Loading dataβ¦
Solid = actual complete weeks Β· faded last point = partial week (excluded from model fitting) Β· dashed = next 8 weeks. Switch methodology below β the demand plan table follows your selection.
Each method forecasts the last 4 complete weeks of the current view using only prior data. Lower MAPE = better fit. β = recommended.
Demand for the current view using the selected method.
Week-by-week breakdown of the current view. WoW compares each week to the previous one; * marks the partial trailing week.
Last 4 complete weeks vs the prior 4, within the current view. Minimum 20 units recent volume to qualify.
Velocity is units per active week since launch; 30-day projection uses launch-adjusted run-rate.
Default order: Flat Back SKUs AβZ, then Hoops AβZ. Net Req = max(0, demand β live stock β inbound FBA); Plan Units round the Net Req up to box multiples (150s first, remainder in 50s) β so the boxes shown are only what's actually missing. UPCs fill automatically from an optional upc.csv (columns: sku,upc) placed next to data.csv β blank otherwise so the team can complete them.
Shipped units by purchase weekday β useful for ad scheduling & replenishment timing.
Shipped units by SKU category (all data).
Share of shipped units in current view (top 8).
Units sent to FBA by week the shipment was created. Bars follow the active category view above.
Units sent per month with month-over-month direction.
Week-by-week outbound with WoW change. Weeks with no shipments are skipped.
Lead time = days from shipment creation to Amazon check-in. In Transit = not yet checked in.
Most replenished SKUs in the current view, all time.
Orange line = shipped sales units, blue bars = units sent to FBA, over the overlapping data window. When the line runs above the bars for several weeks, stock is drawing down.
Per SKU: forecast demand (selected method & horizon on the Sales tab) vs real inventory position = live Stock + Inbound FBA. Cover = position Γ· weekly forecast. Shortfall converts to your 150/50 box plan. Outbound in-transit is not added on top of inbound to avoid double counting Amazon's inbound figure.