01Notes · 27 August 2026
Microsoft Copilot: The Surface Nobody Measures
Quick answer
Copilot has a smaller audience than the Google surfaces and a better position: it is embedded in Microsoft 365 and Windows, which means it is already open on the desk of the enterprise buyer you are trying to reach.
It grounds on the Bing index, and Bing publishes citation data that almost nobody looks at. That combination — short causal chain, free data, no competition for attention — makes it the most under-worked surface in the category.
Key takeaways
- Placement beats reach for enterprise. Copilot is inside the tool your buyer already has open.
- It grounds on Bing, which makes Bingbot access and Bing indexation direct inputs rather than proxies.
- Bing Webmaster Tools ships AI Performance datacitations, cited pages and grounding queries, in public preview since February 2026.
- That data is free and mostly unopened. Teams verify in Search Console and never check Bing at all.
- Check you are properly indexed in Bing first. A surprising number of sites are not.
- The causal chain is shorter here than on any other assistant, which makes it a good place to learn.
Placement beats reach
Every comparison of AI assistants leads with user numbers, and on that measure Copilot loses to Google’s surfaces by a wide margin.
For enterprise selling, user numbers are the wrong measure. Copilot is embedded in Microsoft 365 and Windows. The procurement lead evaluating your category does not open it deliberately. It is already there, in the application they are working in, at the moment the question occurs to them.
That is a different kind of advantage from audience size, and it is the reason this surface deserves measurement in enterprise categories even though it will never top a usage chart.
The Bing connection, and why it simplifies things
Copilot grounds on the Bing index. That produces something rare in this category: a short, inspectable causal chain.
| Question | Most assistants | Copilot |
|---|---|---|
| What corpus does it draw on? | Proprietary, undisclosed | The Bing index |
| Can I check I am in it? | No | Yes, in Bing Webmaster Tools |
| Which crawler matters? | Several, some undocumented | Bingbot |
| Is citation data published? | No | Yes — AI Performance, public preview since Feb 2026 |
| Can I see grounding queries? | No | Yes |
| What does it cost to check? | Sampling time or a tool subscription | Nothing |
Six rows where Copilot is the only assistant that answers plainly. That is the argument for looking at it, whatever its user numbers say.
The data almost nobody opens
Bing Webmaster Tools has shipped AI Performance reporting since February 2026 in public preview. It shows citations, which of your pages get cited, and the grounding queries behind them.
In our experience most teams have a Bing Webmaster Tools account created years ago, verified once, and never opened since. Everything is set up for Google, verified in Search Console, and Bing is left as a box that was ticked.
Meanwhile the one AI surface that publishes citation data for free is sitting inside it.
A twenty-minute check
- Confirm Bing indexation. Open Bing Webmaster Tools and check your page count against what you expect. Discrepancies here explain a lot.
- Check Bingbot is allowed in robots.txt. Verify it rather than assume it — blocked crawlers are the most common finding in any audit.
- Open AI Performance. See which pages get cited and what queries ground them.
- Compare against your Google picture. If the cited pages differ, that is a finding, not noise: the two most similar platforms shared only 24.1% of cited pages in 2026 research.
- Run five buyer prompts in Copilot itself, three times each, and record whether you appear.
Twenty minutes, no tools, no spend. If you sell to enterprise it is the highest return-per-minute check available in this category.
What this means for you
- Open Bing Webmaster Tools this week. The account probably already exists.
- Verify Bing indexation separately from Google. Nobody does, and it is often where the gap is.
- Weight Copilot by buyer, not by audience. In enterprise categories it is worth more than its user numbers suggest.
02Common questions
Questions this raises
Why does Copilot matter if its audience is smaller?
Because of where it sits. Copilot is embedded in Microsoft 365 and Windows, which means it is already open on the desk of the enterprise buyer you are trying to reach. Reach is not the same as placement.
What index does Copilot use?
It grounds on Bing. That makes Bingbot access and Bing indexation a direct input to Copilot visibility, which is a simpler causal chain than most AI surfaces offer.
Can I actually see Copilot citation data?
Closer than on any other assistant. Bing Webmaster Tools shipped AI Performance reporting in public preview in February 2026, covering citations, cited pages and grounding queries. It is free and most teams have never opened it.
Is my site even in the Bing index?
Worth checking, because a surprising number are not properly. Teams optimise for Google, verify in Search Console, and never confirm Bing coverage at all.
Does optimising for Bing differ from optimising for Google?
Less than it used to. The main practical difference is that Bing indexation has to be verified separately, and Bingbot has to be allowed separately.
Where should Copilot sit in our priorities?
Below the Google surfaces on reach, and above them on effort-to-insight if you sell to enterprise. The data is free and the causal chain is short.
Sources
Every figure on this page comes from one of the following. Where a claim is ours rather than a source’s, it carries an evidence label instead.
- Bing Webmaster Tools — AI Performance reporting, public preview announced February 2026, covering citations, cited pages and grounding queries. Bing — Elevating the Role of Grounding on the AI Web, February 2026. Primary
- Rankability — State of AI Search, 1,645 citation observations across 31 topics, 2026. Source for the 24.1% cross-platform overlap and structural citation signals. Independent study
- Searlixa sampling across client engagements, labelled Observed with sample and date. Our own
Ask an assistant
Hand this page to an assistant and ask it to summarise or challenge the argument. If it gets something wrong about us, that is itself a useful finding.
→Next step