Bing AI Performance Report: How to Read Citations and Topics

The Bing AI Performance report in Bing Webmaster Tools shows how often Microsoft's AI answers cite your pages. On October 13, 2026, Bing added a beta Topics Portfolio tab to it. The tab groups those citations into themes and lists queries where competitors get cited and you do not. Meanwhile, Google now gives every site a generative AI performance report in Search Console.

These reports answer different questions, so their numbers do not add up. This guide explains what each Bing metric means and how to read Topics Portfolio. It also compares Bing's data with Google's report and sets out a simple monthly routine. Every feature description comes from Microsoft and Google documentation as of October 14, 2026.

What the Bing AI Performance report measures

Microsoft launched AI Performance as a public preview on February 10, 2026. It covers Microsoft Copilot, AI-generated summaries in Bing and select partner integrations. In short, it counts citations: moments when an AI answer shows your page as a source.

• Total citations: how many times your content appeared as a source in AI answers during the selected period.
• Average cited pages: the average number of unique pages from your site cited per day.
• Grounding queries: the key phrases the AI used when it retrieved your content. Bing notes that this data is a sample.
• Page-level citation activity: citation counts for each URL, so you can see which pages get used.
• Visibility trends: a timeline of citation activity across the supported AI experiences.

However, a citation count tells you frequency, not importance. Microsoft says the numbers do not show placement in an answer, ranking or authority. Also, the report shows no clicks, so it cannot tell you whether a citation sent anyone to your site.

How Topics Portfolio groups your AI citations

Topics Portfolio sits next to the Citations & Pages tab inside AI Performance. It clusters your grounding queries into broad themes, such as taxes or running shoes. As a result, you can review your AI presence by subject instead of one query at a time.

The four columns in the topic table

The table lists each topic with three supporting numbers. Topic Size estimates total citations across all sites for that topic in the period. So it hints at how much AI answer demand the theme has. Sites Cited estimates how many sites appear for the topic's queries, which shows how crowded it is.

The fourth number, Queries, estimates the distinct grounding queries inside the topic. All three figures are approximate. In fact, Microsoft asks site owners to treat them as directional signals for planning, not as a precise ledger.

You can pick a one-, two- or three-month window. The tab appears only when Bing has enough data for your site. Therefore, a small or new site may see few topics, or none at all.

Grounding queries and opportunity queries

Open a topic to see three summary cards: Topic Size, your own citations in that topic and the number of queries. Below them, Your Grounding Queries lists the top queries where your site already appears. These match the data in Citations & Pages.

Next comes Opportunity Queries. These are queries in the topic where other sites get cited and yours does not. Bing shows their citation volumes as ranges, not exact counts. A portfolio-level list of top opportunity queries also sits under the topic table.

Read these lists with care. An AI model assigns queries to topics, and Microsoft admits the classification can be imperfect. Branded queries for a competitor's product can also show up as opportunities, even though you could never rank for them.

How Google's generative AI performance report differs

Google announced its generative AI performance reports on June 3, 2026. By August 31, 2026, they had reached all websites. Unlike Bing's report, Google's counts impressions: how many times links to your site appeared in AI Overviews or AI Mode.

The Search version groups data by page, country, date and device. In addition, a separate report covers generative AI features in Discover. Since September 24, 2026, a search type filter splits text queries from multimodal ones. Multimodal covers Google Lens, Circle to Search, image uploads and Chrome image search.

Google's report shows no query data and no clicks. Usual Search Console limits also apply, such as the 1,000-row table limit. The same impressions also stay inside the main Performance report under the Web search type, so do not add the two together.

Read Bing and Google data without mixing them

A Bing citation and a Google impression measure different events on different platforms. For that reason, never add them up or compare them as one share of voice. Instead, track each trend against its own baseline.

• Units: Bing counts citations in answers, while Google counts link impressions.
• Queries: Bing's grounding queries are retrieval phrases, so they may differ from what a person typed.
• Precision: Bing's query data is a sample and Topics figures are estimates, whereas Google aggregates by property or by page.
• Outcomes: neither report shows clicks, so add analytics data before you judge business value.

Our guide on how to measure AI search visibility explains how these reports fit into a wider measurement plan. Moreover, our review of repeated AI citation measurements shows why single readings mislead.

A monthly routine for the Bing AI Performance report

A short, repeatable review beats occasional deep dives. The steps below take about 30 minutes once your site has enough data. Keep the same window each month so the numbers stay comparable.

1. Note total citations and average cited pages for the last full month.
2. List your ten most cited pages and check that each one is current and accurate.
3. Open Topics Portfolio and pick the two topics closest to your business.
4. In each topic, review opportunity queries and drop branded or irrelevant ones.
5. For the remaining queries, decide whether an existing page should answer them better or whether you need a new page.
6. Record each change with its date, then compare the same topic three months later.

Treat any shift as a lead to investigate, not proof that a change worked. If traffic also falls, our guide to investigating AI Overview click loss walks through the checks in order.

Before you block AI features on either platform

Both search engines let you limit how AI answers use your pages. Bing follows robots.txt and other supported controls. Since October 2025, Bing has also supported the data-nosnippet attribute for hiding parts of a page.

Google now offers a Search generative AI control under Settings in Search Console. It can exclude your site from AI Overviews, AI Mode and Discover's AI features. Exclusion usually takes effect within one to two days. However, it does not stop AI training; for that, Google points to Google-Extended.

So measure first. Check what the reports above show before you opt out, because exclusion also removes the links and impressions you earn there. If you change crawler rules, test them before they go live.

Conclusion

The Bing AI Performance report counts citations across Copilot and Bing's AI answers. Topics Portfolio adds a themed view of those citations and hints at gaps. Google's report, by contrast, counts AI Overview and AI Mode impressions without query data.

Use each report for the question it answers, and keep their numbers apart. Start with one monthly review, act on two topics at a time and record every change. Then, after three months, you will have your own evidence instead of a guess.