iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Bing Webmaster Tools’ AI Performance report shows when pages from your site are cited in AI-generated answers. It covers Microsoft Copilot, AI-generated summaries in Bing, and selected partner integrations. Read it as an observation of source usage—not as a new ranking report, traffic report, or content-quality score.
The useful workflow is to establish a consistent date range, identify pages and trends with repeated citations, then use grounding queries, Intents, Topics, Citation Share, and Compare to form and test hypotheses. None of those signals, by itself, proves ranking position, traffic impact, or the cause of a visibility change.
What the AI Performance report measures
Microsoft introduced AI Performance in Bing Webmaster Tools as a public preview on February 10, 2026. The report records observed citations: instances in which an AI-generated answer displays a page from your site as a source. The supported surfaces are Microsoft Copilot, AI-generated summaries in Bing, and selected partner integrations.
Recommended Free Tools
A citation means your URL was used as a displayed reference in an answer. It does not mean that the page ranked first in a conventional search result, received a visit, influenced every answer for a query, or was judged to be the best page on the subject. AI Performance complements ordinary search-performance reporting; it does not replace impressions, clicks, crawl, or indexing reports.
#1 Best Overall
Because the report aggregates activity across supported AI experiences, a single metric cannot describe the page’s position or role in a particular answer. Treat the data as evidence about where your content is being used, then investigate the underlying pages and retrieval phrases.
How to read each original dashboard metric
| Metric | What it shows | Useful interpretation | What it does not establish |
|---|---|---|---|
| Total Citations | The number of citations displayed as sources in AI-generated answers during the selected date range. | Use it to measure the volume of observed references and to locate periods that merit investigation. | It is not a search ranking, click count, impression count, or measure of how prominently a citation appeared in an answer. |
| Average Cited Pages | The average number of unique pages from your site displayed as sources per day over the selected period. | Use it to see whether citation activity is concentrated on a few URLs or distributed across more of your site. | It does not indicate authority, ranking, or the importance of any page in an individual answer. |
| Grounding Queries | A sample of phrases used when the AI retrieved content that was later cited. | Use the phrases as directional evidence of the needs and wording associated with your citations. | It is not a complete query export and should not be treated as the total demand for a topic. |
| Page-level citation activity | Citation counts for individual URLs. | Use it to find pages already functioning as references and to compare them with indexed pages that receive few or no citations. | A high count does not prove that a page drives traffic, ranks highly, or is the only page used for a subject. |
| Visibility trends | A timeline of citation activity across supported AI experiences. | Use it to locate sustained increases, declines, or abrupt changes that need a cause analysis. | A movement on the line does not identify the reason for the movement. |
Microsoft’s launch description defines Total Citations as the total number of citations displayed as sources in AI-generated answers during the selected time frame. That wording is important: the count is about displayed references, not every retrieval event that may have occurred behind an answer.
What the June 16, 2026 preview additions add
Microsoft announced four globally rolling preview capabilities on June 16, 2026: Intents, Topics, Citation Share, and Compare. They add context to the original counts but do not turn citation data into a conventional ranking system.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Intents: the type of need behind a grounding query
Intents groups grounding queries into categories such as Informational, Commercial, Navigational, Learn and Solve, Research, Creation, and Local. Use the grouping to distinguish the job a cited page appears to help with. For example, a page cited for a Learn and Solve query may need clear procedural steps, while a page cited for a Research query may need definitions, evidence, and links between related concepts.
Intent labels describe the query group, not the quality of your page or the likelihood of a visit. A single topic can appear under several intents, and the available phrases are still based on the report’s sampled grounding-query data.
Topics: broader themes across related phrases
Topics clusters related grounding queries into broader subject areas. This helps reveal themes where your site is gaining visibility and themes where coverage is thin. Compare the themes represented by repeatedly cited URLs with the themes represented by indexed URLs that rarely appear as citations. A gap may indicate that the page does not answer the wording being retrieved, that another page is a better source for the same need, or simply that the sample has not captured enough activity to support a conclusion.
Citation Share: an observational proportion, not a ranking
Citation Share is the percentage of citations attributed to your site out of all citations shown for the same grounding query. In simplified form:
Rank #2
Citation Share = your site’s citations for a grounding query ÷ all citations shown for that grounding query × 100.
Microsoft explicitly describes Citation Share as an observational metric, not a ranking system or competitive scoreboard. It is also not traffic share, market share, or a content-quality score. A higher percentage means that your site accounted for a larger portion of the displayed citations in the observed sample for that query; it does not mean your page will rank higher in web search or receive a corresponding percentage of visits.
Do not compare Citation Share values across unrelated queries as though they were a single league table. Query wording, the number of sources shown, the AI surface, and the period observed can all differ.
Compare: a way to inspect change over time
Compare overlays an earlier period on the current view. Use it to identify what changed, then investigate before assigning credit or blame. Microsoft cautions that AI citation ecosystems are inherently dynamic. A difference between periods can reflect changed demand, seasonality, model behavior, competing pages, freshness signals, content edits, or refresh timing in a partner integration.
Free tools Windows power users keep installed
One-click scans. No signup required.
Compare shows association between periods; it cannot prove that one edit caused the movement. A causal claim requires a controlled change and checks for other explanations.
A practical workflow for interpreting the report
- Choose and record a consistent date range. Select the period you intend to analyze and write down the report date, site, and filters. Keeping the range and review cadence consistent prevents a short-lived spike from being mistaken for a durable trend.
- Start with Total Citations and the trend line. Look for a sustained increase, decline, or abrupt step. A one-day movement is a prompt for investigation, not a verdict. Note whether the change affects the whole site or only a particular period.
- Open page-level citation activity. List URLs that receive repeated citations. Separate pages with recurring activity from pages that appear once. The recurring group is your current set of pages functioning as references in the observed AI answers.
- Inspect the grounding queries for those URLs. Read the actual phrases and group them by the need they express. Check whether the page answers the words and task implied by the query, rather than only matching a broad keyword.
- Use Intents and Topics to add context. Record which intent categories and topic clusters are associated with each URL. This can expose a mismatch, such as a page receiving research-oriented citations even though its content is written only as a short definition, or a topic cluster for which your site has no complete guide.
- Compare cited and less-cited indexed pages. Review pages that are indexed but receive few or no citations against pages cited repeatedly. Examine headings, concise definitions, supporting evidence, tables or FAQs where appropriate, entity clarity, completeness, and the date of the latest meaningful update.
- Make one controlled content change. Choose a specific improvement tied to the observed need—for example, clarify a definition or add evidence for a question shown in the grounding queries. Keep a record of the URL, the exact change, and the publication date.
- Revisit Compare in a later period. Use the same or a deliberately documented comparison window. Treat any movement as correlation until you have checked demand, seasonality, model changes, competing pages, freshness, and partner refresh cycles.
- Check discovery controls. Ensure robots.txt and other owner controls intentionally allow or restrict the pages you want indexed. Use IndexNow when pages are added, updated, or removed so participating engines can discover the current version faster. IndexNow is a discovery aid, not a guarantee that a page will be cited.
How to investigate a rise or fall in AI visibility
First determine the shape of the change
Use the trend line and page-level view together. If many URLs move in the same direction, investigate site-wide or ecosystem factors. If one URL changes while others remain stable, start with that page’s content, indexing state, canonical signals, and recent edits. If the change appears only for one topic or intent, inspect the associated grounding queries rather than assuming a global algorithm event.
Check demand and seasonality
Grounding queries reflect needs expressed to the AI systems. A decline can occur because fewer people are asking about a subject during the comparison period. Seasonal events can create the reverse pattern. Compare equivalent periods when seasonality is plausible, and avoid attributing a change to an edit solely because the edit happened shortly beforehand.
Check model and surface changes
The report aggregates Microsoft Copilot, Bing AI-generated summaries, and selected partner integrations. Changes in a model, answer format, source-selection behavior, or a partner’s refresh cycle can alter which citations are displayed without any change to your page. The report does not identify a single cause, so record the affected surface when that information is available and keep conclusions proportional to the evidence.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Check competing and fresher sources
AI answers can select different sources as competing pages are published, updated, or made easier to retrieve. Compare your page’s update date, evidence, structure, and coverage with the pages now appearing for the same grounding-query themes. This is a content diagnosis, not proof that a competitor displaced you for a fixed rank.
Check indexing and owner controls
Confirm that the cited or declining URL is still accessible to the intended crawlers and has not been unintentionally blocked by robots.txt or another owner control. If the page changed, submit the updated URL through IndexNow where supported. Faster discovery can help an engine find the current version, but it cannot force citation.
What to improve on pages that rarely receive citations
The report is most useful when it changes a concrete editorial decision. Microsoft’s guidance points to clear structure, evidence, completeness, and freshness. Apply those principles to the needs visible in your grounding-query sample.
- Make the answer explicit early. Put a precise definition or outcome near the beginning of the page, then expand with details. This helps a retrieval system and a human reader identify the page’s subject.
- Use descriptive headings. Organize the page around the questions and tasks represented by the grounding queries. Avoid headings that are clever but do not state the subject.
- Support important claims. Add primary evidence, dates, specifications, or clearly labeled assumptions where the topic requires them. Do not imply certainty that the page cannot support.
- Cover the complete task. Include prerequisites, steps, exceptions, costs or limits when relevant, and recovery paths for common failures. A page that answers only the first sentence of a question may be less useful as a source.
- Clarify entities and relationships. Spell out product names, standards, versions, locations, and the relationship between them. Resolve ambiguous terms instead of relying on context that a retrieval system may not share.
- Use tables and FAQs only when they add information. A comparison table can expose differences quickly; an FAQ can address distinct unanswered questions. Do not repeat the same answer in multiple formats.
- Keep meaningful dates current. Update instructions, compatibility details, and definitions when they change. An updated date without substantive changes is not evidence of freshness.
These changes improve the page’s usefulness and retrievability, but the report cannot promise that any particular edit will produce more citations.
How to interpret a worked pattern without overclaiming
Suppose one guide has repeated page-level citations, while several related guides are indexed but rarely cited. Begin by reading the grounding queries attached to the cited guide. If the phrases consistently ask for a definition and a procedure, check whether the other guides omit one of those elements. If the phrases fall into a different Topic or Intent, the pages may serve different needs rather than compete for the same visibility.
Next, compare the cited guide’s structure, evidence, entity wording, and update history with the less-cited pages. Choose one change on one URL, document it, and use a later Compare view. If citations rise, that is evidence consistent with the change, not proof that the change alone caused the rise; demand, model behavior, competing sources, and partner refresh timing still need to be considered.
Common misreadings and the correct response
- “More citations means a higher search ranking.” No. Citations count displayed source references in AI answers; ranking is a different measurement.
- “Average Cited Pages tells me which page is authoritative.” No. It is an average of unique cited URLs per day across supported AI surfaces and does not assign authority.
- “Grounding Queries is my complete AI keyword report.” No. Microsoft describes it as a sample. Use it for direction, not demand forecasting.
- “Citation Share is a competitor leaderboard.” No. It is the percentage of observed citations for the same grounding query and is explicitly not a ranking system, competitive scoreboard, traffic share, or quality score.
- “Compare proves my last edit worked.” No. It shows period-over-period movement. Check seasonality, demand, model changes, competing pages, freshness, and partner refresh cycles before claiming causation.
- “IndexNow guarantees a citation.” No. It helps participating engines discover additions, updates, and removals faster; it does not guarantee indexing or source selection.
- “A sudden drop must be a penalty.” The report does not establish penalties. First check query demand, affected URLs, controls, content changes, model or surface changes, and competing or fresher sources.
Build a measurement routine that remains useful
Keep a dated record of the selected range, Total Citations, Average Cited Pages, notable URLs, representative grounding queries, Intents, Topics, and any Compare period. Add a change log for content edits, publication dates, indexing changes, and major site-wide releases. This makes it possible to distinguish a recurring pattern from an isolated observation.
Review pages with repeated citations for opportunities to maintain accuracy, evidence, and clarity. Review indexed pages with little activity for topic or intent gaps, but do not assume every page should be cited. AI answers use a selection of sources, and the report measures what was displayed in the observed period rather than every page that could have been retrieved.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteUse the report to prioritize questions: Which reader needs are represented? Which topics have strong or weak coverage? Which URLs are repeatedly useful as references? What changed between two periods? Answer those questions with the narrowest claim the data supports.
The limits you should keep in every conclusion
AI Performance is a public preview. The published material describes the metrics and four preview capabilities, but it does not provide independent statistics for usage, accuracy, traffic impact, or lift. Do not fill those gaps with estimated percentages or claims that the report has demonstrated a return.
Its data is observational and aggregated across supported AI experiences. Citation ecosystems are dynamic, and the report cannot by itself identify why a page was selected, how prominently it appeared, whether a user clicked it, or whether an observed change will persist. Conventional search reports, analytics, crawl and indexing diagnostics, and a documented content-change process are still needed to understand the broader outcome.
The most defensible reading is therefore practical: citations show where your pages were used as displayed sources; grounding queries, Intents, and Topics show the needs and themes associated with the observed sample; Citation Share describes your share of displayed citations for a query; and Compare helps you investigate change without pretending it proves causation.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteQuick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
