Bing Topics Portfolio: how to find your AI citation gaps

The answer in brief

Topics Portfolio is a beta view inside the AI Performance report in Bing Webmaster Tools. It groups the grounding queries behind Copilot and Bing’s AI answers into topics where your site already appears, then adds two things Bing has not shown before: how much citation activity each topic gets across all sites, and a list of queries where other sites are cited and yours is not.

That second list is the useful part. It turns AI Performance from a report on what you already have into a short list of pages worth writing or fixing. Microsoft itself says to treat the figures as “directional signals for planning, not as a precise ledger”, so use them to set priorities, not to report progress to the board.

What is Bing Topics Portfolio?

Microsoft announced it on the Bing Webmaster blog on 13 October 2026. It sits inside AI Performance, the report Bing opened in public preview in February 2026, which shows how often your pages are cited in Microsoft Copilot, AI summaries in Bing and some partner services.

The report works from grounding queries. These are the phrases Microsoft’s AI uses to retrieve the content it cites in an answer. Topics Portfolio clusters those queries into themes. Microsoft’s examples are football, taxes and AI chatbots.

Three details matter before you go looking for it:

• It only appears “if there is sufficient data for your site”. Smaller sites may see a short list or nothing at all.
• The list is built from your own presence, “not from a generic ecosystem list”. It shows topics where you already have a meaningful footprint, not every topic you could enter.
• Microsoft’s post calls it the Topics Portfolio tab next to Citations & Pages in one paragraph and the Topics tab in another. If you cannot see the first name, look for the second.

What does each column tell you?

The landing page shows one row per topic. Microsoft describes every figure as approximate.

The landing page also carries a list called Top Opportunity Queries. These are the strongest queries across your topics where other sites are cited and yours is not, with citation volumes shown as ranges rather than exact counts.

Click a topic and you get a detail page with three cards (Topic Size, your citations in that topic and the number of queries), then two tables: Your Grounding Queries and Opportunity Queries. Microsoft calls these “the two halves of a topic: what you already own, and what is still open.”

You can choose a one, two or three month window.

How is this different from the Topics view Bing added in June?

In June 2026 Bing added Intents, Topics, Citation Share and Compare to AI Performance. That version of Topics grouped the queries you were already cited for. It told you what you had, and nothing about what you were missing.

Citation Share, added at the same time, gives your share of all citations for a grounding query. Microsoft called it “an observational metric” and said it does not reveal which other domains are cited.

Topics Portfolio goes further in two ways. It puts all-site totals next to each topic, so you can see whether a theme is big or small and busy or quiet. And it lists the queries where you are absent. Until now, AI Performance only ever showed queries where you had already been cited, so the gap was invisible. We cover how the June features fit into a wider AI search plan in our guide to GEO for business.

How should a UK team use it this week?

Microsoft’s own advice is to start with the topics where your citations are already concentrated, then work down the opportunity queries. We agree, and would add a filter at each step.

1. Pick the three month window and write down your top five topics. The longest window smooths out a single busy week. Check that the labels match how you think about your own content. If a topic looks wrong, use the thumbs down control on that row; Microsoft says those signals go straight to the team fixing the classification.
2. Compare your topics by demand and crowding. Plotting Topic Size against Sites Cited gives a rough grid. Our rule of thumb is that a big topic with few sites is where a strong page has the best chance, and a small, crowded topic is rarely worth a new page.
3. Copy the opportunity queries into your content backlog. The October announcement does not say whether this view exports. The wider AI Performance report does export grounding queries, pages and trends to CSV and Excel, according to the AI Performance help page, so keep those exports for your baseline.
4. Strike out queries that name another brand. Microsoft warns that branded queries can show up as opportunities. Its example is a running shoe site being shown another shoe brand’s query. Unless you sell or review that brand, a rival’s product name is not a gap you can close.
5. Match each remaining query to a page, or brief a new one. For each query, ask whether you have a page that answers it plainly in the first two sentences. Microsoft’s guidance on content for AI search answers says Copilot breaks pages into smaller pieces and assembles answers from them, so clear headings, direct answers and tables help. Recheck the topic in four to six weeks with the same window.

What should you not read into the numbers?

Microsoft is open about the limits, and they are worth repeating to anyone who sees a screenshot.

The topics are drawn by AI, so a query can land in a theme you would label differently. Topics with too little data are hidden. Topic Size, Sites Cited and Queries are approximate, and opportunity volumes are ranges on purpose.

The help page adds two points that apply to everything in AI Performance. A citation is not a click or a visit, and the data is sampled, so totals can differ between views. In Microsoft’s words, AI Performance “does not measure rankings, authority, performance, or importance.”

Two more of our own. The figures cover Microsoft Copilot, AI summaries in Bing and partner services that Microsoft does not name. They say nothing about Gemini or Google’s AI Overviews. OpenAI’s help page on ChatGPT search lists Microsoft’s privacy statement among the search providers’ policies that apply, which is why our ChatGPT SEO guide suggests checking Bing indexing, but a Bing citation count is not a ChatGPT one. For visits from AI assistants, use your analytics; our GA4 audit for ChatGPT traffic covers how to check those figures. And a citation is not the same as being named in the answer. Our Semrush Spotlight 2026 write-up shows how often brands are named in AI answers without being cited.

Does Google Search Console show anything like this?

Not at query level. Google’s generative AI performance report, rolled out to all sites by 31 August 2026, shows impressions in AI Overviews and AI Mode by page, country, device and date. It lists no query view and nothing on other sites.

That makes Bing the only one of the two that currently shows which AI search questions other sites are cited for and you are not. Keep the two reports in separate columns. They count different things on different platforms, and adding them together produces a number nobody can explain. If your Google rankings are strong but AI answers ignore you, our piece on why ranking on Google does not guarantee AI citations is the place to start.

Where should you begin?

Open the report, note whether the view exists for your site at all, and record your top five topics today. That record is your baseline. If the view does not appear, the likeliest reason is too little AI citation activity, which is itself a finding.

If you would like help turning an opportunity list into a content plan, see our answer engine optimisation service or talk to AIWIZ.