AI Search

How to Read Google’s Generative AI Performance Report

Use page, country and device views to understand Search Console AI impressions—without double-counting them against overall Web search.

Abstract search results and an insight chart under a magnifying glass

Google’s Generative AI performance report shows impressions for links to your site in supported generative AI features in Google Search. Start with the trend, then examine the Page, Country and Device dimensions to see where those impressions are concentrated. Keep the AI figure separate from your overall Web search total: Google says the report’s data is already included in the Search results performance report’s Web search data. [1]

Google’s announcement describes the report as a dedicated view of visibility in features including AI Overviews and AI Mode. It does not establish that every impression in the report came from AI Mode, so label the metric “generative AI impressions,” rather than “AI Mode impressions.” [3]

Start with the trend and reporting settings

  1. Choose the Search report and a date range. Google provides separate generative AI reporting for Search and Discover; use the Search report for this analysis. [1]
  2. Read the chart total. By default, the chart aggregates impressions at the property level. The latest data may be preliminary and is shown with a dotted line; it can change. [1]
  3. Note the search type. The report offers Web: text and Web: multimodal filters. Keep the setting consistent when comparing periods. [1]
  4. Compare like with like. Review an earlier period of the same length, using the same property and settings. Record the dates and whether the latest data is preliminary.

An impression is not a visit or a click. Counting rules depend on the result type: in some cases a link may need to be scrolled into view or expanded before an impression is recorded. [2]

Analyze impressions by page

Select the Page tab and review the pages with the most and fewest reported impressions. Compare the list with the pages your content team considers important. A page with relatively few impressions may be worth investigating, but the report alone does not explain why it appeared less often.

Check which URL receives the data before assigning work. Google groups pages by the final landing URL after redirects, and most performance data is attributed to the canonical URL rather than a duplicate. A page-level view is aggregated differently from the property-level chart, so the table and chart totals can differ. [1]

Useful next step: Export the page table, note the reporting period, and flag important pages for a content review. Treat the list as a starting point, not proof that a particular edit will increase AI visibility.

Analyze impressions by country

Select Country to group impressions by the country or region where searches originated. Compare the leading countries with the markets your team serves, then review the same dimension for a comparable earlier period. [1]

Useful next step: For a market that matters to your team, record whether its reported impressions are rising, falling or stable. Use that observation to prioritize a review of relevant content; do not assume the country table explains the cause of a change.

Analyze impressions by device

Select Device to see impressions grouped by the device used to search: desktop, tablet or mobile. Compare the distribution across periods using the same date length and search-type setting. [1]

Useful next step: If one device accounts for a substantial share of reported impressions, include that device in your review of important pages. Device impressions describe visibility in search, not the quality of the subsequent on-site experience.

Page, globe and mobile device symbols representing three analysis dimensions
Review generative AI impressions by page, country and device.

Why AI and overall Web search totals must stay separate

The generative AI report is a dedicated view of data already included in overall Web search reporting. Adding its impressions to the overall Web search total would count overlapping data twice. [1] [3]

Illustrative example: If a dashboard shows 1,000 overall Web search impressions and 200 generative AI impressions for a period, report the two figures separately. Do not add them and claim 1,200 total impressions.

Before comparing the figures, check the property, dates, search type and aggregation. In particular, page-level rows and a property-level chart do not necessarily use the same counting basis. Google notes that when two results from the same site appear in a generative AI result feature, the property-level chart can count one impression, while the page table is aggregated by page. [1]

Why the report may be missing

Google says the report is usually absent when a site has not received enough impressions in Google Search’s generative AI features. Google also notes that a site may be excluded from those features; check whether the site is eligible if exclusion is a concern. A missing report should not be treated as proof of zero impressions. [1]

Availability should be verified before publication. Google’s initial announcement described a release to a limited set of sites, while the supplied help excerpt states that the insights were rolled out globally to all sites by August 31, 2026. Check the current documentation and the property itself rather than assuming either statement describes today’s availability. [3] [1]

Checks before sharing a report

  • Label the source and period: Identify the Search Console property, date range, search type and dimension. Search report dates use Pacific Time. [1]
  • Do not force totals to match: The chart and table can differ because of aggregation. The report also inherits regular performance-report limits, including a 1,000-row table limit. [1]
  • Review exports carefully: Values shown as “~” or “-” in the report appear as zero in downloads. Do not automatically interpret every exported zero as a measured absence. [1]
  • Keep scope clear: Search Labs experiments are excluded, and Discover has separate generative AI reporting. [1]

Use the resulting analysis to decide what to investigate next: which pages deserve a review, which markets warrant attention and which devices to include in testing. Report the observations separately from any explanation or proposed action.

Sources

  1. 生成式 AI 效果报告(搜索)Accessed
  2. 什么是展示次数、排名和点击次数?Accessed
  3. Search Console に検索の生成 AI のパフォーマンス レポートを導入 | Google Search Central Blog | Google for DevelopersAccessed

Discussion

Join the discussion

Start the conversation

Share a question or tell us what worked for you.

Comments appear after review. How we handle comments

The Ansvra journal

Keep reading

Browse all articles

Ansvra

A clearer next step.

MEET YOUR AI GUIDE

A little clarity.
A clearer next move.

Ask about Ansvra, AI visibility, or your next step with content.

Opening chat…