When Microsoft Clarity made AI citations available to all users, it opened up a new playground for SEOs to harvest AI visibility data. Finally, we can see the exact “grounding queries” an AI engine uses to pull our content.
It raises a massive question because this is a Microsoft tool: Are the insights useless if your audience doesn’t touch the Bing ecosystem?
Microsoft Clarity Grounding Queries
When you ask Copilot a question, it translates your words into simple search terms called grounding queries to find facts on the web before it answers. You can use this data to improve your own website and content.
- Finding gaps where your content does not match what the AI searches for.
- Simplifying pages that the AI reads but does not link to.
- Using these simple layouts to help your Google search results.
Copilot Vs. Gemini
Both Copilot and Gemini use retrieval-augmented approaches. Instead of generating answers using only pre-trained parameters, they dynamically query external search indexes to retrieve real-time data, which they then use as context to ground their final responses.
| Feature | Microsoft Copilot | Google Gemini |
| Structure | Uses a query translator, Bing index search, and OpenAI models to write the final text. | Uses a query translator, Google Search, and Google’s Gemini models to write the final text. |
| Pulling Sources | Uses the Bing index and Microsoft Graph to scan web pages, emails, and Microsoft 365 files. (With permissions enabled) | Uses Google Search and Google Workspace to scan web pages, Google Drive files, and Gmail. (With permissions enabled) |
| Synthesising Answers | Focuses on direct answers. It uses structured lists, tables, and bullet points to show facts quickly. | Focuses on creative, conversational answers. It is built to handle text, images, and code at the same time. |
Does Ranking In Bing Matter?
Yes (Correlation).
One of my websites was doing extremely well in Copilot, with over 36,000 citations across all queries. Now, Clarity doesn’t give you the prompts/queries themselves, but it does give you the Grounding queries (grounding queries and key phrases used to retrieve your site’s content).
My website has a history, running for years with a previous domain merged in 2019, and boasts over 1,000 articles. Given that Google barely sends traffic, and third-party SEO tools often label it as spam due to non-English backlinks (it covers search engines like Baidu, CocCoc, SwissCows, attracting an international audience), I never expected 36,000 citations.
So, why the Copilot love? I took the 147 grounding queries and tracked their rank in Google and Bing.
Of the 147 queries, Bing ranked all but 6, with the majority in traffic-driving positions (top 20). Google didn’t rank a single one.
So, If This Is Heavily Dependent On Bing Indexing, Is Clarity’s Data Useful Outside Of The Bing/Microsoft Ecosystem?
Because this is a Microsoft tool, the backend data feeding this dashboard is primarily capturing how your site is cited across Microsoft’s AI surfaces (like Copilot and Bing generative search).
It is not giving you a direct window into how OpenAI’s ChatGPT (using its own search), Google Gemini, or Perplexity are citing your links, because those platforms do not share their internal grounding logs with Microsoft.
And historically, we as an industry have been neglectful of Bing.
Even though the data collection source is skewed toward Microsoft’s AI engine, the insights themselves are highly transferable to your broader, platform-agnostic AI optimization strategies.
Can We Assume Other LLMs Retrieve Data In The Same Way?
AI engines, whether Google Gemini or Microsoft Copilot, use similar RAG frameworks to fetch data.
If the Bing ecosystem flags that a specific page on your site has a high “Share of Authority” for a complex query, it means that page is structured perfectly for AI consumption (clear tables, bullet points, direct answers). Data suggests that you can replicate that formatting across your site to appeal to Google Gemini as well.
However, this can be argued against as other research suggests that the similarity between LLMs is dependent on positional biases, and some may use the SDSR method rather than RAG.
Researchers in SEO have also found that ChatGPT has started to use Google Search as a fallback, when it was initially Bing.
In Summary
If your audience doesn’t touch the Microsoft ecosystem, this dashboard won’t give you a perfect 1-to-1 reflection of your total AI traffic, but it doesn’t make the data useless.
What grounding queries reveal is how AI systems distill user intent into retrievable search terms. That process is broadly consistent across platforms, even when the underlying indexes differ. A page earning citations in Copilot is doing something right structurally with clear answers, well-scoped topics, content aligned with how AI engines translate questions into queries. The Bing dependency tells you where the data comes from. The structural patterns tell you something more transferable.
The gap data is equally instructive. Pages your site ranks for in Bing that never appear as grounding queries signal a mismatch. Either the content isn’t structured for AI retrieval, or the topic isn’t one AI engines are actively grounding answers around.
Treat Clarity’s Citations dashboard as a useful proxy or “lab environment” and window into how LLMs read, slice/chunk, and credit your website’s content. Even if Copilot isn’t your primary AI traffic source, the patterns it surfaces are worth paying attention to.
More Resources:
Featured Image: Prostock-studio/Shutterstock
