We Measured 6,933 AI Citations. Your Homepage Isn't What Gets You Cited.
Published October 4, 2026 · Adam Rock
The short version
We measured 6,933 citations (a snapshot taken 30 Sep 2026) across five AI engines and 10 monitored brands over seven weeks. The sources AI cites most are platforms and publishers - YouTube, Reddit, Forbes, TechRadar - and not one brand's own website made the top five.
That is the whole finding. Being described by AI and being read by AI are different things, and the difference is whether someone other than you has already published the claim.
What we measured
- Universe: 10 monitored brands, 71 monitored queries.
- Engines: ChatGPT, Gemini, Google AI Overviews, Perplexity, Microsoft Copilot (the five that returned data).
- Window: 13 Aug 2026 to 30 Sep 2026.
- Measure: citations the engines made to a publication (classified info_site), grouped by registered domain.
- Sample: 6,933 cited sources (as of 30 Sep 2026).
- Excluded: our own domains and test domains.
The full dataset, method box and the SQL we used are published at /data/ai-citation-source-domains-2026-09/. The published total is the 30 Sep 2026 snapshot; citation rows are periodically reprocessed, so a live re-run returns a comparable but not identical figure.
The finding
Top 12 source domains of 6,933 total citations (snapshot taken 30 Sep 2026).
| Source domain | Citations | Engines |
|---|---|---|
| youtube.com | 510 | 4 |
| reddit.com | 383 | 3 |
| forbes.com | 127 | 3 |
| techradar.com | 110 | 3 |
| linkedin.com | 55 | 3 |
| zapier.com | 50 | 5 |
| whathifi.com | 48 | 3 |
| bankrate.com | 44 | 3 |
| finance.yahoo.com | 44 | 1 |
| pcmag.com | 43 | 2 |
| blog.hubspot.com | 39 | 4 |
| nerdwallet.com | 39 | 2 |
YouTube and Reddit lead every engine we measured. Winning AI visibility is often testing and discussion, not marketing copy.
Why your homepage isn't what gets cited
An AI answer is assembled from sources that already corroborate a claim in public. Your homepage asserts; a third-party review, a Reddit thread, a YouTube test or a Forbes roundup evidences. When an engine has to choose what to stand behind, it leans on the source that is not the interested party.
That is why a brand can rank and still be invisible in AI answers: the engine is not quoting you, it is quoting the people who checked you.
What to do about it
- Get named in the third-party roundups and reviews that already rank for your category. That is the evidence layer engines reuse.
- Build a YouTube and Reddit footprint - the two domains at the top. Testing, comparisons and honest discussion get cited; press releases rarely do.
- Earn media. A citation is not an endorsement, but an unaffiliated mention is what makes a claim quotable.
- Make your own pages machine-readable - schema, clean semantic HTML, a real last-updated date. That is what lets an engine quote you cleanly once it does cite you.
- Track it per engine. Coverage differs by engine; a single blended score hides where you are actually missing.
The limits
- This is our measurement, not a survey of the AI industry.
- The monitored brands are a mix of internal, pilot and demo accounts. The AI answers are real; the brand set is not a representative sample of all categories.
- A citation is not an endorsement, and a citation is not a recommendation.
- Counts are citations to a domain, not unique pages.
- The window is roughly seven weeks, not a year.
The dataset
The numbers above, the method box and the SQL we used are published as a first-party dataset: /data/ai-citation-source-domains-2026-09/.