Observatory · Engine
Which sources does Microsoft Copilot cite on buying questions?
Microsoft Copilot takes 48.5% of its cited sources from rankings and comparison pages on buying questions, the highest share of the four engines tracked via Profound. Data: 35,017 answers collected between July 12 and October 11, 2026.
Updated
90%
answers citing sources
3.6
sources per answer
49%
rankings among sources
52%
first-cited brand changes week to week
| Month | Sources per answer | Rankings among sources | YouTube |
|---|---|---|---|
| July 2026 (partial month) | 3.6 | 48% | 0% |
| August 2026 | 3.6 | 47% | 0% |
| September 2026 | 3.5 | 50% | 0% |
| October 2026 (partial month) | 3.9 | 49% | 0% |
The Observatory tracked Microsoft Copilot on 35,017 answers to buying questions, in two sectors only: B2B software and tech (28,007 answers) and marketing agencies (7,010 answers). That is Copilot's main audience, enterprise users on Microsoft 365, but it also means its figures do not describe consumer sectors.
Copilot writes the longest answers of the four engines where length is measured, with a median of 437 words, but cites few sources: 3.6 per answer, with at least one source in 89.7% of answers. Its mix is split almost evenly between brand sites, media and blogs (48.9%) and rankings and comparisons (48.5%). Reviews and directories weigh 2.6%; YouTube, LinkedIn and Wikipedia are at 0.0%.
That reliance on rankings is stable: 48.3% in July, 46.7% in August, 50.0% in September and 48.8% in October. It peaks at 57.8% on marketing agency questions and 45.7% on B2B software questions.
Copilot is also the most stable of the four engines where stability is measured: the brand it cites first changes from one week to the next in 51.6% of cases, against 64.3% for ChatGPT and 65.5% for Google AI Overviews. Once you hold a position in Copilot, you tend to keep it longer. See our Copilot optimization service.
Method
406,626 AI answers collected every day from July 12, 2026 to October 11, 2026 on over 1,700 purchase-intent questions tracked by Archipel (mostly in French, plus the United States and the Netherlands). ChatGPT, Google AI Overviews, Perplexity and Copilot: Profound data. Claude and Gemini: Peec AI data. Sources are classified automatically from their address and title. Figures are aggregated: no brand or client is published. Free to reuse with a link to this page.
FAQ
Frequently asked questions
Which sources does Microsoft Copilot cite on buying questions?
How many sources does Copilot cite per answer?
Is Copilot more stable than ChatGPT?
Which sectors does the Copilot data cover?
Shall we get started?
Free GEO audit: your rank on ChatGPT, Claude and Google AI Overviews, your competitors and your first actions. Reply within half a business day.