Live data · scanned weekly

Viclaro Atlas

Public rankings of how ChatGPT, Claude, Gemini, and Perplexity recommend businesses in each category. 25 prompts × 4 models × 2 samples per scan run. 95% Wilson confidence intervals. Methodology versioned. Reproducible.

Live rankings

Reproducible

Every citation traces to a verbatim LLM response. Same prompts in, same numbers out. Public JSON API.

Honest CIs

95% Wilson intervals shown next to every rank. Long-tail firms flagged below threshold. No false precision.

Cross-LLM consensus

Per-model breakdowns surface which LLMs disagree — a firm cited only by Perplexity is a different signal than one cited by all four.

How it works

For each (city, category), we send 25 buyer-style questions to ChatGPT, Claude, Gemini, and Perplexity — twice each, with different sessions — for a 200-slot scan run. We parse which businesses each model cites, fuzzy-match against a pre-harvested firm database, and rank by recommendation share.

Prompts are grouped into 5 intent buckets (broad authority, specific need, process, style/approach, conversational). Per-bucket counts surface which buyer questions each firm wins on. Methodology is versioned — changes increment the version and snapshots from different versions are not directly comparable.

Full methodology →