AI Visibility Index — Phase 1 Dataset & Analysis Bundle: 1,000 Businesses Across Five Generative AI Search Systems

Companion dataset and analysis bundle for the Phase 1 audit reported in "Business Visibility in Generative AI Search" (Zenodo DOI 10.5281/zenodo.20048424).

Contents:

- anonymized-dataset.csv — per-business measurements for 1,000 businesses across 12 industries, including outcomes (mention status across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews) and 25+ on-page, off-page, and emerging-convention signals (schema markup, FAQ content, indexed page counts, Google review counts, llms.txt presence, AI crawler accessibility, and others). Identifiers anonymized as BIZ_xxxx; URLs and brand names removed.

- 10 JSON analysis outputs — descriptive statistics, raw and partial correlations, OLS-residual strict-isolation analysis, logistic regression coefficients, per-industry breakdowns, source-URL citation correlations, and predictive-model feature importance.

- README.md / codebook — column-by-column definitions, missingness notes, collection methodology, and reproduction instructions for the headline results.

Each business was queried across five AI search systems using curated intent prompts per industry; mentions were extracted via canonical-name string matching and aggregated to a 0–100 visibility score. On-page signals were measured via direct site crawl; off-page signals via Moz, Google Maps Places API, and a directory aggregator.

Phase 2 enrichment data (n=2,729 businesses across 14 verticals × 4 metropolitan markets) is held in the source database and will be released as a versioned update to this record.

Released under CC-BY-4.0. The author is the founder of MentionLayer, the generative engine optimization platform whose underlying methodology motivated this work; data and analysis outputs are released openly to enable independent verification.