Microsoft added automatic topic classification to the AI Citations experience inside Clarity on July 22, 2026, grouping raw grounding queries into a smaller set of generated themes and scoring each one on citation volume and share of authority. The change arrives thirteen days after the same product added competitive topic analysis, and it targets a problem publishers created for themselves the moment citation data became abundant.

The premise of the release is unusually plain for a product announcement. Citation queries have become numerous enough that reading them individually no longer works. According to Microsoft, as AI-powered search and chat experiences cite more websites as sources, individual citation queries can quickly grow into large, unstructured datasets, making it difficult to review and organise citation activity at scale.

That is a measurement problem produced by a measurement success. Two years ago, publishers had no visibility whatsoever into whether their pages were being pulled into AI-generated answers. Microsoft closed that gap in stages. Now the gap has inverted: the data exists, and it exists in volumes that defeat manual inspection.

What the feature does

Query Topics groups related grounding queries automatically, based on the citation activity attached to a given Clarity project. The classification is not drawn from a fixed taxonomy. According to Microsoft, a broad topic such as Sports may be broken down into more specific themes like Winter Skiing or other query patterns that emerge from a site's own AI citation data.

The distinction matters, and Microsoft draws it explicitly against its own sibling product. Bing Webmaster Tools recently introduced Topics, which organises AI citation queries into predefined topic categories. According to Microsoft, Query Topics in Clarity takes a more granular approach by generating groupings from the project's actual citation record rather than sorting queries into pre-existing buckets.

The example illustration published with the announcement makes the mechanic concrete. Four raw queries - best all-mountain skis, beginner ski boots, top snowboard brands, waterproof ski jackets - collapse into four topics with attached metrics. All-mountain skis carries 10.8 thousand citations at 24.1 percent share of authority. Ski boots shows 8.7 thousand citations at 27.5 percent. Snowboards records 4.3 thousand at 14.8 percent. Ski jackets sits at 1.2 thousand and 23.6 percent.

Read against each other, those four rows demonstrate why the two metrics are reported together. Ski boots generates fewer citations than all-mountain skis but a higher share of authority, meaning the site's content carries proportionally more weight within a smaller pool of queries. Snowboards produces roughly four times the citations of ski jackets while holding a materially lower share. Volume and authority move independently, and a dashboard reporting only one of them would mislead on the other.

The four components

Microsoft lists four changes shipping with the beta.

The topics card now appears directly inside the AI Citations dashboard, surfacing the main themes a domain is being cited for and allowing filtering into the underlying queries.

Topic-level performance metrics attach citation count and share of authority to each theme, which is what enables the volume-versus-authority comparison described above.

A topic list with deeper analysis allows filtering across a dedicated list of topics and comparison of share of authority alongside page-level citation distribution.

Export support permits downloading topic-level data for reporting or analysis outside the dashboard - a small addition on paper, though it is the one that determines whether the data can enter existing reporting workflows rather than remaining trapped in a Microsoft interface.

According to Microsoft, the dashboard now emphasises the metrics that best reflect visibility and authority while making large query sets easier to browse, filter, and export.

Two products, two questions

Query Topics does not stand alone. It sits alongside Topic Insights, which Clarity added on July 9, 2026, and Microsoft is careful to frame the pair as answering different questions rather than duplicating each other.

Topic Insights operates on topics a team defines in advance. It evaluates visibility and opportunity around those chosen themes, including where competitors are winning citations. Query Topics runs in the opposite direction: it reports the themes already driving citations, whether or not anyone thought to look for them. According to Microsoft, the two experiences together answer where a site is currently earning visibility in AI experiences and where the biggest opportunities to improve lie.

The distinction is between a hypothesis and an observation. A content team defines topics based on what it believes its coverage areas to be. The citation record may disagree - surfacing themes the team never targeted, or revealing that a flagship coverage area generates negligible grounding activity. Query Topics is the instrument that exposes that mismatch.

Topic Insights was itself notable for its pricing, arriving free to every Clarity user at a moment when competing AI visibility vendors reserve comparable competitive analysis for paying tiers. Query Topics follows the same distribution model, entering beta as part of the broader AI Visibility experience with, according to Microsoft, additional enhancements and expanded insights planned for upcoming releases.

An eighteen-month accumulation

The sequence behind this release runs longer than the feature itself suggests, and PPC Land has documented each stage.

The groundwork was behavioural rather than citation-based. Clarity introduced AI channel groups on August 29, 2025, separating referral traffic from ChatGPT, Claude, Gemini, Copilot, and Perplexity into distinct sources. That measured what happened after an AI system sent someone to a website. It said nothing about what happened before.

In December 2025, Clarity's research team published an analysis of eight months of platform traffic finding AI-referred traffic had grown 155 percent over the period while still accounting for under 1 percent of total visitors in the dataset. The same research proposed AI citation counts as an emerging metric category - a proposal that has since become product.

Bot Analytics arrived in January 2026, exposing crawler-level activity and path-request patterns. Microsoft later added a violations layer flagging robots.txt non-compliance, extending the picture from what bots requested to whether they respected stated preferences.

The citation-side tooling began at Bing rather than Clarity. The AI Performance dashboard launched in public preview in Bing Webmaster Tools on February 10, 2026, exposing total citation counts, average cited pages per day, grounding queries, and page-level citation activity. Microsoft acknowledged at launch that query data represented a sample rather than a complete record. Two days later, the company formally defined generative engine optimisation and positioned grounding as the infrastructure layer beneath most major AI assistants.

On March 23, 2026, Microsoft connected grounding queries to specific cited pages - the most-requested enhancement following the February launch, according to the company. Clarity's own Citations feature reached general availability on May 13, 2026, bringing grounding queries, cited pages, share of authority, and AI referral trends to every Clarity project, with domain verification available through either Bing Webmaster Tools or Google Search Console.

The June 17, 2026 Web IQ announcement placed Clarity's citation reporting inside a wider infrastructure argument. According to Microsoft, AI traffic is growing eight times faster than human traffic, while many businesses still block AI systems by default.

Each stage added a layer. None reduced the volume of what publishers had to read. Query Topics is the first release in the sequence whose stated purpose is compression rather than exposure.

Why the aggregation layer matters

The practical consequence is that AI visibility reporting can now be structured the way search reporting has been structured for two decades. Keyword-level data has always been aggregated into themes before it reaches anyone making decisions, because individual queries are too numerous and too sparse to interpret directly. Citation data has lacked that layer entirely.

Share of authority at topic level is the more consequential of the two metrics, because it is comparative in a way citation counts are not. A citation count says a domain was referenced. Share of authority says how much of the total citation weight for that theme the domain captured relative to every other cited source. Applied per topic, it distinguishes between broad but shallow presence and narrow but dominant presence - two positions that identical citation totals would obscure.

There are limits worth stating. Microsoft has said its citation query data represents a sample rather than a complete record, which means topic-level aggregates inherit whatever sampling characteristics the underlying queries carry. Automatic classification also involves editorial judgement executed by a system: the boundary between Winter Skiing and an adjacent theme is a modelling decision, and Microsoft has not published the classification methodology. Topics that a site considers distinct may merge; topics it considers unified may split.

The measurement question also remains partly unresolved on the commercial side. AI referral traffic, by Microsoft's own December 2025 numbers, sat below 1 percent of total visitors in its analysed dataset. Citation volume and share of authority describe presence inside AI answers, not what that presence is worth. The sequence of releases since February 2026 has made AI visibility increasingly legible without settling what it converts to.

For content teams, the immediate effect of Query Topics is narrower and more procedural: the citation record becomes something that can be filtered, compared, exported, and reported on periodically rather than inspected query by query. That is not a change in what Microsoft measures. It is a change in whether the measurement can be used.

Query Topics remains in beta. Microsoft has not published a general availability date.

Timeline

Summary

Who: Microsoft, through the Clarity analytics team, publishing via the Clarity blog under a Clarity Staff byline. The release affects publishers, brands, and content teams tracking how their pages are cited in AI-generated answers.

What: Query Topics, a beta feature that automatically groups AI citation grounding queries into generated topics, attaches citation count and share of authority to each topic, provides a filterable topic list with page-level citation distribution, and supports topic-level data export.

When: Announced July 22, 2026, thirteen days after the July 9 Topic Insights launch and roughly two months after Clarity Citations reached general availability on May 13, 2026.

Where: Inside the AI Citations dashboard in Microsoft Clarity, part of the broader AI Visibility experience, available in beta.

Why: Citation query datasets have grown large and unstructured enough that manual review is impractical. Topic-level aggregation gives citation reporting the summarisation layer keyword reporting has long had, allowing comparison of themes by volume and by authority rather than inspection of individual queries.