For teams that already built those custom groups, the immediate value is the ability to run the two setups side by side and compare results. Google has not published a full list of recognized referrers, which means the custom regex patterns the company recommended last August remain necessary for covering platforms outside the small set already included. The two approaches are complementary, not redundant, and the prudent move is to maintain both until the native referrer list expands to a degree that makes the custom version superfluous.<\/p>\n\n\n\n
The larger significance of the change is structural. AI assistant traffic now appears as a distinct line item in acquisition reports, user reports, and channel group reports, directly comparable to organic search, direct traffic, and referrals. That removes the friction of manual filtering and makes it straightforward to analyze conversion behavior, session duration, and volume without workarounds. For organizations that have been trying to quantify how much of their audience is arriving through AI interfaces, this eliminates a major blind spot. The data was always somewhere in the logs, but it was scattered across referral strings and required interpretation. Now it surfaces by default, in the same reports, next to the channels it is being measured against.<\/p>\n\n\n\n
Kevin Indig, Growth Advisor at Growth Memo, described the change as overdue, noting that he had publicly raised the issue shortly before the announcement. Johan Strand, Senior Digital Analyst and Partner at Ctrl Digital, advised teams that already have custom channel groups in place to audit and adapt them now, since the native grouping may not capture every source their custom setup was designed to track.<\/p>\n\n\n\n
The open question is how quickly Google will expand the recognized referrer list as new AI platforms gain adoption. For now, the named examples cover the dominant players, but the category is evolving rapidly, and organizations tracking less common assistants should keep their custom groups active.<\/p>\n\n\n\n Google has completed the deprecation of FAQ rich results<\/a>, a process that began years ago with an initial reduction in eligibility and has now reached its endpoint. The FAQ search appearance no longer appears in Google Search results, and the company has added a notice to its structured data documentation confirming the change. No separate blog post or public explanation accompanied the final step.<\/p>\n\n\n\n The practical consequences are phased. The FAQ search appearance filter in Search Console, the rich result report, and support in the Rich Results Test will be removed in June. API support will end in August. For organizations whose reporting pipelines extract FAQ-specific data through Google’s API, those API calls must be updated before the August cutoff to avoid errors. The markup itself can remain in place without causing technical issues, but it will no longer generate the visual rich result that once made it valuable for click-through optimization.<\/p>\n\n\n\n The deprecation leaves an open question about whether FAQ schema influences AI-generated search responses. Google’s move does not answer that question one way or the other. Organizations that kept FAQ markup on the assumption that it might benefit visibility in AI Overviews or similar surfaces will have to look elsewhere for evidence. The deprecation itself is not evidence of harm, but it is also not evidence of continued utility.<\/p>\n\n\n\n A study published by Ahrefs<\/a> has injected a dose of empirical caution into the conversation around structured data and AI visibility. The research tracked 1,885 pages that added JSON-LD schema and compared them against control pages that never added schema, measuring changes in citations across Google AI Overviews, AI Mode, and ChatGPT over 30-day windows.<\/p>\n\n\n\n The results were essentially flat. AI Overviews citations for the treated pages declined by 4.6 percent relative to the control group. AI Mode showed a 2.4 percent increase, and ChatGPT showed a 2.2 percent increase, but both movements were small enough to be indistinguishable from statistical noise. The study found no evidence that adding schema caused a meaningful lift in AI citations<\/a> for pages that were already being cited.<\/p>\n\n\n\n This matters because the correlation between schema usage and AI citation frequency has been widely cited as a reason to prioritize structured data implementation. Sites that use schema tend to invest more in content quality, site authority, and link building, and those factors may explain the correlation better than the markup itself. The Ahrefs study does not claim that schema never helps, nor does it address whether structured data could benefit pages that are not yet visible to AI systems. That is a different population requiring its own study. But for pages that already appear in AI citations, the evidence suggests that adding JSON-LD is unlikely to be the factor that moves the needle.<\/p>\n\n\n\n![\"Google](\"https:\/\/i.ibb.co\/zHRjb473\/image.png\")
FAQ Rich Results Are Gone, and the Deprecation Is Final<\/h2>\n\n\n\n
New Research Questions Whether Schema Actually Drives AI Citations<\/h2>\n\n\n\n