For most of the past year, optimising for AI search has felt like flying blind. You could see that ChatGPT, Gemini, Perplexity and Google's AI surfaces were answering questions about your market, and you could guess you were sometimes cited and often not, but you had almost no instrumentation. You were tuning an engine by ear. That is starting to change, and the first useful dashboard is arriving from a slightly unexpected place: Bing Webmaster Tools.

Bing is rolling out four new AI visibility reports, in preview globally. The SEO consultant Aleyda Solís, who flagged the rollout, put the significance plainly: the Bing team is starting to give the search community "the AI search data needed to better understand and optimise for AI search." Bing powers Copilot and a meaningful slice of the wider AI answer ecosystem, so this is real signal, not a vanity metric. Here is what each report actually shows, and what to do with it.

Intents: why you were cited, not just that you were

The first report groups the queries that triggered a citation into broad intent categories: Informational, Commercial, Navigational, Learn and Solve, Research, Creation and Local. Instead of a flat list of keywords, you get the shape of the demand you are showing up for.

That distinction matters more than it first appears. Being cited on an Informational or Research intent tells you the model trusts you as a reference. Being cited on a Commercial intent, the queries closer to a purchase decision, is the one that pays the rent. If your citations cluster entirely in the informational bucket and disappear the moment intent turns commercial, you have a very specific, very fixable gap: the model knows you as an explainer but not as a vendor.

Topics: the themes a model files you under

The second report clusters related queries into thematic groups rather than treating each one as an isolated keyword. This mirrors how modern AI systems actually work. They reason across concepts and entities, not exact-match strings, so seeing your visibility organised by topic is a far truer picture than a keyword export ever was.

Read it as a map of the territory a model associates with your brand. Where the themes match what you sell, good. Where there are obvious adjacent topics you are absent from, that is your content and entity roadmap writing itself.

Citation Share: the number that finally lets you keep score

This is the one to care about. Citation Share shows the percentage of citations attributed to your site out of all the citations shown for a given query. For the first time you have a competitive denominator. Not "were we mentioned," but "of all the sources the answer drew on, how much of that space was ours."

Stop asking whether you were cited. Start asking what share of the citation space you own, and who is taking the rest.

A low citation share on a query you should dominate is a direct, prioritised instruction: someone else is being treated as the better source, and you can go and find out who and why. It turns a vague anxiety about AI search into a metric you can move and report on.

Compare: trend, not snapshot

The fourth report lets you overlay a previous time period onto the current view. Modest on its own, important in aggregate, because a single reading of any of the above tells you almost nothing. Citation share that is quietly eroding month over month is a very different story from one that is climbing, and only the trend distinguishes them. Treat these reports as a time series from day one and baseline now, while the data is new.

What this does and does not change

A few honest caveats, because the temptation is to over-read a new dashboard:

  • It is Bing, not the whole picture. This is genuinely useful coverage of Bing and Copilot's answer surface. It does not report what ChatGPT, Gemini, Perplexity or Google's AI Overviews are doing, so it is one instrument on the panel, not the panel.
  • It is in preview. Definitions, categories and numbers may shift as it matures. Useful for direction and relative comparison today; treat absolute figures with the usual caution.
  • It measures the outcome, not the cause. The reports tell you where you stand. They do not tell you why, or which content, entity and citation gaps to close to improve it. That diagnosis is still the work.

None of that diminishes the headline. The thing that made AI search so frustrating to optimise for, the absence of measurement, is finally cracking open. The teams that win the next two years will be the ones that start keeping score now, while most of their competitors are still arguing about whether any of this matters.

How we put it to work

Our AI Visibility Audit reads signals like Bing's new citation share alongside how your brand actually shows up across ChatGPT, Gemini, Perplexity and Google's AI surfaces, then tells you where competitors are being cited instead of you and which entity, content and citation gaps to close first. The new reports give you the scoreboard. The audit tells you how to change the score.