Microsoft Adds AI Visibility Insights to Webmaster Tools

According to a Microsoft blog post dated June 16, 2026, Bing Webmaster Tools expanded its AI Performance report with four new preview capabilities: Intents, Topics, Citation Share, and Compare. Intents classifies grounding queries into categories such as Informational, Commercial, Navigational, Learn and Solve, Research, Creation, and Local. Topics groups related grounding queries into thematic clusters. Citation Share reports the percentage of citations attributed to a site out of all citations shown for a given grounding query. Compare overlays a prior time period to show how citation activity shifts. Microsoft says the features are rolling out in preview globally and build on the AI Performance report it launched in February 2026.

A raw citation count conflates frequency with relevance and says nothing about query context or how a system selected a passage as evidence. Intent classification maps citations onto query-type distributions; Topic clustering aligns reporting with the thematic groupings retrieval systems use internally rather than isolated keywords. Search Engine Journal's Duane Forrester, who previously built Bing's original Webmaster Tools, argues this is the clearest public confirmation yet that "ranking a page and citing a passage are different jobs" - the index owner itself now runs two separate dashboards (Search Performance and AI Performance) with different definitions of success.

Teams instrumenting SEO and content analytics can now treat citation events as structured data: compute intent-weighted visibility, use Citation Share as a directional proxy for relative presence in a citation set, and watch Compare windows for topic-level drift. That makes testing content variants against generative visibility more tractable, because the metrics separate where a page is used (topic) from why (intent) and how much it contributes relative to other cited sources (citation share). Search Engine Journal's analysis cautions against treating Citation Share as a single scoreboard, however: it only reflects Microsoft's own surfaces (Copilot, Bing, partner integrations) and says nothing about how ChatGPT, Gemini, or Perplexity treat the same content, since Microsoft's Web IQ grounding infrastructure powering some of this does not report competitor citations back to publishers.

Key indicators as the preview matures: whether Microsoft documents classification methodology for intents and topics (reproducible definitions determine how reliably teams can backtest content changes), whether sampling and geographic breakdowns are exposed, whether API access for programmatic reporting arrives, and whether Google Search Console's own Generative AI report - which currently shows impressions but not clicks - moves toward comparable depth.

This is a meaningful step toward measurement parity for generative search visibility, not a change to model architectures or grounding fidelity. Its main limitation is structural rather than temporary: first-party reporting from any single platform, however good, only ever describes that platform's own house. Practitioners who rely solely on Bing's dashboard risk mistaking one instrument's view for the state of the field; cross-engine, third-party measurement remains necessary to see how content performs across Copilot, ChatGPT, Gemini, and Perplexity together.