Google’s AI Impression Data Tells You Where to Look

TL;DR: Google separated AI Overviews and AI Mode impression data into its own Search Console report on June 3, 2026. The report shows which URLs appear inside generative search features, segmented by page, country, device, and date — but no clicks, no CTR, no queries. This is a diagnostic lens, not a performance scorecard, and operators in high-CAC verticals need to treat it that way from day one.

What Google Actually Released

Google’s new Generative AI performance report inside Search Console tracks impressions from two features: AI Overviews and AI Mode. Experimental Search Labs features are excluded. A separate report covers generative features in Discover. The rollout is currently limited to a subset of site owners while Google collects feedback, so not every account will see it yet.

What the report includes: impressions, URLs, country, device, and date. What it does not include: queries, clicks, click-through rate, average position, citation placement, which passage was used, or any conversion data. Google has signaled that additional metrics may come later, but nothing is confirmed.

Google is also testing a Search Console toggle that lets site owners opt their content out of AI Overviews, AI Mode, and generative Discover features entirely — without affecting traditional organic listings. The default is opted in. Before anyone reaches for that switch, the new report gives you a baseline to consult first. Opting out forfeits any traffic those features might eventually send. That is a real cost in verticals where every acquisition channel matters. If you want a structured view of how your current content mix is performing before making that call, a full channel audit is the right starting point.

How AI Impressions Are Counted (And Why the Math Looks Wrong)

Google defines an AI impression as an instance where a link to your site is shown to a user inside a generative feature. The counting logic shifts depending on the aggregation level.

At the property level (the chart view), if two URLs from the same site appear in a single generative response, they count as one impression. At the page level (the table view), each URL may receive its own impression count. This means summing page-level impressions will rarely match the property-level total. The math is not broken — the dimensions are different.

AI impressions and organic impressions are also not the same unit. In conventional search, an impression means a discrete listing was presented. In AI Overviews, a link must be scrolled or expanded into view before it registers. In AI Mode, a follow-up question triggers a new query, and links shown in the follow-up response generate additional impressions. Blending these into a single “total search visibility” number produces a figure that means nothing actionable. Do not calculate a blended CTR either. The number is computable; it is not meaningful.

The Four Patterns Worth Investigating

The report’s real value is comparison, not totals. Cross-referencing AI visibility against conventional organic performance reveals four diagnostic patterns.

High organic, low AI visibility. A page ranks and receives solid organic impressions but rarely appears in the generative report. This does not automatically indicate a problem — not every query triggers an AI response. But it justifies a content review: Does the page answer questions directly? Can individual passages stand on their own without surrounding context? Is the key information buried in JavaScript, tabs, or images that Google cannot parse? Ranking gets a page into consideration; it does not guarantee Google can extract a clean answer from it.

Low organic, high AI visibility. A page with modest conventional rankings receives a disproportionate share of generative impressions because it contains a precise definition, a useful statistic, or a clear comparison. These pages are worth studying first. They show where Google finds content easy to use. Look for recurring structural characteristics: direct answers near the top of sections, strong heading structure, original data, focused topical scope, and language that matches how users phrase questions. One successful URL is a clue, not a playbook.

Impressions concentrated in a few URLs. If most generative impressions come from a handful of pages, group those pages by topic, content type, and template. One template may expose content cleanly in HTML while another hides it behind scripts or decorative formatting. The report will not diagnose the issue — it will tell you where to start looking.

Visibility shifts after content revisions. The report can track whether meaningful revisions correspond with sustained changes in AI impressions. Meaningful revisions include adding a summary or definition, updating outdated data, consolidating overlapping pages, improving headings, and making important content visible in HTML. A sustained increase following a substantial revision is useful evidence. A three-day spike after changing one heading proves nothing — demand, seasonality, and Google’s own systems all shift simultaneously.

What This Means for High-CAC Vertical Operators

Operators in forex, iGaming, crypto, and legal run markets where a single converted lead can be worth hundreds to thousands of dollars. AI search impressions without click data may feel like a half-built instrument, but the diagnostic value is real if you apply it correctly.

For forex acquisition teams, the report surfaces which regulatory explainers, broker comparison pages, or spread analysis content Google is pulling into AI Overviews. If those pages are getting AI impressions but your analytics show no referral uptick, the content is being cited but not generating navigation — a content structure problem, not a traffic problem.

For operators running iGaming campaigns, generative search is already reshaping how users research bonus structures and game mechanics. Knowing which pages appear in AI Mode gives your content team a list of formats and structures to replicate across thin or underperforming pages.

Legal operators handling mass tort or personal injury intake should check which intake-adjacent content — symptom explainers, eligibility criteria pages, statute-of-limitations guides — is getting generative exposure. Those pages directly influence pre-qualified intent. If you run law firm lead generation, this data tells you where Google is already positioning your authority; the question is whether the page then converts the reader into a form submission.

Crypto operators managing content for exchanges or token projects will find the pattern analysis especially useful. Crypto lead generation depends heavily on trust signals and explainer content. The AI impression report can identify which technical explainers or risk-disclosure pages are being surfaced, helping content teams prioritize updates that extend that visibility across similar topics.

Across all these verticals, the report is most useful when combined with audience-level targeting data and your analytics referral breakdown. AI impressions without conversion tracking attached are an input, not an output. Build the measurement stack before drawing conclusions about ROI.

A Practical Six-Step Workflow

You do not need a new tool stack. You need exports, consistent date ranges, and discipline about what the data can and cannot claim.

Step 1: Export AI-visible URLs. Use a date range of at least three to four weeks. Rollout data from the first few days will be noisy.

Step 2: Add conventional search data. For the same URLs and dates, pull organic impressions, clicks, CTR, average position, and top queries from the standard Performance report. The purpose is not to treat the two impression types as equivalent — it is to identify pages where AI visibility diverges sharply from conventional performance.

Step 3: Categorize the pages. Add attributes — page type, topic, intent, template, author, publication date, last revision, funnel stage. A list of URLs and impression counts is inventory, not analysis.

Step 4: Investigate outliers. Look at visible HTML structure, answer placement, internal link architecture, and source quality for pages that overperform or underperform in generative search. Someone has to actually read the pages. That part does not automate.

Step 5: Add analytics separately. Use your analytics platform to assess identifiable AI referral traffic and conversions. Search Console and analytics have never reconciled perfectly on organic data — AI search will not change that.

Step 6: Track trends, not daily fluctuations. Multi-week or monthly comparisons are more reliable than daily movement. If your dashboard is alerting on Tuesday dips in AI impressions, the dashboard is the problem. For operators who need structured oversight across multiple channels simultaneously, managed performance reporting keeps the signal-to-noise ratio functional.

What Not to Put in Your Executive Dashboard

The moment a large number labeled “AI” appears in Search Console, someone will want it at the top of the dashboard in green. Resist that. Avoid reporting a blended total of generative and traditional impressions, a blended CTR, revenue attributed to impressions without supporting conversion data, or a visibility increase labeled as proof that a single optimization worked.

A responsible dashboard can include generative impressions over time, the number of pages receiving impressions, topics and page types represented, the relationship between AI-visible pages and organic performance, meaningful revisions made during the period, and identifiable AI referral traffic reported separately from impression counts.

The useful question is not whether the number went up. It is: what does this new visibility reveal about how Google is choosing to use the content? That question drives decisions. The number alone does not.

Originally reported by Search Engine Journal, August 2026.