Demand Planning

Demand forecasting produces the number; demand planning decides what happens to it. A working cycle runs monthly or weekly: cleanse sales history, generate a statistical baseline, collect what sales, category and marketing know that the model cannot see, agree the adjustments, publish the plan, and measure the result against the baseline it started from. The plan then feeds sales and operations planning, where supply and finance commit to it.

• Baseline first: generate the statistical forecast before any human input, otherwise there is nothing to compare adjustments against
• Forecast value added (FVA): accuracy of the final plan minus accuracy of the untouched baseline – a negative FVA means the step is removing accuracy
• Naive benchmark: last period's demand repeated; a process that cannot beat it is not earning its cost
• Plan at the level decisions are made: replenishment needs store and day, not a national monthly total

Demand planning in practice

Two failure modes are worth checking for first, because they hide each other. The first is a demand plan that is a sales target in disguise: the number arrives from the budget commitment, so the error is structurally positive. MAPE conceals this – bias has to be measured separately, per category, or the plan keeps overstating demand for years while the reported accuracy looks acceptable. The second is the reflex to adjust everything. When planners touch nearly every item, the baseline is being overridden on items where it was not wrong, and only an FVA measurement separates the overrides that added accuracy from those that removed it. The fix is not to ban adjustments. It is to restrict them to items where a person holds information the model does not – a promotion, a listing change, a one‑off B2B order – and let the rest run untouched. That also frees planner time for the items that genuinely need judgement.

Veritico STOCK generates the baseline per SKU, store and warehouse, lets planners adjust the forecast and then confirm or freeze it, and turns the agreed plan directly into replenishment orders – see demand forecasting and inventory optimization and replenishment and allocation management. Mondelez raised forecast accuracy from 50 % to 70 % after unifying planning across functions, and Kofola reached 76 % accuracy on promotional forecasts.

Related terms

• Demand forecasting
• Sales and operations planning
• Replenishment
• Inventory optimization
• Forecast accuracy
• MAPE