85.7% of what AI says about a brand comes from sites the brand doesn't own
Source structure changes by market: for Polish brands the top source is YouTube, and Lithuanian is the one language where Wikipedia isn't first.
Key numbers
- Sample: 128 brands across 12 home markets and 13 languages — 167,551 URL-grounded citations (189,974 attribution rows).
- 85.7% of citations point to third-party sites the brand does not own; owned properties account for just 14.3%.
- Sources are highly concentrated: about 18% of domains supply 80% of citations, fitting a Zipf distribution (α = 0.86, R² = 0.983).
- Wikipedia is the most-cited domain in 11 of 12 languages, with Lithuanian the exception. Market-level differences are larger: for 46 Polish national brands the top source is YouTube, and recruiting portals supply 637 citations versus 297 for Polish Wikipedia.
Method and sample
arXiv paper “How Large Language Models Source Brand Reputation Across Languages and Markets” (2026-06-24): three merged datasets covering 128 brands, 12 home markets and 13 languages; 167,551 URL-grounded citations classified by domain and source type, analysed per language and per market.
What it means for brands
What AI says about you is mostly built from what third parties say — and which third parties matters changes by market. One English content set pushed globally will miss the sources that actually decide each market.
How SourceSight handles it
SourceSight runs one project per target market, generates prompts in that market's language, and builds the off-site list from that market's platforms — Chinese projects never get overseas platforms, German projects never get Chinese communities.
Limits
The sample skews to Europe and North America plus parts of Central and Eastern Europe; Chinese engines are not covered. Shares are citation counts, not the probability that a user sees the brand.
References
Run it on your own project
What the same prompts return across 10 engines is only knowable by running them.
Related research
What Chinese generative search actually cites: 8.3% brand selection, and app vs web cite different sources
Across 160,860 citation-level records, the same platform surfaced systematically different source sets on its app and web interfaces.
Sources · 2026-10-06Incumbent advantage: with identical specs, a well-known brand is recommended 100% of the time
Three experiments on skincare: identical specifications produce near-total dominance for known brands — and less than a +0.1-star edge for a competitor breaks it.