Microsoft Clarity splits AI citation queries into branded and non-branded

Microsoft Clarity has added branded and non-branded segmentation to the Citations view of its AI Visibility dashboard, announced on 3 August 2026. Individual queries are now marked as branded, both types can be filtered, and the Share of authority card breaks its results out by each. Microsoft defines branded queries as those that “reference your brand directly”, against “broader, generic queries the AI system uses to help answer a user’s prompt”.

The split is the standard way to read every other query dataset, so extending it to observed AI citations is more useful than the size of the release suggests. It separates two things that a single citation count runs together: being cited because someone asked about you by name, and being cited on a generic question where the model had the whole web to choose from. The second is the one that shows reach beyond an audience that already knows you. A high citation count made up almost entirely of branded queries is a retention signal rather than an acquisition one, and it will rise on its own as AI answers push more people to search for you directly.

Two limits carry over from the underlying data and neither is mentioned in the announcement. Clarity’s citation dataset is described by Microsoft as a representative view rather than a complete log, and Share of authority only counts days on which your domain was cited for a given query, which pushes the figure higher than a broader calculation would. The dashboard also names its source as Microsoft Copilot and unspecified partners, so a branded/non-branded split describes Copilot behaviour, not AI search generally. Read the ratio as a trend on one surface, and do not present it to stakeholders as an AI-wide number.

Sources

• Branded and non-branded queries in AI Visibility — Microsoft Clarity