Bing’s New Citation Share Metric: A GEO Guide

Bing Webmaster Tools AI Citation Share is the standout in a June 16, 2026 Microsoft update that, for the first time, lets publishers see what percentage of citations they own across an AI-generated answer — not just how many times they were cited. Microsoft shipped four new capabilities together: Intents, Topics, Citation Share, and Compare. All four are free and, per Microsoft, in preview globally.

That matters because generative-engine optimization has, until now, been measured almost entirely with estimates. Practitioners scraped AI answers, ran prompt panels, and inferred visibility from third-party tools. Bing’s AI Performance report — first launched in public preview on February 10, 2026 — was the first time a major AI search provider handed publishers direct performance data. The June 16 update layers analytical structure on top of that foundation.

This guide covers exactly what shipped, what Citation Share does and does not tell you, why grounding queries are a new mental model, how to turn the Intents categories into a content-priority matrix, and where Bing’s native reporting still leaves gaps you have to fill with estimation or third-party tools. Every figure below is sourced; vendor-stated and self-reported numbers are labeled as such.

Microsoft announced the update on its Bing blog on June 16, 2026, authored by four Microsoft AI product managers. The framing was deliberate: this is part of an “ongoing effort to provide greater transparency into how content appears across AI-powered experiences.” The four new capabilities sit inside the existing AI Performance report, which covers citations across Microsoft Copilot, AI-generated summaries in Bing, and select partner AI experiences. It does not cover ChatGPT, Perplexity, or Google AI Overviews — an important scope boundary to set with stakeholders before they read too much into the numbers.

The February 2026 launch shipped four original metrics: Total Citations, Average Cited Pages, Grounding Queries, and Page-Level Citation Activity. The June 16 release is additive — Intents, Topics, Citation Share, and Compare are analytical lenses on top of that base. Microsoft is explicit that everything here is in preview, and that preview-phase classification will keep improving as the models learn from real-world usage.

Citation Share answers a question raw counts cannot: of all the citations an AI answer used for a given query, what fraction were yours? Per Microsoft, the calculation is simple — your citations for a grounding query, divided by the total citations shown across all sites for that same query, multiplied by 100. If your site receives 3 of the 10 citations an answer drew on, your Citation Share for that query is 30%.

The reframe is the point. A raw count of “cited 40 times this month” tells you nothing about competition. A Citation Share of 12% versus 60% on the same query tells you whether you are a footnote or the primary source. It moves GEO measurement from volume toward share of voice — the same shift display advertising made decades ago when impressions gave way to share-of-shelf thinking.

Microsoft is careful about what Citation Share is not. The company states plainly that it is an observational metric, not a ranking system or a competitive scoreboard — it does not expose competitor domains, represent traffic share, or assign quality scores to your content. That disclaimer is load-bearing. Citation Share tells you how much of the citation pie you own per query; it cannot tell you whether that query drives traffic, leads, or revenue. Treat it as a visibility signal to be triangulated with your own analytics, not a standalone business metric.

The single most important conceptual distinction in the whole report is also the most underexplained in trade coverage. The Grounding Queries metric shows the key phrases the AI generated internally when retrieving content to support an answer — not the question a user typed. When someone asks Copilot a messy, conversational question, the model decomposes it into one or more cleaner retrieval phrases. Those internal phrases are what the report surfaces.

This is a mental-model shift as significant as the move from keywords to topics in traditional SEO. You are no longer optimizing for what humans type; you are optimizing to be retrievable for the phrases an AI system constructs on a human’s behalf. Two practical consequences follow. First, a grounding query may look nothing like your target keyword, so do not expect a clean one-to-one map. Second, content that answers a precise sub-question cleanly — with a clear heading, a direct answer, and supporting evidence — is exactly what an AI system can lift into a grounded answer. Microsoft’s own February 2026 guidance pointed the same way: strengthen depth and expertise, improve structure and clarity with headings, tables and FAQs, support claims with evidence, keep content fresh, and reduce ambiguity across formats.

Citation Share gets the headlines, but the other three capabilities are what make it actionable. Each one is a different way of slicing the same citation data.

Intents classifies grounding queries into seven-plus categories — Informational, Commercial, Navigational, Learn and Solve, Research, Creation, Local, and more. The unstated “and more” means the taxonomy is not fully published yet, so treat the seven named categories as the working set. The value is diagnostic: you can finally see which intents you win citations on and which you are invisible for.

Topics groups related grounding queries into thematic clusters using AI/ML classification. Microsoft’s example maps “solar panels,” “solar energy efficiency,” and “residential solar installation” to a single “Solar Energy” cluster. The company is candid that during preview, labels “may still be broad — especially for highly specialized or niche domains,” and will sharpen over time.

Compare overlays a previous period on the current view — the current 30 days against the prior 30, or a custom range. It is the one feature with a dependency: it is only useful once you have accumulated history. That single constraint is the strongest argument for getting into the dashboard now rather than later.

Read together, the three lenses turn a flat citation count into a map you can act on. Intents tells you where your citations cluster by purpose, Topics tells you what themes you own, and Compare tells you which direction both are moving. For a deeper grounding in the discipline these metrics serve, our GEO (Generative Engine Optimization) fundamentals guide is the companion read, and our generative engine optimization guide walks the broader 2026 playbook.

Microsoft named the intent categories but stopped there — it offered no guidance on what to do when your citations are low in one of them. That is the gap worth filling. Map each category to the content formats AI answers tend to cite, the type of publisher it matters most to, and the lever to pull when you are under-cited. An eCommerce site might find it has strong Informational citations and near-zero Commercial visibility — a concrete, actionable gap that the raw dashboard would never surface on its own.

The table below pairs Microsoft’s named categories with our inferred content-strategy guidance. The category names are vendor-stated; the format, audience, and optimization-lever columns are Digital Applied’s interpretation, not Microsoft’s.

The practical workflow: pull your Citation Share by intent, find the categories where your share is lowest relative to your business goals, and reallocate content effort there. A B2B SaaS brand that already dominates Informational citations gets more marginal value from closing its Commercial gap than from defending a lead it already holds. This is the same prioritization logic behind a brand citation audit checklist — applied now with native, per-intent data instead of scraped estimates.

Honest measurement means naming the boundaries. Bing’s reporting is the most transparent native AI-visibility tooling any major platform has shipped — and it still leaves real gaps. The map below charts each GEO measurement dimension against where it is available: at the February 2026 launch, after the June 16 update, whether it is still missing from every native platform, and whether third-party tools cover it. Cells are sourced from the two Bing blog posts and the Adobe CX Enterprise announcement.

The bottom half of the map is where GEO work still depends on estimation. There is no competitor-domain visibility — by design, since Microsoft positions Citation Share as observational, not a scoreboard. There is no click-through attribution and no traffic share, which is why you must triangulate citation data with your own analytics rather than treating share as a revenue proxy. Enterprise tools are starting to fill some of this: Adobe launched a brand visibility AI search tool inside Adobe CX Enterprise in June 2026 that shows brands where they win and lose across AI surfaces. Adobe’s own analytics data also reports steep AI-referral growth — directional evidence that the surface is worth measuring, even if the precise percentages are vendor-reported rather than independently verified.

That is the clear-eyed framing: Bing’s tool is a genuine leap for native transparency, and it is not a complete GEO measurement stack. Knowing exactly where it ends is what lets you choose, deliberately, which gaps to close with estimation, which with third-party tools, and which to leave alone. Our AI citation visibility audit data and tracking your brand’s AI share of voice cover the estimation methods that fill in the missing columns.

Here is the original-analysis case for acting immediately rather than waiting for the feature to leave preview. The Compare lens is only useful with history — and history is the one input you cannot backfill. A brand that establishes a citation baseline in Bing’s AI Performance report today will hold months of trend data by the time any equivalent reporting matures elsewhere. As of mid-June 2026, Google has not provided equivalent direct citation-level reporting to publishers via Search Console, and Microsoft’s tool is the first free, native GEO measurement window from a major AI search platform. We are not going to predict when, or whether, a competitor ships a match — but the asymmetry favors whoever started collecting first.

The demand-side timing reinforces it. A Pew Research Center study conducted February 17–23, 2026 (n=5,119 U.S. adults) found that 60% of Americans read AI-generated summaries at the top of search results, and roughly half use AI chatbots overall — up from about a third in 2024. AI referral traffic is climbing fast across categories; Adobe’s analytics data points to triple-digit and higher growth in AI-driven visits to retail and travel sites since late 2024, which we read as directional rather than precise. The combined signal is simple: AI answers are where a growing share of attention now lands, and Bing just handed you a free instrument to measure your slice of it. The cost of starting is near zero; the cost of a missing baseline compounds.

The feature is free and live in preview. Here is how to turn it into a repeatable practice rather than a one-time look. Pick the entry point that matches where your team already is.

If your team would rather have this run as a managed practice — the baseline, the per-intent diagnosis, the content work, and the period-over-period reporting — that is exactly what our agentic SEO services are built to deliver. We also maintain a reusable GEO operating framework that slots Bing’s native metrics into a broader technical and editorial spec.