Data coverage
Clean, structured, and reconciled.
This view shows what is captured, what is excluded, and which retailer price views are ready to compare.
Coverage snapshot
Catalog ceiling
The category tree reconciles to selected leaf pages.
Selected product-list categories are tracked separately from parent navigation pages and empty shelves.
Data dictionary
Field coverage is profiled across current datasets.
Each dataset is checked for the fields needed to power product, promotion, and comparison views.
Costco catalog audit
Canada and USA catalog rows are checked before comparison.
Canada and USA Costco rows are kept in separate price views so catalog, delivery, savings, and warehouse offers do not get mixed.
Empty categories
Selected leaf pages that returned zero product rows.
These pages were checked and returned no product rows during the crawl.
Promotion integrity
Savings, discount math, and date ranges reconcile.
Promotional rows are checked so savings and date ranges stay consistent.
Promotion source coverage
Offer rows reconcile by category, signal type, and source fields.
Promotion source coverage proves the captured promotion rows reconcile by category, source URL, product URL, currency, observed timestamp, and signal type. It does not claim every future flyer or store-specific promotion until another crawl observes it.
Promotion manager queue
Every branded offer has an action label.
Promotion manager queue contains only brand-actionable T&T promotional rows. No-brand promotion rows remain excluded, and open windows stay open until a later crawl proves an end date.
Product identity
SKU, slug, category, package signal, and variant coverage.
Product identity is source-backed from current T&T listing rows. Package and variant text are extracted only when visible in listing names; missing package text is counted, not inferred.
Brand filter audit
Raw rows reconcile to surfaced brand rows.
Readiness
What is ready and what still needs source proof.
Open proof gates
Useful now, with stronger claims held back.
These are the remaining places where more source evidence is needed before stronger price or comparison claims are shown.
Goal matrix
The original ask mapped to current proof.
This matrix separates ready work from claims that still need another source pass.
Price snapshot archive
Dated observations are preserved before trend labels change.
The first dated snapshot is preserved; true observed movement starts after another dated crawl.
Retailer match review queue
Near matches stay held until evidence is strong enough.
Near matches stay in review until brand, product text, package size, and price view agree.
Dataset map
Sources, outputs, and app surfaces.
Each artifact is mapped to its source layer and the pages or APIs that use it.
Live app check
Pages and data feeds are reachable.
Latest local route check from http://localhost:4117.
Data readiness
What is ready, partial, or waiting for stronger source evidence.
Source coverage
Every layer shows where it came from.
Checks
Data checks
Data contracts
Required fields by dataset
Source proof still needed