Demand forecasting software: questions
The questions we are asked most often about demand forecasting software, answered without the marketing layer.
What data does demand forecasting software need?
Transaction-level sales history, stock-out records and the promotion calendar. The second is the most commonly skipped and the most expensive to skip: when a product was not on the shelf, the sales data reads as low demand, and demand forecasting software trained on it mistakes your inventory error for customer behaviour.
Does demand planning software replace our existing ERP?
No, it sits on top of it. The ERP records what happened; demand planning software predicts what will happen and recommends accordingly. Confusing the two means mistaking a system of record for a system of decision — systems of record are designed to keep the past accurate, not the future.
What does demand forecasting software need from us?
Sales history at transaction level, stock-out records, and a promotional calendar. The second is the most commonly skipped and the most damaging to skip: if an item was unavailable, the sales data reads as low demand, and a model trained on it learns your stock errors as customer behaviour.
How far ahead does it forecast?
Two horizons, because they serve different decisions. A statistical forecast covers capacity and purchasing months out; a sensing layer reacts inside the lead time, where allocation is still open. Running the first and calling it the second is the common mistake.
Our forecast accuracy is already good. Is this needed?
Possibly not — but check what the number is measured at. A forecast that is accurate at category-month can be useless at SKU-week, because errors cancel as you aggregate. If accuracy is measured at a level nobody makes decisions at, it is describing something other than the business.
Related: Inventory Intelligence · Performance Intelligence