A New Report, With a Catch

On June 3, 2026, Google rolled out a dedicated “Search Generative AI performance” report inside Search Console, giving site owners a first look at how their pages show up in AI Overviews and AI Mode. It started with a limited test group in the UK, then expanded on June 23, 2026 to the US, India, Switzerland, and additional markets — though the rollout is still ongoing and not yet available to every property. For an industry that has spent the better part of two years guessing at AI visibility from indirect signals, this is a meaningful shift: it is the first time Google has offered anything resembling official data on the topic.

The catch is in what the report leaves out. It breaks down impressions by page, country, device, and date, which is genuinely useful for spotting which content is being pulled into AI-generated answers. But it does not include click data, click-through rate, average position, or query-level detail — the exact numbers most teams actually use to judge whether visibility is translating into anything. You can see that a page was surfaced. You still cannot see what someone searched to get there, or whether they visited your site afterward.

Why This Report Exists Now

The timing lines up with a broader shift in how people search. New research from SparkToro and Similarweb, based on US data from January through April 2026, found that 68.01% of Google searches now end without any click at all — up from 60.45% just two years earlier. AI Overviews are a major part of that story: they now appear on more than 20% of all Google searches, and when one shows up, the long-standing #1 organic result loses roughly 58-60% of the clicks it used to get.

Interestingly, standalone “AI Mode” — the more chatbot-like search experience — still accounts for a small share of sessions, around 0.34%. That suggests the bigger disruption to organic traffic right now isn’t a separate AI product people are switching to; it’s AI Overviews quietly reshaping the results page most people already use. Google adding an AI-specific report to Search Console, alongside a new toggle that lets site owners opt out of having their content used in generative AI features, reads as a direct response to that pressure — an acknowledgment that “your page ranked” and “your page got a visit” have become two different claims that need two different metrics.

What You Can Actually Do With the Data

Even without clicks or queries, the impressions data is a useful diagnostic. If certain pages show high AI impressions but flat overall traffic, that’s a signal those pages are being used as source material for AI answers without earning the visit — useful context when planning content investment. Comparing AI impressions against your existing Performance report (which does have queries and clicks) can also help you infer, even if not confirm, which topics are shifting toward zero-click behavior.

The llms.txt Question, Settled

One tactic that gained traction through 2025 was publishing an llms.txt file — a proposed standard meant to guide AI crawlers toward preferred content. Google’s John Mueller has now confirmed directly that Google Search ignores llms.txt files entirely; they neither help nor hurt visibility or rankings. Separate research from Ahrefs backs this up from the other direction, finding that 97% of published llms.txt files receive zero requests from AI crawlers in practice.

That doesn’t mean the file is actively harmful to have, but it does mean it isn’t the lever some 2025-era guidance suggested it was. If a team is choosing between spending time on an llms.txt file and spending time on the content itself, the content wins, decisively.

What Actually Seems to Move the Needle

Google’s own webmaster guidance, reinforced in a July 2026 report from Search Engine Roundtable, keeps pointing at the same traits for content that continues to earn citations and clicks in an AI-saturated results page: real named authors with identifiable expertise, first-hand experience or evidence rather than secondhand summary, and specific, direct answers instead of generic or templated coverage. None of this is new advice, exactly, but it matters more now than it used to, because thin or interchangeable content is precisely what AI systems can already synthesize on their own without sending anyone to the source.

This is the space where Generative Engine Optimization (GEO) work is focused — structuring and strengthening content so it is more likely to be selected and cited by AI Overviews and similar systems in the first place, rather than reacting after the fact. It’s worth distinguishing that from the more traditional goal of ranking for clicks: GEO is about being the answer, while classic Answer Engine Optimization (AEO) work overlaps but keeps an eye on capturing the click when a click is still the outcome you want.

Getting Started

The practical move for most site owners is not to chase a new trick, but to treat this report as one more diagnostic in a fuller picture: check AI impressions alongside standard Performance data, watch for pages with visibility but no traffic lift, and prioritize the kind of authoritative, specific, clearly-authored content that both classic search and AI systems have consistently rewarded. If your team wants help interpreting what the new report is showing for your site, or building a content strategy suited to an increasingly zero-click landscape, Context Root’s GEO team can help translate the data into a plan.