker, and GQ is now building budgets around those numbers. The barbell observation is worth testing against your own client portfolio or publishing operation. Brands without deep category authority or a strong niche focus lack a clear path forward as AI Overviews, commerce links, and sponsored results fill the page before organic listings appear. Kevin Indig noted that there is no escape hatch for publishers in the age of answer engine optimization.

The theme of the week is clear: the measurement infrastructure is finally catching up to the problem. The tools and signals that defined search visibility for years are being deprecated, questioned, or abandoned by the publishers who depended on them. FAQ rich results are gone. Schema’s role in AI citations is weaker than the correlation suggested. A major publisher is planning as if search traffic won’t recover. Each story involves an environment where the old measurement infrastructure no longer matches the landscape.

The GA4 update is the other side of that coin. Google is building native tracking for the traffic source that’s growing while the traditional one contracts. AI assistant traffic is still a fraction of what search delivers, but it’s now visible by default, in the same reports, next to the channels it’s measured against. That’s a shift worth watching, and acting on.