Bing's AI Performance report is the clearest first-party way to see where your site is already being cited in Bing's AI answers and Copilot, and where competitors are being cited instead. Use it for two jobs: first, to find which pages Bing already trusts enough to cite, and second, to mine the grounding queries, which are the real prompts and search intents behind those answers. Start with total citations, cited pages and query-to-page mapping. Then review grounding queries every month, group them by topic, compare them with the pages Bing chose, and fill the gaps with better answer content, honest comparison pages, pricing pages, review signals and stronger distribution. Since Bing Webmaster Tools added Topics Portfolio in beta on 13 October 2026, it is also the only first-party place to see AI citation gaps where competitors appear and you do not.
Bing's AI Performance report is a measurement tool, not an AI ranking report.
The first thing to know is straightforward: there is no such thing as an AI ranking position. Bing's report doesn't give you a rank. It gives you evidence of where your pages were used in AI answers.
Microsoft first put the Bing Webmaster Tools AI Performance report into public preview on 1 February 2026. It included total citations, average cited pages, grounding queries and page-level citations. Microsoft then added query-to-page mapping on 23 March 2026. Those dates matter because they tell you what the report can now do, and what older advice may miss.
If you are reporting to a board, a founder or a commercial team, treat this as a sample of visibility. Don't treat it as complete traffic data. Microsoft presents it as citation reporting, not as a clicks report.
If you need the wider measurement picture, pair this with your analytics and with Google's own reporting. We explain the broader setup in How to measure your AI search citation performance and the Google side in What Google Search Console's AI report shows, and what it doesn't.
Grounding queries are the most useful part of the report for content planning.
Grounding queries are the underlying searches or prompts Bing used to build an AI answer. In plain English, they show the real questions behind the answer, not just the page that got cited.
This matters because many businesses look only at the cited page and miss the intent. A pricing page may be cited for a comparison query. A product page may be cited for a trust query. A help page may be cited for a buying decision query.
That is why grounding queries are more valuable than a simple mention count. They tell you what the buyer was trying to decide.
What you can learn from grounding queries
You can learn four useful things very quickly.
- Which topics Bing already associates with your business.
- Which page types win citations, such as pricing, comparisons, product pages or guides.
- Which commercial questions you are missing, such as alternatives, costs, reviews or suitability.
- Which competitors are present in the same topic space.
Why this matters more after October 2026
On 13 October 2026, Bing Webmaster Tools added Topics Portfolio in beta. Microsoft describes it as the grounding queries where competitors are cited and you are not. That makes it the first first-party AI citation gap report.
For a marketing leader, that changes the job. You are no longer guessing which AI answer spaces you are missing. You can now review a concrete list of topic gaps each month.
The report is most useful when you read it in a fixed order.
If you jump straight into individual queries, you will lose the commercial picture. Start broad, then narrow down.
1. Start with total citations
Total citations tell you whether Bing is using your site at all. A low number does not always mean failure. It may mean Bing prefers a small number of strong pages. Still, if the number is flat for months, you likely have a trust, coverage or access problem.
Don't compare this number directly with web sessions. It is a visibility signal.
2. Check average cited pages
This shows how many different pages are earning citations. If one or two pages carry everything, your visibility is fragile. If a wider set of pages appears, your authority is spreading across topics.
A healthy pattern is breadth across your key commercial themes, not random blog coverage.
3. Review page-level citations
This tells you which pages are doing the work. In many accounts, the winners are not the pages teams expect. Honest comparison pages, pricing pages, category pages, product detail pages and review-rich pages often matter more than generic homepage copy.
This fits what HelloWorld sees in practice. Roundup placements, YouTube explainers, honest comparison pages and Google Business Profile are high-weight levers. Cheap technical tidy-up work still matters as hygiene, but it is not the main growth lever.
4. Use query-to-page mapping
Query-to-page mapping shows which page Bing matched to which grounding query. This is where you find mismatches.
For example, if Bing keeps citing your homepage for a detailed product comparison query, you probably need a proper comparison page. If it cites an old blog post for a pricing query, you may need a live pricing page with clear terms.
5. Mine grounding queries by topic
Now group queries into themes. Typical themes include price, alternatives, reviews, best-for, local availability, compliance, setup time and support.
This step turns the report into an editorial plan. Without it, you just have a spreadsheet.
The best way to use grounding queries is to turn them into page decisions.
Don't answer every query with a new article. First decide which page type should own the topic.
Use pricing queries to improve or create pricing pages
If grounding queries contain words like cost, price, fees or how much, your pricing page is either missing, weak or unclear. Buyers ask these questions before they shortlist suppliers.
For many established firms, a better pricing page is more valuable than another top-of-funnel blog post.
Use alternatives and comparison queries to build honest comparison pages
If queries mention alternatives, versus or competitors, create comparison pages that explain fit, trade-offs and who each option suits. Keep them fair. AI assistants are more likely to trust pages that acknowledge limits and context.
This is especially important for B2B software and services. If that is your market, see What is Answer Engine Optimisation for B2B software companies? and How B2B buyers use ChatGPT to build a vendor shortlist.
Use review and trust queries to strengthen proof
If queries mention reviews, ratings, trusted, best or recommended, your issue may be proof rather than copy. Improve named authors, customer evidence, review profiles, comparison coverage and third-party mentions.
For software firms, review platforms can matter. We cover that in Do G2 and Capterra reviews affect what ChatGPT recommends?.
Use local intent queries to strengthen local signals
If you serve a place-based market, grounding queries often include towns, counties or near-me intent. That tells you to improve local landing pages, local proof and your business profiles.
For local businesses, your Google Business Profile, Apple Business Connect and Bing Places can support discoverability. They are not magic on their own, but they are practical trust signals.
Topics Portfolio is your monthly gap report against competitors.
Topics Portfolio, added in beta on 13 October 2026, shows grounding query areas where competitors are cited and you are absent. That makes it ideal for monthly review.
Use it like this: export the missing topics, sort them by commercial value, then decide whether the fix is content, proof, distribution or page type.
How to prioritise topic gaps
Not every missing topic deserves action. Prioritise in this order.
- High buying intent topics, such as pricing, alternatives, reviews and product fit.
- Topics tied to your highest-margin services or products.
- Topics where competitors are repeatedly cited.
- Topics where you already have a page, but Bing is choosing someone else.
What not to do with topic gaps
Don't create thin pages for every phrase. Don't rewrite copy “for AI”. Google says this is unnecessary in its generative AI optimisation guidance. Google also says its systems understand multi-topic pages, so forced content chunking is not required.
Instead, build pages that answer the commercial question clearly, with visible question-and-answer text, named authors, honest dates and sourced figures.
Access and coverage problems can hide good content from Bing's report.
If a strong page is missing from AI citations, the problem may be access, not quality.
Check whether the page is in Bing's phone book, which is the index that Copilot and Bing's AI summaries read. Also check your AI welcome sign, robots.txt, which is the small file that tells AI readers whether they are invited in. Then check whether your site's table of contents, called a sitemap, includes the page.
Keep the explanation simple inside your team: if the page is hard to find or blocked, Bing can't cite it reliably.
Why this matters even more for other AI assistants
Don't assume Bing visibility equals ChatGPT visibility. Since 4 September 2026, ChatGPT search has run on OpenAI's own index, called Labrador, with result scraping as a fallback and Bing used mainly in Deep Research. That means a page can be visible in Bing's world and still weak in ChatGPT's world.
It also means IndexNow is a Bing and Copilot lever, not a ChatGPT one. Keep it because it is cheap and checkable. Don't sell it to your team as the main growth lever.
The report works best when you combine it with content, proof and distribution.
Bing's AI Performance report tells you where the gaps are. It does not fix them. The fixes usually sit in one of four buckets.
Answer content
Add visible answers near the top of key pages. Use plain language. Include sourced figures with named sources and dates. Add named authors where trust matters. If a page answers a commercial question, make that answer easy to quote.
If you are unsure about trust signals, see E-E-A-T and named authors: building trust AI can verify.
Page type coverage
Make sure you have the page types buyers expect: pricing, alternatives, comparisons, reviews, category pages, location pages and strong product or service pages. Many firms have plenty of blog content and very weak buying pages.
Third-party proof
AI assistants often cite pages beyond your own site. That is why honest comparison pages, review platforms, roundups and business profiles matter. Buying mentions is not the answer. Google's guidance says directly that seeking inauthentic mentions is not as helpful as it might seem.
Basic hygiene
Your machine-readable business card, called schema markup, your sameAs links and your one-line summary in search are still worth keeping tidy because they are cheap and checkable. They are hygiene. They are not the main lever. If you want the detail, read What is schema markup, and does it matter for AI search?.
A simple monthly workflow will get more value from the report than ad hoc checks.
The teams that benefit most from this report review it on a schedule. Once a month is enough for most established businesses.
- Export total citations, cited pages and grounding queries from Bing Webmaster Tools once a month.
- Group grounding queries into commercial themes such as pricing, alternatives, reviews, fit, setup and support.
- Mark which themes already have a strong page on your site, and which do not.
- Review query-to-page mapping to spot mismatches, such as a homepage being cited for a detailed buying query.
- Check Topics Portfolio for competitor-cited themes where your business is absent.
- Prioritise fixes by revenue value, not by query volume alone.
- Update or create the right page type, such as a pricing page, comparison page or local service page.
- Add visible answers, named authors, current dates and sourced figures to the page you want cited.
- Strengthen proof with reviews, roundups, business listings and fair comparisons where relevant.
- Check that key pages are in Bing's phone book and not blocked by your AI welcome sign.
- Track whether more pages, not just more citations, start appearing over time.
The biggest mistake is treating the report as a dashboard only.
If you only watch the numbers, you will miss the opportunity. The real value is in the decisions the report supports.
Use it to choose what to publish next. Use it to improve pages that already have a chance. Use it to find competitor gaps before your next quarter starts. Use it to explain to leadership why a pricing page, a comparison page or a review strategy may matter more than another generic awareness article.
For local firms, the same logic applies on a smaller scale. The report can show whether Bing connects your business with local intent and service terms. If local visibility is your priority, our regional guides such as How Limerick businesses can appear in AI search results show how to build that foundation.
Is Bing's AI Performance report only useful if most of my customers use Bing?
No. It is useful because it is first-party evidence from Microsoft about AI citations and grounding queries. It helps with Bing and Copilot directly. It also shows which page types and topics AI search systems may trust. Just remember that Bing visibility does not guarantee ChatGPT visibility, because ChatGPT search has used its own index since 4 September 2026, according to OpenAI reporting cited by industry coverage at the time.
How often should I review grounding queries?
Once a month is a good rhythm for most established businesses. That is frequent enough to spot new topic gaps and page wins, without turning it into daily noise. If you publish often or run active campaigns, you may check key themes more often.
What is the difference between a grounding query and a keyword?
A keyword is usually a search term you track for search engine optimisation. A grounding query is the underlying prompt or search intent Bing used to build an AI answer. It is often closer to a buyer question than to a classic keyword list.
Should I create a new page for every grounding query?
No. Start by grouping queries into themes. Then decide which existing page should answer
