Microsoft Clarity now shows whether AI cites you by name — or finds you on merit.

What Microsoft shipped

The announcement adds three things to the Citations dashboard that went generally available on 2026-05-13: individual queries in the queries card are “now clearly marked as branded”; Share of Authority — defined by Microsoft as your citations divided by total citations across all domains, computed daily — now shows “where your authority is strongest: queries that mention your brand versus more general queries”; and you can “filter dashboard data by branded or non-branded queries to compare how your visibility performs.” Grounding queries are the retrieval queries an AI assistant generates to look up supporting content before it answers — per Microsoft's docs, they “may differ from the exact phrases users type.” The dashboard is free; domain verification takes the Clarity tracking snippet, Google Search Console, or Bing Webmaster Tools.

Branded vs non-branded is the split that matters

A citation on a branded grounding query means the user (or the assistant on their behalf) already had your name — the AI is looking you up, not choosing you. A citation on a non-branded query means the AI selected you while answering a generic buyer question, against every competitor at once. Those are different businesses' realities wearing the same “we appear in AI” label: our own 144-answer instrument found local businesses were named 99% of the time when asked about by name, but appeared in only ~20% of answers when the question was phrased the way a customer would actually ask. Until now you needed synthetic prompt-sampling to see that gap. Clarity now shows one first-party version of it on a free dashboard: if your Share of Authority is healthy on branded queries and near zero on non-branded ones, AI knows you exist and recommends someone else.

What it can and cannot see

Two limits worth stating plainly. First, Microsoft does not enumerate which AI experiences feed the dashboard — the data comes from its own grounding infrastructure, so assistants that ground their answers elsewhere (Gemini on Google's index, Claude, Perplexity's own crawl) are invisible to it. A clean Clarity report is not a clean bill of AI health. Second, grounding-query data only exists where retrieval happened — it cannot tell you about the buyer questions where you were never retrieved at all, which is precisely the non-branded discovery you want to win. Treat Clarity as the free first-party floor of an AI-visibility stack (here is how it compares to the paid trackers) and pair it with question-level sampling across engines — checking where you actually show up across ChatGPT, Perplexity, Gemini and Claude is the cross-engine view Clarity structurally cannot give you.

The cadence is the other signal: topic insights on 2026-07-09, query-topic grouping on 2026-07-22, branded segmentation on 2026-08-03. Microsoft is building AI-citation analytics into free tooling at a monthly clip — which means “how often does AI cite you, and on what queries” is about to be a number your clients can look up themselves. Agencies should be the ones explaining it to them first.