Skip to content
AI Search Measurement

Google’s Generative AI Report Is Here. Read It Without Inventing a Story.

Google’s new report adds useful visibility data and a fresh opportunity to overclaim. Here is what its impressions and dimensions show—and what still belongs in analytics and the CRM.

Search analyst comparing generative AI visibility with source documents

Field note

By XenGrowth EditorialPublished Reviewed 9 min read

Key takeaways

  • Google announced dedicated Generative AI performance reports in June 2026, but the rollout initially covers only a subset of websites.
  • An impression shows that a URL appeared in a generative AI feature; it does not prove the page shaped the answer, earned a click, or influenced revenue.
  • Compare pages, countries, devices, and time periods, then annotate material publishing and technical changes before drawing a conclusion.
  • Use Search Console for Google Search visibility and analytics or CRM data for what happened after a visit; the numbers serve different jobs.

01

Put the rollout note above the chart

Google introduced dedicated Search Console views for generative AI features in Search and Discover on June 3, 2026. The first release is a test across a subset of websites, so absence of the report is not evidence that a site never appears in AI Overviews or AI Mode. Record whether the property actually has access before making the report part of a client promise or recurring dashboard.

The dedicated report sits alongside the broader Performance reporting. Treat it as a narrower lens on Google visibility, not a replacement for the rest of Search Console. A useful baseline records the property, report availability, date range, country, device, page group, and any major site changes during the period.

02

Know exactly what each dimension answers

The announced report includes impressions, pages, countries, devices for Search, and time views ranging from hourly to monthly. Those dimensions help answer where visibility appeared and whether it moved. They do not expose the generated prompt, the wording of the answer, citation absorption, or the complete path a person took after seeing it.

Start page-first. Group URLs by genuine topic, template, commercial role, and publication cohort. Then compare country and device patterns only when the sample is large enough to support the cut. A five-impression movement across six segments is not a trend; it is a reminder to wait.

Swipe to compare every column

Report signalUseful questionDo not claim
ImpressionsDid a URL appear more or less often in Google generative AI features?That the page was quoted, clicked, or preferred
PagesWhich URLs are receiving observable visibility?That every section of the page contributed
Country / deviceWhere and how did exposure occur?Why Google selected the source
TimeDid visibility change around a documented event?That the event caused the change

03

Use annotations to resist convenient causality

Mark releases, migrations, indexing incidents, canonical changes, major updates, PR coverage, and seasonal events. Compare equivalent periods and separate newly published URLs from mature ones. Generated results can vary by query, context, model, interface, and available sources, so a before-and-after line without controls is suggestive rather than causal.

Read representative pages as well as totals. A rising impression count can come from an irrelevant query family; a flat total can hide one strategically important guide replacing many low-value appearances. The editorial decision lives in that context, not in the direction of the aggregate arrow.

04

Join visibility to the journey without forcing attribution

Google recommends using Search Console as the source for Search performance and analytics for behavior on the site. Compare landing-page engagement, qualified inquiries, assisted journeys, and commercial outcomes where privacy and volume allow. Expect totals to differ because the systems measure different events and apply different processing.

A sound monthly note separates observation, interpretation, action, and confidence. For example: “Three implementation guides gained generative-AI impressions after their evidence sections were refreshed; we will review the visible query themes and downstream quality before deciding whether the edit contributed.” That sentence is less dramatic than a fabricated AI ranking win—and far more useful.

Primary sources and further reading

Use the source material to validate details against your own context and current platform configuration.

This field note follows the XenGrowth editorial policy: primary sources where available, visible limitations, material review dates, and no invented first-hand experience.

Stay with the problem

Explore AI search & GEO