Original post by Dmitrij Żatuchin
30-second rundown
Key learnings:
- Third parties hold the microphone: Most AI citations about brands in this study came from third-party websites rather than brand-owned pages.
- A small group carries the load: Identify the recurring sources before chasing hundreds of minor mentions, because a relatively small group of domains produced most citations.
- Every market has its own informants: Check each target country and language separately, because local publishers, video sites, and careers portals can matter more than global sources.
This week: Ask three important brand questions in each AI platform and market you care about. Record every cited domain, count which sources recur, and flag inaccurate or missing information on the ten most common third-party sites.
Coming 3 months: Repeat the audit by language and platform, then coordinate SEO, PR, communications, and local teams around the small set of high-frequency sources. Keep the company site accurate, improve legitimate reference material, and earn credible local coverage. Treat changes as directional evidence, not guaranteed causation.
Dmitrij Żatuchin followed 167,551 AI citations across 128 brands, 12 markets, and 13 languages.
The trail did not lead back to corporate headquarters. It led into the open web, where strangers, encyclopedias, video platforms, and local newspapers testify about your brand while your own site sweats under a dead ceiling fan.
1. Third Parties Hold the Microphone
The study found that 85.7 percent of URL citations pointed to third-party sites, while 14.3 percent pointed to brand-owned domains. The pattern held across business and consumer brands, although the exact rates varied. Your site supplies necessary facts, but AI systems usually want an outside witness before they repeat the story.
2. A Small Head of Sources Carries the Load
About 18 percent of domains generated 80 percent of all citations, and half of the citations came from only 547 hosts. Wikipedia led in 11 of 12 languages, with YouTube, Statista, Reddit, and market-specific publishers also prominent. Scattered mentions in the long tail may feel productive, but the machines keep returning to a smaller circle of familiar saloons.
3. Every Market Has Its Own Informants
Lithuania’s leading source was the national business daily vz.lt rather than Wikipedia. For 46 Polish brands, YouTube led overall, while four recruitment and careers sites produced 637 citations compared with 297 for Polish Wikipedia. An English-only citation plan can miss the local sources shaping the answer in the very market you are trying to win.
This is a measurement study, not proof that a citation causes a recommendation or reflects human reputation. The datasets also used related but not identical collection methods, and model behavior will change.
Even with those limits, the strategic signal is loud enough to wake the dead: audit the sources AI actually reads, market by market, and earn your place among them.
