Reports Observe Bing Webmaster Tools AI Performance Dashboard in Public Preview

Product change and affected accounts

The Bing Webmaster Tools team announced on February 10 2026 that AI Performance is in public preview, exposing citation activity for publisher content across Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations . The post is attributed to Krishna Madhavan, Meenaz Merchant, and Fabrice Canel on the official Bing Webmaster Blog. Because the release is labeled a public preview rather than a general availability launch, the affected account scope is publishers who can opt into the dashboard inside Bing Webmaster Tools during the preview period; no separate enrollment form is described in the source documents.

This change is a platform reporting expansion, not an algorithm update. Bing Webmaster Tools has historically covered indexing, crawl health, and search performance; AI Performance extends those surfaces to AI-generated answers, framed by the team as an early step toward Generative Engine Optimization (GEO) tooling inside Bing Webmaster Tools. The list of “select partner integrations” is not enumerated in the source post, so which specific third-party AI surfaces contribute to the dashboard is not specified in the available evidence. Practitioners can pair this dashboard with Bing Webmaster Tools launches AI Performance in public preview — a citation dashboard, not a ranking signal when cross-checking framing.

Metric semantics and validation

The dashboard exposes five named metrics: Total Citations, Average Cited Pages, Grounding queries, Page-level citation activity, and Visibility trends over time . Each metric carries explicit scope disclaimers in the source post that practitioners must absorb before treating any number as a performance signal.

Total Citations counts the total number of citations displayed as sources in AI-generated answers during the selected time frame; the post states this metric highlights how often content is referenced, without indicating placement or presentation within a specific answer . Average Cited Pages reflects the average number of unique pages from the site displayed as sources per day over the selected range, aggregated across supported AI surfaces, and does not indicate ranking, authority, or the role of any page within an individual answer . Page-level citation activity shows citation counts for specific URLs during the selected date range and reflects citation frequency, not page importance, ranking, or placement . The same claim carries the caveat that Grounding queries represent a sample of overall citation activity that Bing will continue to refine .

The clustering of caveats matters. Three of the five metrics explicitly disclaim placement, ranking, authority, and page-role interpretation. Reading Total Citations as a ranking proxy, or Average Cited Pages as an authority indicator, would import a meaning the post explicitly withholds. Reviewers auditing the dashboard should treat the numbers as citation frequency distributions, with grounding query data as a sampled subset that the vendor expects to evolve.

Operator workflow and measurement limits

The source post positions the dashboard as a content-improvement input rather than a ranking diagnostic. The recommended workflow is to review cited pages and grounding query phrases to validate which pages are already referenced in AI answers, identify content that appears frequently, and spot opportunities to improve clarity, structure, or completeness on pages that are indexed but less frequently cited. The team also lists content practices — strengthening depth and expertise, improving structure and clarity, supporting claims with evidence, keeping content fresh and accurate, and reducing ambiguity across formats — and points to a separate “Optimizing Your Content for Inclusion in AI Search Answers” guide for deeper guidance.

Two measurement limits are stated up front. First, Bing states that it respects all content owner preferences expressed through robots.txt and other supported control mechanisms when reporting AI Performance insights, so publishers who block Bingbot or specific AI crawlers should expect underreporting in their citation counts . Second, the post ties freshness to IndexNow, which notifies participating search engines whenever content is added, updated, or removed, helping ensure AI systems reference the most current version of a page when generating answers . For local businesses, the post directs registration with Bing Places for Business so address, hours, and contact information remain current and eligible for inclusion in AI-generated responses .

Practitioners should also keep in mind that the supporting blog archive from February 10 2026 sits alongside earlier posts on IndexNow adoption, sitemaps in AI-powered search, the data-nosnippet HTML attribute, and AI-search conversion measurement, indicating that AI Performance is being positioned as one report inside a broader Bing tooling stack rather than a standalone product. The workflow is consistent with a Serp Snippet Preview check for snippet-shaped rendering, used alongside the citation dashboard rather than as a substitute for it.

Knowledge Delta: new evidence, mechanism, decision, and falsifiable follow-up signal

Mechanism evidence. The dashboard aggregates citation events from supported AI surfaces into five metrics, applies an explicit non-ranking disclaimer to the citation-frequency metrics, and samples grounding queries. The post treats IndexNow as the freshness lever that keeps cited URLs aligned with the version AI systems retrieve. The decisions the team is signalling to publishers are: treat citation frequency as a content coverage signal, treat grounding query phrases as a topic-completeness input, and treat robots.txt and Bing Places controls as gating conditions for what shows up at all.

Alternative explanations to keep on the table. First, the “select partner integrations” wording leaves a possibility that citation counts are dominated by a small set of surfaces, which would mean a single partner changing its citation policy could move a publisher’s Total Citations without any change to the underlying content. Second, because grounding queries are explicitly a sample that will be refined, a stable grounding query set over two consecutive weeks could reflect a stable sample rather than stable publisher relevance. Third, AI surfaces can cite content without fetching it through IndexNow, so a publisher with no IndexNow adoption could still register citations if those surfaces independently discover URLs.

Uncertainty and falsifiable follow-up signal. The post does not specify reporting latency, the date range covered by “selected time frame,” how “select partner integrations” is filtered, or whether API access to AI Performance metrics is included in the public preview. The most concrete falsifiable follow-up signal is whether Page-level citation counts move in step with on-page edits submitted through IndexNow within a known freshness window; if they do not, freshness is not the controlling variable, and citation frequency is being driven by retrieval-side selection independent of IndexNow. A second signal is whether grounding query phrasing changes after Bing refines the metric; stability of phrasing across two refreshes would indicate that the sample is mature, while a wholesale rewrite would indicate that prior grounding data should be treated as preliminary.

Public Action Brief

txt does not block the crawlers whose citations would feed the report, and confirm IndexNow submission is enabled for content that is being edited. LABEL: WHAT NOT TO CHANGE YET: Do not rewrite high-citation pages solely because Total Citations moved; the post explicitly disclaims ranking, placement, and authority interpretation, so a citation delta is not evidence of a ranking shift and is not a confirmed ranking signal. LABEL: MEASUREMENT BASELINE: Capture the current Total Citations, Average Cited Pages, and Page-level citation activity for the property as the public-preview baseline.

LABEL: MEASUREMENT METRICS: Total Citations, Average Cited Pages, Grounding queries, Page-level citation activity, Visibility trends over time. LABEL: MEASUREMENT SEGMENTS: By URL, by grounding query phrase, and by AI surface where the dashboard distinguishes them. LABEL: OBSERVATION WINDOW: Two consecutive weeks inside the public preview, long enough to span at least one IndexNow resubmission cycle for any edited URLs.

LABEL: WHAT WOULD CHANGE THIS CONCLUSION: A future Bing announcement that explicitly ties AI Performance metrics to ranking, placement, or authority; an API surface that exposes per-surface citation counts and resolves the “select partner integrations” ambiguity; or a vendor statement that grounding queries are no longer a sample. LABEL: WHEN TO REVIEW: At the end of the two-week observation window, or sooner if Bing publishes a clarification on metric semantics or partner coverage. LABEL: APPLICABILITY: Publishers with verified Bing Webmaster Tools property access during the public preview; local businesses should pair the dashboard with Bing Places for Business registration.

LABEL: RISK BOUNDARY: The dashboard measures citation frequency, not ranking, placement, or authority; reading the metrics as ranking signals would misdiagnose the underlying search behavior.