Stock accuracy, forecast inputs, store-to-DC replenishment signals, supplier lead times, and shrinkage attribution — where retail inventory data fails and how to govern it.
Wrong stock figures produce wrong orders: stockouts on one line, excess on another, shrinkage buried in a group total until stocktake. The fixes are specific — authoritative quantity by location, clean history for forecasting, timely signals into replenishment, and variance codes that tie loss to store or cause.
Retail Data Strategy covers the full picture. Product and POS data must be sane; inventory inherits every gap.
Movement without capture — Receipt to backroom to shelf to sale to return: any step handled informally leaves the system wrong. Small retailers often skip cycle counts; large ones may trust the DC while store stock is stale.
Forecast history poisoned — Promotional weeks treated as base demand, or stockouts recorded as zero demand, distort the next order quantity.
Multi-echelon blind spots — Store order depends on DC stock, in-transit, lead time, and MOQ. If store sales feed replenishment on a lag or DC availability is wrong, the optimiser works on fiction.
Shrinkage rolled up — Group shrinkage hides which store or category bleeds. Without reason codes and someone accountable for investigating variance, operations only sees a journal line.
Per location (and channel if relevant): who owns accuracy, how often physical count reconciles to system, what happens when variance exceeds a threshold. POS, WMS, and finance will disagree until one record is declared primary for replenishment.
SKU must match product master definitions or “units on hand” is not comparable.
Sales history at the right grain, promotion dates stripped or flagged, stockout periods not read as zero demand, new and delisted products handled so they do not pollute the series. Someone owns the calendar and sign-off on overrides — not only the model.
For warehouse-led retail: store sales rate, store on-hand, DC on-hand, in-transit, supplier orders. Latency matters — overnight batch to replenishment while the shop sells today creates systematic error. VMI requires clarity on whose system is authoritative.
Stale lead times distort safety stock. Short-ships not recorded against the vendor repeat the same miss on the next buy. Ties back to supplier data.
Useful shrinkage is split by cause and location, not one lump sum. That needs count discipline, variance codes, and ownership when the gap is material.
Promotions baked into “normal” demand — over-order, then markdown.
Batch replenishment while stores move stock in real time — system optimises yesterday.
DC accurate, stores fuzzy — seasonal peaks hit before the full count exposes the hole.
Shrinkage booked by finance only — operations never gets a store- or category-level view.
| Decision | Why it matters |
|---|---|
| Authoritative stock record per location | One replenishment signal, not three |
| Owner for forecast inputs | Stops unmanaged calendars poisoning the series |
| Documented store–DC data latency | Automation expectations match reality |
| Shrinkage reason codes and review cadence | Write-offs become operational signals |
| Alignment with product master | Same SKU everywhere |
Retailers with heavy DC and transport operations should align retail–DC handoffs with how logistics firms govern warehouse and shipment data — same class of boundary problem.
Customer and demand signals interact with forecasting when segments or channels shift mix.