How do you track brand mentions in Microsoft Copilot? Run a fixed list of buyer prompts in Copilot, then log whether your brand is named, which URLs are cited, and which competitors appear. Keep that log separate from ChatGPT. Same question, different answer, often.
Copilot sits inside Microsoft's products and on the web at copilot.microsoft.com. Enterprise tenants can see a different Copilot than a personal account. Write down which one you checked. "We checked Copilot" is not a method.
What to log
| Field | Why |
|---|---|
| Account type | Personal web Copilot, or a work tenant. They are not interchangeable |
| Prompt text | Must match the other engines if you want a comparison |
| Mention | Name in the answer |
| Citation | A source URL on your domain |
| Competitor | Who got the recommendation |
| Date | So a later run is comparable |
Bloomiro can run Copilot as part of an AI presence check, alongside ChatGPT, Gemini, Perplexity, Google AI Overviews, and Google AI Mode. Use the same prompt group so the comparison is fair.
Prompts that are worth the row
- How do I track brand mentions in Copilot?
- Does Microsoft Copilot recommend my competitor for [category]?
- How can I see what Copilot says about my business?
- Which sources does Copilot cite for [problem you solve]?
These are warmer than a generic "AI visibility tools" query. People ask them when they already have a Microsoft stack and they want to know if Copilot sends buyers somewhere else.
How Copilot differs from the ChatGPT column
ChatGPT and Copilot can both name a brand with no source list you can act on. When Copilot does show links, treat those links like Perplexity sources: the cited domain is the asset, the brand mention is the outcome. If Copilot cites a Microsoft Learn page, a marketplace listing, or a news site, your blog is not the only lever.
Read where AI answers get their sources before you commission five new articles.
If the same prompt is fine in ChatGPT and empty in Copilot, do not "average" them into one score. A blended score hides the engine you are losing. Definitions: AI visibility metrics. The ChatGPT-only method is how to track ChatGPT brand mentions.
What to change after a loss
- Confirm the run completed. A failed Copilot fetch is not a lost mention.
- Open the cited URL. Note the sentence Copilot could have quoted.
- If you have a page on that question, tighten the first answer. If you do not, write one only when you know something the cited page does not.
- Re-run the same prompt two to four weeks later. Cadence: how often to re-check.
Search Console still matters. Copilot will not tell you that Google already shows your site at position 15 for the same wording. That page is usually the one to edit.
