Demand Forecasting That Isn't Wishful: A Practical How-To
Most demand forecasts go wrong before any math happens. Someone decides what the number should be — the target the business needs, the figure that keeps the boss calm — and then the spreadsheet is steered toward it. That is not forecasting; it is wishful thinking with decimal places. A real forecast is a disciplined estimate of what customers will actually buy, stated with an honest sense of how wrong it might be, and useful precisely because it is not the same thing as a goal.
The 2022 environment punishes wishful forecasts harder than usual. With inflation reshaping buying behaviour, lead times stretched, and supply still volatile, a forecast that quietly assumes last year's patterns will repeat sets up either expensive overstock or empty shelves at the worst possible moment. The point of forecasting is not to be right to the unit; it is to be roughly right and clear about the range, so procurement and operations can plan around it.
Build the forecast in layers
A forecast you can defend is built up deliberately, not pulled out whole. Work through these steps and you will know which part of the number to trust and which part to hedge.
Start from clean demand history. Use actual demand, not shipments, and strip out the one-offs — a stockout that hid real demand, a one-time bulk order, a promotion that pulled sales forward. A history full of unexplained spikes will teach the model the wrong lessons.
Find the underlying pattern. Separate the baseline level, any trend, and any seasonality. Even a simple method — a moving average or exponential smoothing with a seasonal factor — beats a flat guess when the pattern is real and stable.
Layer in what the past cannot know. Add the things history will never show you: a planned price increase, a competitor exiting, a new customer onboarding, a product approaching end of life. This is where sales and account knowledge earns its place, applied as explicit adjustments you can point to later.
State the range, not just the point. Give a likely figure plus a high and low. The width of that range is information — a tight range invites lean inventory, a wide one tells procurement to keep options and safety stock open.
Reconcile top-down and bottom-up. Check the sum of item-level forecasts against the business-level expectation. When they disagree, the gap usually exposes a hidden assumption worth arguing about before it becomes a purchase order.
Measure the forecast so it can improve
A forecast nobody scores never gets better. Track accuracy with a simple, consistent metric and review it on a regular cadence.
Watch bias separately from error: a forecast that is consistently too high (or too low) signals a process or incentive problem, not bad luck.
Use a measure like MAPE or a tracking signal so the conversation is about numbers, not opinions.
Hold a short, blameless monthly review where planning, sales, and procurement reconcile what they each expect.
Forecast at the level you actually decide on — by item and location if that is how you buy and stock, not just a comforting company total.
Write down the assumptions behind each adjustment so that when reality differs, you can learn instead of guess again.
The discipline that separates a useful forecast from a hopeful one is the willingness to keep them apart from targets. The sales target can be ambitious; the forecast must stay honest. When the two are forced to be the same number, inventory and cash pay for the optimism. Keeping them distinct lets the business chase a stretch goal while the supply chain plans for the demand it actually expects.
Turning the forecast into decisions
A forecast only matters when it changes what you order, when, and how much buffer you carry. Tie it directly to reorder points, safety stock, and supplier lead times, and revisit it whenever a major assumption shifts rather than only on the calendar. In a volatile year, the team that updates its forecast quickly and acts on the range — not just the midpoint — is the one that avoids both the panic buy and the write-off.
If your forecasts and your purchasing keep drifting apart, XNM's procurement, sourcing & contract management can help you connect a realistic demand picture to the way you actually buy.