Dataïads

Dataïads — Product-page-level AI-readiness corpus for e-commerce catalogues: for each scanned public product URL, a Readiness Score decomposed into a 2x3 matrix — two parallel readings (agent-agnostic and major-AI-specific) across three sub-scores: (1) Reach & Read (fetch success, robots.txt, sitemap, raw vs rendered HTML divergence, plus confirmation that Gemini/OpenAI actually read the URL); (2) Product Data (crawler-side Schema.org Product/Offer presence, and strict-JSON facts extracted independently on the Gemini and OpenAI sides with multi-provider confidence); (3) Semantic Decisioning (…

Dataïads offers (Alternative, Reference, Sentiment) — Unknown at absolute scale, but each scanned URL yields a rich multi-part record (raw + rendered HTML, crawler markup, two per-provider JSON extractions, six sub-scores), so a 1,000-URL sector sample is already a large capture corpus. The Baromètre E-commerce France 2026 implies a systematic multi-retailer cohort crawl — its sample size is the key number to request; not stated on the page..

Evaluation benchmarks for shopping agents and LLM retrieval over commerce content (which page attributes survive extraction); catalog-quality impact models on AI recommendation and visibility; automated remediation/trainable-optimisation data for product feeds; cross-model extraction-divergence datasets for model comparison and reliability testing; GEO/LLMO benchmarking feeds for retail brands; measurement of LLM-mediated traffic shift at sector level.

Coverage spans Other, US, UK; Internet & Direct Marketing Retail, Marketing; alternative, reference, sentiment; equities.