How AI Search Works
1
Attention Is All You Need: The Google Paper That Accidentally Ended Google's Search Monopoly.
On June 12, 2017, eight researchers at Google Brain published a 15-page paper. It accidentally made Google's own search business obsolete.
2
How Large Language Models Actually Work: Tokens, Context Windows, and Why Your Content Gets Ignored.
The answer to "how do I get cited by AI search?" is technical, not strategic. It's rooted in how large language models actually process information.
3
What Is RAG: the Only Thing That Matters for AI Search.
Once you see how Retrieval-Augmented Generation works, the rest of AI search visibility clicks into place. It's the mechanism that connects your content to every answer these systems produce.
4
The AI Search Race: ChatGPT, Gemini, Claude, Copilot, and Perplexity. A Technical Timeline.
The models, the milestones, and the architectural decisions that shaped every AI search product in market. A reference for anyone trying to make sense of where this came from.
5
How Google AI Overviews and AI Mode Actually Work: Why Your SEO Rankings Don't Guarantee AI Visibility.
AI Overviews aren't a ranking bonus layered on top of existing positions. In May 2024, Google restructured how most people receive information. The blue links became supplementary.
6
How to Be Surfaced in AI Search: What the Platforms' Own Documentation Actually Says.
Most GEO content produces a list of tactics and implies they work across all platforms. This article goes directly to what each platform has actually published about how its own system works.
7
8
The GEO Documentation Gap: Why Three Platforms Haven't Told You How to Rank in Their Systems.
Google published a guide. Microsoft published measurement tooling. Anthropic, OpenAI, and Perplexity have published nothing equivalent. Understanding why is more useful than waiting.
9
Constitutional AI vs RLHF: Why the Alignment Method Affects What Gets Cited.
ChatGPT is aligned with RLHF. Claude is aligned with Constitutional AI. These aren't interchangeable, and the difference affects what each model treats as trustworthy and citable.
10
Training Data Is the New Ranking Factor: The Common Crawl AI Visibility Audit Explained.
Before RAG, before indexing, a site has to be reachable by the crawlers that feed AI training data. The Common Crawl AI Visibility Audit is the framework for checking that upstream layer.
11
Blocking AI Training Can Cost You Search: Cloudflare's September Defaults Explained.
Cloudflare split AI bots into Search, Agent and Training. Multi-purpose crawlers resolve to the most restrictive rule, so a training block can take your search visibility with it.
12
Google Finally Reports AI Visibility. It Still Won't Tell You What It's Worth.
Search Console's Generative AI report gives you impressions and nothing else. No clicks, no queries, and no split between AI Overviews and AI Mode. Here's how to use it honestly.
13
The AI Search Statistics Everyone Repeats, and Where They Actually Came From.
Gartner's 25% prediction rests on a survey of 299 people. The 14% AI Mode figure isn't Google's. AI traffic converts 3x, not 23x. The audit, and the method for running it yourself.
14
97% of llms.txt Files Are Never Read. Here's the Data.
Ahrefs studied 137,000 domains and found 97% of valid llms.txt files received zero requests. No major AI provider has adopted it, and Google says you don't need one.
15
Schema Markup Doesn't Drive AI Citations. It Still Matters.
1,885 pages that added schema, against 4,000 matched controls, tested with difference-in-differences. The effect on AI citations was nil. Keep your schema anyway, and here's why.
16
The AI Browser Died. Now Agents Browse as Your Reader.
OpenAI shut down Atlas and moved agentic browsing into a Chrome extension. The winning model runs the agent inside your reader's session, using their logins, which breaks how most people think about blocking bots.
Writing  ·  Series: How AI Search Works  ·  Part 7 of 10

Why the Bing AI Performance Report Is the Most Underused Tool in GEO.

Microsoft's Bing AI Performance report is the only first-party citation dashboard from any AI search platform. Most people working in GEO aren't using it. Here's what it shows, what it doesn't, and how to turn it into a content strategy signal.

Published2 June 2026
Updated5 September 2026
ByThomas Cox
Read time10 minutes
Filed underSeries · GEO Measurement · Part 7 of 10
Correction · 5 September 2026

When I wrote this, I treated the Bing AI Performance report as a window into AI citation behaviour generally. That was too broad. Search Engine Land's August 2026 teardown of ChatGPT's retrieval stack analysed 1,200 answers, 88,000 results and 26,900 distinct pages, and found that only 1.5% of the URLs ChatGPT retrieves appear in Bing's top 20 for the same queries. ChatGPT runs on OpenAI's own retrieval hub, not on Bing's index.

So read what follows as a Copilot report, because that's what it is and that's what it's good at. It's still the most detailed first-party citation data any platform publishes, and the grounding queries are still the most useful thing in it. It just isn't a proxy for ChatGPT visibility, and I shouldn't have implied it was.

Everyone working in GEO in 2026 is dealing with the same measurement problem: first-party data on how AI systems use your content is thin, and what exists counts outcomes rather than explaining them. Google Search Console added a Generative AI performance report in June 2026, rolled out to all sites by 31 August, and it gives you impressions only: no clicks, no CTR, no position, no queries, and no split between AI Overviews and AI Mode. You can see that you were surfaced. You cannot see what you were surfaced for.

Microsoft has changed this for its own platform. The Bing AI Performance report, launched in public preview in February 2026, is still the only first-party tool that shows you which queries actually caused a citation, which is precisely the part Google withholds. If you work anywhere near Copilot, that's the most detailed citation data anyone publishes and it's sitting there unused. It's also why it sits in a different category to the estimate-based trackers: I compare the two in Bing AI Performance vs the third-party AI trackers.

What the report shows.

The AI Performance report is accessible via Bing Webmaster Tools (bing.com/webmasters) for any verified site. The metrics it provides:

Microsoft Clarity users who have linked their Clarity workspace to Bing Webmaster Tools also get the AI citation data surfaced inside Clarity's reporting interface, useful if Clarity is already your primary analytics view.

The grounding queries: the most valuable data point.

The grounding queries data is the most strategically valuable element of the report, and the most commonly misunderstood.

When a user asks Microsoft Copilot a question, Copilot does not send the user's verbatim question to Bing for retrieval. It generates a set of distilled search terms (the grounding queries) that the system determines will retrieve relevant information. These grounding queries are shorter, more keyword-like than the user's original conversational prompt.

The grounding queries visible in the AI Performance report are the actual retrieval signals that caused your content to be cited. They show you what Copilot was looking for when it found your pages, which is often different from what you expected.

Practically: if a grounding query that consistently cites your content is "enterprise data security compliance framework 2025" but your page is titled "Our Approach to Data," the grounding query is telling you that Copilot is finding your content for a specific informational need that your title and metadata do not clearly signal. That is an optimisation signal, not just for Copilot, but for the semantic clarity of your content more broadly.

What the report does not show.

Microsoft's documentation is honest about the report's limitations. Be equally honest with clients and stakeholders.

How to use the report strategically.

The five steps below are the workflow I use when onboarding a client to the report. They're ordered by impact per hour.

1
Verify in Bing Webmaster Tools
Verification is free (XML file, meta tag, or DNS record) and is the prerequisite for Bing indexation monitoring, the baseline for Copilot web citation eligibility.
Tool: bing.com/webmasters
2
Map cited pages to grounding queries
Sort grounding queries by volume and map each to the page it most frequently cites. This shows what Copilot perceives each page as authoritative about, which may not match what the page is optimised for.
3
Identify citation gaps
Look for grounding queries where Copilot retrieves from other sources. These indicate either content gaps (the content doesn't exist) or structural gaps (it exists but isn't structured clearly enough to be retrieved).
4
Use citation trends to detect freshness issues
A declining citation trend for a previously well-cited page often means more recently updated content has displaced it. A freshness pass (updated data, new examples, refreshed dates) can recover lost citation frequency.
5
Feed grounding queries into content strategy
Grounding queries show what AI systems believe users are asking about in your domain, a research signal independent of traditional keyword data. Use them to find content opportunities keyword research won't surface.

The broader point.

The Bing AI Performance report is a first-party measurement tool from one platform, covering one ecosystem. It is limited. But it is the most direct empirical data available anywhere on how AI systems actually use web content.

More importantly: it demonstrates what platform-level citation measurement can look like. When I first published this I said that if Microsoft had built it, Google would follow. That took about four months, and Search Console's Generative AI performance report arrived in June 2026. What Google shipped is impressions and nothing else, which tells you plenty about how much of this data the platforms are willing to hand over.

In the meantime, using the data that exists beats waiting for the data you wish existed.

Using the data that exists beats waiting for the data you wish existed.

The Bing AI Performance report is limited to one platform, doesn't show click data, and uses sampled queries. But it's the only first-party citation measurement tool that exists from any AI search platform. Use what you have.

Frequently asked

Where do I find Bing AI search reporting?

It's the AI Performance report inside Bing Webmaster Tools at bing.com/webmasters, available for any verified site. That report is Bing's AI search reporting: it shows how often your pages were cited as sources in Copilot's AI-generated answers, and the grounding queries that retrieved them.

Do Bing and Yahoo share ranking reports?

Yahoo's web search has run on Bing's index for years, so the pages Bing indexes are the same ones that surface on Yahoo. There's no separate Yahoo reporting tool to chase: verify in Bing Webmaster Tools and the indexation and citation data covers both surfaces. Bing is where the reporting lives.

What is the Bing AI Performance report?

It's the only first-party citation dashboard any AI search platform currently offers, launched in public preview in February 2026. It shows total citations, average cited pages, page-level citation activity, grounding queries, and citation trends over time for your verified site.

Is Bing Webmaster Tools free?

Yes, completely. Verification is free (XML file, meta tag, or DNS record), and the AI Performance report is included. It's a large part of why the report is underused: people assume first-party AI citation data must be paid or gated, and it isn't.

What are grounding queries?

When someone asks Copilot a question, Copilot doesn't search your verbatim prompt. It generates distilled search terms, the grounding queries, that it determines will retrieve relevant content. The report shows the grounding queries that cited your pages, which tells you what Copilot thinks each page is authoritative about.

Does Bing reporting cover Copilot and Edge?

Yes. The report covers Microsoft Copilot, Bing's AI-generated summaries, and the Edge integration, plus select partner integrations. It does not cover Google AI Overviews, ChatGPT, or Perplexity, so treat it as one platform's view, not a cross-platform picture.

Next in the series

Article 8 asks why Anthropic, OpenAI, and Perplexity haven't published any equivalent guidance, and what that silence means for how to approach them. Read Article 8 →

SOURCES ↓
  1. Microsoft Bing, "Introducing AI Performance in Bing Webmaster Tools," February 2026 - blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview
  2. Microsoft Learn, "Microsoft 365 Copilot Architecture and How It Works" - learn.microsoft.com/en-us/microsoft-365/copilot/microsoft-365-copilot-architecture
  3. Microsoft Learn, "Semantic Indexing for Microsoft 365 Copilot" - learn.microsoft.com/en-us/microsoftsearch/semantic-index-for-copilot
  4. Microsoft Learn, "Use Public Websites to Improve Generative Answers" - learn.microsoft.com/en-us/microsoft-copilot-studio/guidance/generative-ai-public-websites
  5. Microsoft, "15 Milestones That Shaped Microsoft's Vision for AI," 2025 - news.microsoft.com/source/features/ai/15-milestones-that-shaped-microsofts-vision-for-ai
  6. Search Engine Land, "Inside ChatGPT's retrieval stack: The index, cache, and pages it actually reads," Olivier de Segonzac, August 2026 - searchengineland.com/chatgpt-retrieval-stack-index-cache-pages
  7. Google Search Console Help, "Generative AI performance report (Search)" - support.google.com/webmasters/answer/16984139

Let's talk about your visibility.

Invisible in AI answers, slipping in search, or both. Thirty minutes is enough to work out where to start.

Book a call →