Bing Webmaster: Measure and Optimize Your AI Visibility

What did Bing just add to AI Performance?

I review Webmaster Tools reports for my clients every week. The AI Performance tab, though, almost always stays closed.

That’s a pity. Since October 13, 2026, there’s a real reason to open it.

According to Search Engine Journal (article by Matt G. Southern), Microsoft is launching Topics Portfolio in beta. It’s a new tab, placed to the right of « Citations & Pages » in the AI Performance section. The principle fits in two sentences. Bing groups the queries for which your site is cited in its AI answers into broad topics. Then, within those topics, it lists the queries where other sites appear and yours doesn’t.

These queries have a name: grounding queries. They are the search terms Bing’s AI uses to locate content, then cite it. They feed Copilot, Bing’s AI summary and some partner integrations.

The timeline helps you see what’s new:

• February: AI Performance enters public preview.
• June: Topics, Intents, Citation Share and Compare arrive. Topic labels then grouped only the queries where your site was already cited.
• October 13, 2026: Topics Portfolio adds the total citations across all sites at topic level, plus the list of queries where you are not cited.

Microsoft admits it: the report was missing two pieces of data. The ecosystem context. And the queries where you aren’t cited.

The change is easy to sum up. The report no longer tells you only where you are. It tells you where you’re absent.

For an SEO engineer, the report moves from thermometer to map. Before, it diagnosed (you’re cited). Now it navigates (you’re missing here).

One useful clarification before going further. The view appears only if Bing has enough data on your site. Topics with too little presence or ecosystem activity are hidden. A small site may therefore see few items, or none. If that’s you, it’s not a verdict on your content. It’s a data threshold, and you’ll cross it by building presence topic by topic. We’ll come back to that.

Bing n’a pas lancé Topics Portfolio du jour au lendemain. Trois étapes clés ont structuré le déploiement de l’AI Performance report. Ce timeline explique pourquoi agir maintenant, alors que la majorité des concurrents regardent encore ailleurs.

How do you read Topic Size, Sites Cited and Queries?

Each topic in the portfolio rests on three metrics. Each one asks a different question.

Taken alone, each of these three numbers offers little. Together, they form a complete picture.

Here’s how I read these three variables. It’s not a rule published by Microsoft, just what I observe. High Topic Size + few sites = a concentrated market where a handful of players dominate. You need precision to get in. High Topic Size + many sites = an open market, with room for good content. Many queries + few citations = a long tail to cover page by page.

The report offers three windows: 1, 2 or 3 months. The choice changes everything.

• 1 month: measure the effect of a recent publication.
• 2 months: smooth out peaks and troughs.
• 3 months: validate an editorial strategy. This is where I decide budgets.

Citation Share (launched in June) calculates your share of all citations on a query. It doesn’t reveal who your competitors are. It shows your own presence while keeping other sites’ data private.

Topics Portfolio works the same way. You know other sites are cited, but which ones stays hidden. It’s a mirror of your blind spots, not a competitive spying tool. To find out who dominates, open the AI answer itself and check.

What should you do with Top Opportunity Queries, concretely?

Below your topics table, Bing displays the Top Opportunity Queries list. Across all your topics, it gathers the queries where other sites are cited and yours isn’t. Citation volumes appear there as ranges.

One click on a topic opens its detail page. There you find the topic size, your citations and the number of queries. Two tables face each other: the queries where you’re already cited, and the opportunity queries.

Two tables. One for the before, one for the possible after.

I work with it inside a semantic cocoon. The cocoon is my way of structuring a site: one pillar page, satellite pages that answer each intent, and internal linking that ties it all together. Opportunity queries fall into four cases.

1. The page exists, but it doesn’t answer the question. The topic is covered, the angle is missing. I rewrite the page opening to give the answer in the first lines, then expand.
2. The page exists, but it’s isolated. The pillar doesn’t link to it. I attach it to the cocoon with anchors that reuse the query’s vocabulary.
3. The page doesn’t exist. It’s a new satellite to create. It joins the editorial plan, ranked by topic size.
4. The query has nothing to do with your offer. I set it aside. Not everything deserves a page.

This triage takes half a day per topic. It sidesteps the classic temptation: publishing everything because the list is long.

I see a recurring pattern with my clients. They have a lot of content and little structure. Their pages answer questions nobody asks the AI, while the questions the AI really asks, through its grounding queries, remain unanswered. The issue isn’t content volume. It’s the alignment between your pages and the queries the engine uses to search.

Topics Portfolio makes this gap visible, topic by topic. For an e-commerce site, the reading is immediate. A « trail running shoes » topic with ten opportunity queries points to a category to enrich: buying guide, comparisons, FAQs, better-described product pages. For a services site, it’s a series of FAQ pages and client case pages to write.

A question to ask yourself right now: how many of your pillar pages have a satellite for each of the queries slipping away from them?

What are the limits of this beta, and why do they help you use it better?

Microsoft is transparent about the limits. According to Search Engine Journal, the topics in Topics Portfolio are AI-generated. A query can therefore land under a different label than the one you would have chosen. Your business classification and Bing’s don’t always overlap.

Second limit: brand queries can show up as opportunities. Microsoft illustrates this well. A shoe site sees a competitor’s brand search surface in the « Running Shoes » topic as an opportunity.

Should you answer it? Rarely. A query containing a competitor’s name signals the intent to find that competitor. Creating a page to capture it yields, at best, an honest comparison piece, at worst, noise. So I filter the list in three passes:

• I remove queries containing a third-party brand, unless I already have a documented comparison angle;
• I remove queries outside my catalog or offer;
• I keep informational or comparison-intent queries that match content I can back up with facts.

Third limit: the ranges. You won’t know whether a query is worth 120 or 140 citations, only which bracket it sits in. That’s enough to prioritize. It’s insufficient to promise a client a numeric result. That’s why I never make that promise.

Last limit, already mentioned: small sites will see fewer items, or nothing. If that’s you, use the rest of AI Performance. The Citations & Pages tab shows you where you’re already cited. Build from those pages, extend the cocoon, and check back every month to see if Topics Portfolio opens up.

The beta is your chance to test before everyone else does.

Why look at Bing now, when everyone talks about Google and ChatGPT?

Because Bing is currently the only major player to show you your citations in its AI answers this precisely, directly in a free tool for webmasters.

Bing’s grounding queries feed Copilot, Bing’s AI summary and partner integrations, according to Search Engine Journal’s description. Bing is a visibility surface to take seriously, with its own citation rules.

My opinion, and I present it as such: the time to position yourself is now. Most SEO teams I work alongside put their energy into Google and into visibility in ChatGPT. The competitors who watch Bing’s AI Performance report are still few. When a feature is new, the first people to use it build their presence before the field fills up. That holds for every new search surface, and I have no reason to believe it will be different here.

There’s a second benefit, less visible. Grounding queries are a window into how a search AI rephrases users’ questions. They reveal the vocabulary, the angles, the sub-questions the engine will look for. My working hypothesis: this vocabulary is useful well beyond Bing. Content that answers well-formulated sub-questions precisely has a good chance of being clearer for any search system, human or AI. It’s a hypothesis. I’m testing it with my clients, and I don’t present it as established fact.

My recommendations to get started:

• Connect or verify your site in Bing Webmaster Tools if you haven’t yet.
• Open AI Performance, then the Topics Portfolio tab. If nothing displays, look at Citations & Pages.
• Use a three-month window to decide, a one-month window to track.
• Record your top three topics and their metrics in a spreadsheet, on a fixed date. In three months, you’ll have your first curve.

That’s all. The first step works with free tools only.

What interests me most about Bing is the clarity. An engine that shows you its queries is a rare opportunity. Make the most of it.

What protocol should you apply starting this week?

Here’s the sequence I suggest, in four stages. It rests on a discipline of reading and acting, with zero promises of results.

Week 1: map. Export or copy your topics, with Topic Size, Sites Cited and Queries. Sort them into three groups: core business, adjacent, off-topic. Off-topic leaves the plan.

Week 2: filter. Run the Top Opportunity Queries list through the sieve described above. Remove third-party brands and queries outside your offer. Attach each remaining query to an existing page or to a page to create.

Week 3: produce. Start with rewrites and internal links. These are the fastest gains, because the page already exists. Then move on to new satellites, starting with the topic with the highest Topic Size.

Week 4: verify. Rerun the view on a one-month window. Check whether your citations move on the topics you worked on. Keep in mind an engine’s reaction time: one week of hindsight is too short to conclude. Ask the question again at three months.

This protocol belongs to a broader logic, the DOSE framework that Guillaume Attias teaches at BMO Academy. The core idea: a visibility system works in layers, each one tested before the next. Topics Portfolio becomes your AI diagnostic layer here. It builds on your cocoon, your internal linking and your editorial plan. It replaces nothing. It guides.

One last field tip. Document what you do. Note the date, the page changed, the topic concerned. In three months, when you return to the long window, you’ll know which action produced which movement. Without this trail, the data turns into noise.

I’m not selling you the method. I show you the pages: in an audit, I open your AI Performance report live, we read your topics together and we set the first three actions.

And you: what are the three topics where Bing already cites your competitors, but not you?

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn