Answer-engine response corpus plus derived visibility panel: for each prompt, the verbatim answer text (text/markdown/HTML) with every parsed citation (position, URL, label, description), the engine's own fan-out sub-queries and shopping/places blocks, and country plus US state-level targeting — collected across up to seven surfaces (ChatGPT with web search, Perplexity, Gemini, AI Overviews, Google AI Mode, Microsoft Copilot, Grok) alongside classic Google Search and Google News results, then aggregated into mention rate, citation position, share of voice versus named competitors, cross-engine coverage and entity-recognition accuracy
Equities in US, UK. Related tickers GOOGL, PERP, SHOP, MCD, AMZN, UBER.
Structured JSON records (unified response shape: answer text, markdown, parsed source objects with position/url/label, country) delivered by REST API, async webhook callbacks, TypeScript/Node SDK and a public MCP server; customer-side loads target BigQuery/dbt/Looker, so the byproduct is warehouse-ready tabular/nested JSON rather than reports or screenshots
~1 billion API calls per month across eight to nine surfaces. On the monitored-prompt product specifically, the company's own pricing arithmetic implies roughly 8,000 cross-engine prompts per month on the ~$100 Hobby plan (250k credits) and ~1.56M credits on the ~$500 Growth plan; multiplied by 250+ paying accounts and a public leaderboard snapshotted weekly across 12 categories, the accumulated prompt-response-citation corpus is in the tens of millions of parsed answers per month
Per prompt x engine x geography: whether the target domain was named (mention), whether it was cited as a source and at which rank position, the competitor set named in the same answer, the engine's own decomposed sub-queries, and the trust set of source domains the engine chose — i.e. human-free relevance/judgement labels on which sources each assistant relies on. The weekly public leaderboard adds curated brand-level labels across 12 software categories with historical snapshots
cloro — Answer-engine response corpus plus derived visibility panel: for each prompt, the verbatim answer text (text/markdown/HTML) with every parsed citation (position, URL, label, description), the engine's own fan-out sub-queries and shopping/places blocks, and country plus US state-level targeting — collected across up to seven surfaces (ChatGPT with web search, Perplexity, Gemini, AI Overviews, Google AI Mode, Microsoft Copilot, Grok) alongside classic Google Search and Google News results, then aggregated into mention rate, citation position, share of voice versus named competitors, cro…
Not yet on file: Sample · Data dictionary · Data frequency · Reporting lag · Point-in-time · License