Citelens study finds Turkish and English AI answers cite different sources

Citelens’ analysis of 444 commercial queries found that only 22% of the domains cited by AI answers overlap between Turkish and English, suggesting language-specific visibility plays a major role in generative search. The findings matter for brands selling in Turkey because English-language AI visibility appears to tell little about performance in Turkish results.

Why it matters: - Brands that sell in Turkey or operate in multiple languages cannot assume AI visibility in English will translate into Turkish. - The study suggests Turkish and English AI answers function like two separate visibility markets. - That changes how brands should measure generative engine optimization, content authority, and local search performance.

What happened: - Turkish AI visibility platform Citelens analyzed 500 commercial queries in English and Turkish. - 444 queries returned AI answers in both languages and were used for the overlap analysis. - The research found only 22% overlap in the domains cited by Turkish and English AI answers. - English queries produced an AI Overview in 96% of cases. - Turkish queries produced an AI Overview in 94% of cases. - The full research and raw data were published at the full report.

The details: - The study compared citations in Google AI Overviews for the same commercial question asked in Turkish and English. - Citelens measured overlap using the average Jaccard similarity of cited domain sets across queries that returned AI answers in both languages. - The company fixed country and language for each query using gl=us, hl=en for English and gl=tr, hl=tr for Turkish. - Citelens said the method shows whether a brand is visible directly to Turkish audiences, not through an English-language proxy. - The platform tracks how often brands are cited in ChatGPT, Perplexity, Google AI Overviews, and other AI engines. - Citelens is developed by Solustiq Yazılım ve Yapay Zeka Teknolojileri A.Ş. in Edirne, Turkey.

Between the lines: - The low citation overlap points to different source ecosystems by language, with Turkish results leaning more on local content, local authority sites, and local community discussion. - For Turkish brands, that means the competitive set in AI search may be more local than global. - Citelens founder Alper Tekin said many brands assume AI works the same in every language, but the data shows otherwise. - Tekin said brands visible in English should not assume they will appear in Turkish AI answers.

What's next: - Brands will likely need separate AI visibility programs for Turkish and English rather than one global strategy. - Citelens positions its platform as a way to track where brands gain or lose citations across languages and markets. - The company’s approach could push more marketers to test AI search performance by market, not just by language.

The bottom line: - In AI search, language is not just a translation issue; it changes which sources get cited and which brands get seen.