This screen compares each store's stock as the till system sees it against as the ERP sees it, item by item, from snapshots taken minutes apart each morning — so a day's trading can't create false differences.
The three levels:
- Stores — every store listed, worst first; clean stores show as ties, so absence from the problem list means checked-and-clean, not unchecked.
- The lines — for a store, each item where the two systems disagree, with the quantity each side holds.
- The why — for a line, the product looks back over recent weeks of movements on both sides and attributes the difference to its cause where the data supports it: a count that reached one system but not the other, a sale missing on one side, a one-sided transfer. Each explained line carries a suggested action (a correcting entry, a store task, or a recount). The same inputs always produce the same verdict — it's rules, not judgement, so it's auditable.
Differences the data can't explain are labelled as needing a count — honestly unexplained, never guessed.