Google Search Console AI Performance Report: What 351 Impressions and 6.9K Bing Citations Actually Mean

Direct answer: Google Search Console's Generative AI report is useful for measuring exposure, but it is not an AI ranking report. It shows how often a site's links appeared in AI Overviews and AI Mode, then breaks those impressions down by page, country, device, and date. It does not show the query, click, citation share, answer position, or conversion behind each impression.

As of August 31, 2026, Google says the Generative AI performance report has rolled out worldwide. The Search report currently covers AI Overviews and AI Mode; Search Labs experiments are excluded.

The dedicated report does not expose AI clicks separately. Google says generative AI activity remains included in the regular Web performance data, where it is not isolated as its own click metric.

FloxoLab's live three-month view recorded 351 Google AI impressions across 63 pages. Over the corresponding three-month preset, Bing Webmaster Tools displayed 6.9K citations and grouped grounding queries. Those totals should not be divided or compared as market share: Google counts link impressions on its supported AI surfaces, while Bing counts visible citations across Microsoft Copilot, Bing AI summaries, and selected partners.

The same “AI performance” label hides different measurement

Microsoft explicitly says a grounding query is a grouped phrase used to summarize retrieval activity—not the full prompt entered by one person. It also says citation volume does not indicate rank, authority, importance, or placement inside an answer. Google likewise reports an impression, not proof that someone clicked or even read the linked source.

What the Google report shows on a live site

The screenshots below were captured on September 6, 2026 from FloxoLab's verified property. The built-in three-month view visibly covers June 5 through September 4. This is first-party account data, not a fabricated dashboard or third-party estimate.

The Pages table is more actionable than the sitewide line. It contains 63 URLs, and the top page—FloxoLab's automation and optimization service page—recorded 117 page-level impressions. That is roughly one third of the 351 headline property impressions, but it should not be treated as a strict share because Google aggregates the property chart and page table differently. The next visible pages span n8n troubleshooting, Philippine SEO pricing, sitemap canonicalization, the homepage, a free-trial guide, a technical case, and a RAG guide. Google is not limiting AI visibility to one content format.

The leading page's own trend is uneven: 117 impressions appear as small daily clusters rather than a stable rank. That is a reason to monitor the page, not evidence that a particular edit caused visibility.

Country and device data add context, not attribution

The Philippines generated 168 of 351 impressions, about 48%. The United States contributed 44 and India 25. Together, those three countries produced roughly two thirds of the measured exposure. This supports a geographic observation: FloxoLab's AI visibility is concentrated in the Philippines but is not confined to it.

Device distribution is unusually clear: 308 desktop impressions, 42 mobile, and one tablet impression. Desktop therefore accounts for about 88% of this sample. That may reflect the types of research questions and users reaching FloxoLab, but the report cannot prove why the mix is desktop-heavy.

What Bing reveals that Google does not

Bing's three-month AI Performance view reported 6.9K total citations and an average of two cited FloxoLab pages per day. Its documentation defines a citation as content visibly referenced or shown as a source. The report is sampled and aggregated, so it is designed for trends and comparisons rather than exact accounting of every answer.

The most important result is concentration. Approximately 6.4K citations—roughly 93% of the rounded site total—belong to one page: FloxoLab's zero-click search measurement guide. Its leading grounding-query groups all describe tools or methods for measuring zero-click impact. Citation share across the visible leading phrases ranges from about 9.5% to 21.4%.

The best cross-platform check is one page, not two site totals

The n8n webhook troubleshooting article provides a cleaner comparison because the same URL appears in both systems. Google recorded 34 AI impressions for the page. Bing recorded 11 citations. The numbers still measure different events, but the overlap confirms that both platforms found the same problem-solving page useful enough to surface.

A read-only Search Console API check adds conventional Web-search context for June 6 through September 3, the latest fully finalized range available during this review. The webhook guide had 706 standard Web impressions, one click, and average position 10.2. The automation services page had 277 impressions, nine clicks, and average position 10.1. These figures should remain separate from the Generative AI report: they describe normal Web performance under Search Console's standard aggregation rules.

Four decisions you can make from the reports

1. Protect proven pages. When one URL carries most citations, check freshness, canonical consistency, internal links, and factual accuracy before redesigning or splitting it.
2. Investigate platform-specific winners. The zero-click guide dominates Bing citations but had only two conventional Google Web impressions in the finalized 90-day API view. That is not a contradiction; it shows that discovery and reuse differ by platform.
3. Use Google's dimensions for QA. Review leading pages, markets, and devices. A country or device mix can expose a content or rendering question, but it cannot explain causality.
4. Use Bing's grounding groups for topic diagnosis. Look for repeated reader tasks, then improve the existing page that owns that task. Do not create one article for every phrase.

After a substantive change, record the intervention date and compare a stable period against the previous one. Bing now provides a comparison view; Google provides date-based trends and exportable dimensions. A before-and-after pattern is still observational evidence, not proof that one edit caused the change.

What the data cannot tell you

• Neither report proves a fixed AI rank.
• A Google impression is not equivalent to a Bing citation.
• Bing grounding queries are grouped phrases, not complete user prompts.
• Neither total proves a click, lead, sale, or recommendation.
• Rounded and sampled totals should not be used for false precision.
• A trend change cannot be attributed to a content update without additional evidence.
• The reports do not provide a complete cross-platform view of ChatGPT, Perplexity, Gemini, or every partner surface.

For a broader measurement process that combines platform reports with repeated prompt observations and citation review, use the AI search visibility audit. For page-level eligibility and evidence structure, use the AI citation guide.

Method and source boundaries

The dashboard screenshots were captured from FloxoLab's Google Search Console and Bing Webmaster Tools accounts on September 6, 2026 using each interface's three-month preset. Google's visible chart runs from June 5 through September 4. Bing's screenshot shows the preset but does not print exact start and end dates, so this article does not claim day-for-day parity. Conventional Google Web metrics were queried separately through read-only Search Console access for the finalized June 6–September 3 range.

Platform definitions and availability were checked against official documentation on September 6, 2026:

• Google Search Console Help: Generative AI performance report
• Google Search Central: introduction and worldwide rollout
• Bing Webmaster Tools: AI Performance definitions and limitations
• Microsoft Bing: AI Performance public preview