Public datasets for AI search visibility research. Free to use, cite, and reproduce.
AI Citation Behavior Dataset (2026 Q3)
CSV · JSONL
20 brands × 50 prompts × 3 platforms. Each row records whether the brand was mentioned, the citation sources returned, and the brand’s rank position in the answer — collected under the governed IndexHot measurement standard.
3,000 answer recordsChatGPT · Perplexity · GeminiLicense: CC BY 4.0
AI mention rates aggregated by industry and platform, covering both branded and non-branded prompts. Use it to benchmark how visible a typical brand in your sector is across generative engines.
Industry-level aggregatesBranded & non-branded splitLicense: CC BY 4.0
Results from a controlled experiment across 50 pages where structured data was added and removed. Includes citation and absorption outcomes by schema type, so you can see which markup actually moves AI citations.
50 test pagesBefore / after comparisonLicense: CC BY 4.0