Séve

Séve — Grocery basket intelligence generated as a loyalty-programme byproduct: OCR-extracted line-item receipts (store, register, timestamp, SKU, size, unit price, basket total) enriched with derived labels for shopping mission, consumption occasion, repeat-purchase sequence and competitive adjacency, all from opt-in shopper receipt submissions rather than surveys.. Small panel, unusually rich grain: 5-person company, no disclosed funding, but the dataset is verified transaction-level grocery purchase data with OCR line items and occasion labels, which is the premium tier of consumer panels. …

Séve offers (Alternative, Sentiment, Fundamental) — Receipt-grain rather than impression-grain: the worked example on the page shows a single $152.14 Whole Foods basket with 6 line items, so each submission yields roughly 5-15 structured rows. At a low-tens-of-thousands panel submitting monthly, the corpus is plausibly 10^5-10^6 line-item rows per year and 10^4-10^5 receipt images..

Shopper panel for CPG brand mix and share-of-basket modelling; retailer shelf-assortment and planogram arguments with verified adjacency evidence; promotional-lift and price-promotion elasticity measured on real baskets; SKU-level demand nowcasting ahead of syndicated panel releases; retail-media audience segments built from verified purchase behaviour; ground-truth purchase labels for evaluating consumer-behaviour models.

The data is with 2 years of history.

Coverage spans US; Food Distributors, Personal Care Products; alternative, sentiment, fundamental; equities.