Free Demand Forecast Excel Template
Build statistical forecasts from history — moving average, weighted moving average, exponential smoothing, with seasonal adjustment. Compare methods on the same data.
What you get
Working demand forecast template with 4 forecast methods, forecast error metrics (MAPE, MAD, bias), and seasonal index calculation. Pick the right method for each SKU based on data.
Free 30-day trial · No credit card required · Used by manufacturers since 1991
Why manufacturers still use Excel for this
Most shops "forecast" demand by extrapolating last quarter and adjusting by gut. The result is forecasts that are biased optimistic (sales' instinct), inconsistent across SKUs, and impossible to improve because nobody measures forecast error.
Statistical forecasting is not complicated math — but it is consistent math. The template provides 4 forecast methods (simple moving average, weighted moving average, single exponential smoothing, double exponential smoothing for trend) and runs each against the same historical data. Forecast error metrics (MAPE, MAD, bias) tell you which method fits each SKU best.
For most manufacturing demand, exponential smoothing wins. For seasonal products (consumer goods, food), seasonal-adjusted forecasts beat naïve methods substantially. The template handles both; the discipline of measuring forecast error is what turns "forecast" from gut guess to managed input.
What's inside the template
Demand history input
Up to 36 months of historical demand per SKU. Weekly or monthly granularity.
Simple moving average
3-month, 6-month, and 12-month moving averages. Stable but slow to react.
Weighted moving average
Recent months weighted higher. Faster reaction to trend than simple MA.
Single exponential smoothing
Smoothing constant α adjustable (typically 0.1–0.3). Better at handling level shifts than moving averages.
Double exponential smoothing
Holt method — handles trend (gradual rise or decline) explicitly. The right choice for growing or declining demand patterns.
Seasonal index
For seasonal products, calculate monthly seasonal index from historical data. Apply to base forecast for seasonal-adjusted forecast.
Forecast accuracy metrics
MAPE (Mean Absolute Percent Error), MAD (Mean Absolute Deviation), bias. Compare methods on the same data.
How to use this template
A practical walkthrough — five steps from blank spreadsheet to a working schedule.
- 1
Use enough history
Minimum 12 months of data; 24+ months for seasonal products. Forecasting from 6 months of history is dart-throwing.
- 2
Pick the method that minimizes MAPE per SKU
Different SKUs respond to different methods. Stable demand: moving average. Trending demand: double exponential smoothing. Seasonal demand: seasonal-adjusted. Test each method on each SKU and pick the winner.
- 3
Track forecast error over time
Update MAPE and bias monthly. Bias drifting positive = forecasts systematically too high. Bias drifting negative = systematically too low. Both correctable.
- 4
Override statistical forecasts with explicit reasons
Sales sometimes knows things history does not (lost a customer, won a contract). Overrides are allowed; they must be documented with reason. Tracking override accuracy separately surfaces whether sales' instinct beats the math (rarely does).
When you outgrow this template
Excel is the right answer for early-stage scheduling — until it isn't. Here are the warning signs that you need a real production scheduling tool.
If three or more of these apply, you have outgrown Excel scheduling. The good news: you do not have to leave Excel behind. Resource Manager for Excel (RMX) is a real finite-capacity scheduling engine that runs as an Excel add-in — so your team keeps the interface they know while gaining the scheduling power of a dedicated APS tool.
Learn about RMXFrequently asked questions
What is MAPE and what is "good" MAPE?+
MAPE = Mean Absolute Percent Error = average of |actual - forecast| / actual. 10% is excellent for most manufacturing demand. 20% is OK. 30%+ means the method is not fitting and you should try another. Intermittent or low-volume SKUs may run higher MAPE — they are statistically harder.
What does "forecast bias" mean?+
Bias = average of (forecast - actual). Bias near zero = unbiased forecasts. Bias positive = forecasts systematically too high (over-ordering, excess inventory). Bias negative = forecasts systematically too low (stockouts, expediting). Bias is the most actionable forecast error metric.
How do I handle intermittent demand?+
Intermittent demand (frequent zero periods, occasional spikes) does not fit standard smoothing methods. Use Croston's method or simply use safety-stock-based logic rather than forecast-based logic. Template includes a basic Croston implementation for these SKUs.
How does seasonality affect forecast?+
Seasonality is the predictable monthly pattern around the base trend. Calculate the seasonal index from history (each month's average vs annual average), then apply: base forecast × seasonal index = seasonally-adjusted forecast. Done well, this drops MAPE by 30–50% on seasonal products.
Get the free template — plus the tool that grew up around it
The template is the starting point. Resource Manager for Excel (RMX) is what manufacturers move to when their Excel scheduler starts breaking. 35+ years in production, free 30-day trial.