Microsoft Clarity Splits AI Citations Into Branded and Earned Discovery. The Non-Branded Number Is Your Real Score.

Microsoft Clarity now segments AI citations by branded and non-branded queries. The split exposes a problem most brands do not know they have: your aggregate citation count is two different numbers pretending to be one. Branded citations measure recognition you already earned somewhere else. Non-branded citations measure whether AI engines choose your content when nobody asked for you by name. Those are different business outcomes, and most brands are only watching the total.

What Microsoft Shipped on August 3

On August 3, 2026, Microsoft Clarity added branded query segmentation to its AI Citations dashboard. The update includes four changes: branded labels on individual grounding queries, a Share of Authority breakdown split by query type, filters that apply the branded/non-branded distinction across all dashboard data, and what Microsoft describes as "more precise citation analysis" from the separation.

The feature is live inside the AI Visibility dashboard for every Clarity user. Free. No beta gate. It is the third citation-side release from the same product in 25 days, following Topic Insights on July 9 and Query Topics on July 22.

Microsoft's framing is careful: "Not all AI visibility comes from the same kind of query." That is an understatement for a distinction that has been quietly corrupting every AI visibility report since the category was invented.

What the Branded Split Reveals About Your Citations

A branded citation happens when an AI system retrieves your page because someone typed your company name into a prompt. A non-branded citation happens when the AI retrieves your page while answering a category question that had nothing to do with you.

The first measures recognition. The second measures reach.

Microsoft's own sample data makes the gap visible. Branded queries in the published example carried 24.2% and 22.3% Share of Authority, while the three non-branded queries sat at 20.3%, 18.6%, and 15.9%. Branded queries dominated authority despite lower citation volume. A brand reading only its aggregate total would see a healthy number and conclude it was winning. A brand reading the split would see that its strongest position depends entirely on users who already knew its name.

The inverse is worse. A domain whose citations are overwhelmingly branded is not being discovered by AI. It is being confirmed. Those citations follow demand generated by advertising, press coverage, or word of mouth. They disappear the moment that upstream spending stops.

Aggregate reporting has been unable to tell these two states apart. Now it can.

Three Citation Releases in 25 Days

The cadence tells you where this is heading.

Topic Insights launched July 9, giving every Clarity user competitive citation analysis at no cost, at a point when rival AI visibility vendors reserved comparable features for paying tiers. Query Topics entered beta July 22, grouping raw grounding queries into automatically ranked themes. Branded segmentation followed twelve days later.

Each feature attacks the same underlying problem from a different direction. Citation data arrived in usable volume and immediately became unreadable. Topic Insights compressed it against a hypothesis you supplied. Query Topics compressed it against patterns the data itself contained. Branded segmentation does not compress anything. It splits the record along a line that changes what the totals mean.

The infrastructure underneath is eighteen months deep. Microsoft introduced AI channel groups in August 2025 to separate referral traffic by engine. Their December 2025 research found AI-referred traffic had grown 155% over eight months while remaining below 1% of total visitors. By June 2026, the Web IQ announcement framed the whole stack as commercial infrastructure, with Microsoft stating that AI traffic is growing 8x faster than human traffic.

They are building measurement for an economy that barely registers in click volume but is compounding fast in influence.

Non-Branded Search Demand Is What AI Is Redistributing

Research from Fractl and Search Engine Land published July 2, 2026 puts a number on why the branded split matters. Across 1,010,848 high-volume keywords covering 379 brands in eight verticals, overall search demand fell 29% year over year. Fintech dropped 38%. The volume classified as non-branded, roughly 90% of the total, overlapped heavily with the informational content most exposed to AI displacement.

Branded query share held its value. Non-branded demand is what moved.

Set against Clarity's new split, the implication is uncomfortable. If your AI citations are mostly branded, you are reading your own resilience as growth. You are standing on the part of search that held while the bigger part shifted into AI answers. If your citations are mostly non-branded, you are capturing exactly the demand that Fractl's data says is being redistributed rather than destroyed.

One is a holding pattern. The other is a growth position.

Writesonic data published July 29 adds another layer. Across 16 million brand appearances tracked over 30 days, 40% of AI citations did not name the source brand anywhere in the answer. Perplexity's ghost rate hit 52%. Google AI Mode followed at 49%. The engine used the content. The brand was invisible. Ghost citations are overwhelmingly a non-branded query phenomenon: when someone types your name, the AI names you back. When someone types a category question, the AI takes your content and may never say who created it.

Google Is Drawing the Same Line on the Paid Side

Microsoft is not alone in splitting these apart. Google launched a native branded search toggle inside AI Max on June 1, 2026 after expanded intent targeting pulled branded and competitor queries into prospecting campaigns and corrupted attribution. Attributed Branded Searches, which Google links to a $31 sales gain per query, went to global availability in late June.

Two of the largest platforms in digital advertising have converged on the same structural conclusion: brand-name traffic and category traffic are different products. Reporting that merges them produces decisions built on an average of two unrelated things.

If both Microsoft and Google need to break these apart for their own analytics to make sense, the aggregate number you are currently watching is structurally misleading.

What to Measure Now

Open Microsoft Clarity. Go to the AI Visibility dashboard. Turn on the branded/non-branded filter.

Look at one number: what percentage of your AI citations come from queries that already contained your brand name?

If that number is above 70%, your AI visibility score is measuring your marketing budget. It is confirming demand you generated through advertising, PR, or direct brand awareness. It is not measuring whether AI engines find you when someone asks a category question without knowing you exist.

The non-branded citation rate is the number that tells you whether your content earns discovery. Whether AI engines select your page as source material for an answer about your category when nobody prompted them with your name. That is citation authority. That is the score that compounds.

Branded citations follow demand. Non-branded citations create it.

If your branded share is high and your non-branded share is near zero, you have a recognition asset and a discovery gap. The recognition holds as long as you keep spending to generate it. The discovery gap means every category question going to AI right now is sending buyers to whoever does show up.

That list does not include you.

FAQ

What is Microsoft Clarity's branded query segmentation for AI citations?

Shipped August 3, 2026, the feature splits the AI Citations dashboard into branded and non-branded views. Branded queries include your company name in the grounding query AI systems use to retrieve source material. Non-branded queries are category or topic queries where the AI found your content without being asked about you by name. The feature is free and live in the Clarity AI Visibility dashboard.

Why does the branded vs. non-branded AI citation split matter?

Aggregate citation counts merge recognition (someone searched your name and the AI confirmed it) with discovery (someone asked a category question and the AI chose your content). Those are different business outcomes with different investment implications. Fractl's July 2026 study found non-branded search volume absorbed the majority of the 29% year-over-year decline in search demand. Non-branded AI citations are where that redistributed demand lands.

How do I check my branded vs. non-branded AI citation split?

Log into Microsoft Clarity, go to the AI Visibility dashboard, open the Citations section, and apply the branded/non-branded filter. Compare your Share of Authority across both query types. If branded citations dominate, your AI visibility depends on existing brand awareness rather than content-driven discovery.

What is the connection between non-branded AI citations and Machine Relations?

Machine Relations is the discipline of earning AI citations through entity clarity, earned media authority, and citation architecture. Non-branded AI citations are the direct output of Machine Relations work: they measure whether AI engines select your content based on its quality and authority when no one prompted them with your name. Jaxon Parrott, founder of AuthorityTech, coined Machine Relations in 2024 to describe this shift from optimizing for human search rankings to earning machine-mediated discovery.