Google’s May 2026 update to GA4 gives AI assistant traffic its own dedicated channel for the first time. Visits from ChatGPT, Gemini and Claude are now grouped and reported under a single “AI Assistant” channel, sitting alongside organic search, paid and direct in your acquisition reports.
For anyone tracking this manually, it is a welcome step forward. But it’s not the complete picture.
We were never flying completely blind
At Loom, we have been reporting on LLM traffic for some time. GA4 already allowed you to isolate referrals from AI platforms using secondary dimensions and custom channel groupings in Looker Studio. Our clients have had dedicated LLM traffic slides in their monthly reports built this way.
The GA4 update formalises that. It makes the data cleaner, more accessible and easier to act on. For teams that have not yet set this up, it removes a significant barrier.
The more significant change is in Search Console
What GA4 captures is referral traffic: users who clicked a link in an AI assistant and landed on your site. That is one part of the picture.
AI Overviews and AI Mode are a different matter. When Google’s own AI features show your content in a response, those impressions and clicks have always been folded into your standard Search Console data. There was no way to separate them from traditional search results.
That is now changing. Google is rolling out a dedicated generative AI performance report in Search Console, currently being tested with a number of UK websites It will show impressions, which pages appear in AI responses and in which countries.
We’ve been using the AI Performance report BETA in Bing Webmaster tools to share insights with our clients. Alongside the Google report, it promises to provide consistent data direct from source.
For SEOs, this is the update that matters most. High impression counts with declining CTRs have long been a signal that AI features may be absorbing visibility without passing clicks. This report will make that pattern measurable, not inferred.
What marketers should actually be reporting on
Start with what is driving traffic to your site. For some clients, ChatGPT is the dominant AI referrer. For others, it is Perplexity. The mix varies by sector and audience, and you will not know until you look.
From there, layer in Search Console data as the generative AI report rolls out. AI Overviews and AI Mode matter strategically even if they do not generate referral clicks in the traditional sense. Visibility without traffic is still visibility, and it shapes how your brand is perceived at the point of search.
The two data sources together, GA4 referrals and Search Console AI impressions, start to give you a connected view of how AI features are influencing the full journey, not just the ufinal click.
This is not separate from SEO
There is a lot of noise at the moment about GEO as a discipline distinct from SEO. The reality is more straightforward. The foundations are the same: good structure, strong EEAT, content written for people rather than for algorithms. Writing in clear, structured sections is good web writing. The human touch is what E-E-A-T has always asked for.
The difference worth acknowledging is technical. How JavaScript is crawled by LLMs differs from standard Googlebot behaviour, so assuming your current SEO setup covers it carries some risks. It is worth auditing specifically for AI accessibility, rather than assuming SEO coverage extends automatically.
The connected picture
A user can start their research in ChatGPT, encounter your brand on social media and complete a branded search three weeks later. Each of those touchpoints sits in a different report. None of them, in isolation, tells you what actually drove the conversion.
These new tools are not the final answer to that problem. But they fill in a gap that has made joined-up performance reporting harder than it should be. For businesses running organic search optimisation and paid media in parallel, this data belongs in the same conversation.
Tracking AI traffic is now a baseline expectation. The teams that will use it well are the ones treating it as part of a connected system, not as a new channel to monitor in isolation.
What this looks like in practice
At Loom, AI and LLM traffic is already part of how we report for clients. It sits alongside organic, paid and direct data as part of a single performance view. Our SEO and data teams work together to make sure performance reporting reflects how people actually find and engage with your business, across every channel that matters.
Looking for a trusted partner to help you navigate these changes? Book a discovery call with Tommy to discuss your current strategy.
