One assumes the level holds; the other assumes the direction continues
The moving-average forecast takes the mean of the three months and applies your growth rate. It is stable and ignores month-to-month noise, which is what you want when demand is essentially flat with random variation.
Forecast = recent average × (1 + growth rate) (simple moving-average method)
The trend forecast extends the direction the three months are moving in. It responds faster to a genuine change and overreacts to a one-off. Seeing both is the point: when they agree, the forecast is reasonably safe, and when they diverge sharply, the recent months contain something worth understanding before ordering against either.
Enough to catch a trend, not enough to see a season
Three periods cannot distinguish a genuine upward trend from the approach of a seasonal peak. A forecast built on August, September and October will extrapolate straight through a Christmas that the data has never seen.
Where seasonality is material, the more reliable comparison is the same months a year earlier, adjusted for overall growth. Use this tool for stable categories and short horizons, and treat its output on a seasonal line as a starting point only.
Everything downstream inherits whatever you enter here
The growth percentage is not derived from the data — it is your assumption about the future, and the forecast is only as good as it. An optimistic rate produces an optimistic order, and the stock sits in a warehouse being wrong for months.
Running the forecast at three rates — flat, your expectation, and half of it — brackets the outcome. Ordering against the conservative end and reordering into demand costs less than the reverse.
What to do with the number once you have it
The forecast feeds the reorder point and the order quantity; it does not decide them. Lead time, safety stock and the economic order quantity all sit between this figure and an actual purchase order.
Track how wrong it turns out to be. A forecast consistently 20% high is more useful than one with no track record, because a known bias can be corrected — and reviewing that gap is what makes the next forecast better.
Sources & References
Figures on this page are checked against primary, authoritative sources. Links open in a new tab.
What the calculation assumes
- The three months entered are representative and free of one-off distortions like a stockout or a flash sale.
- Growth is assumed to continue at the rate you enter rather than accelerating or reversing.
- Demand is treated as roughly stable plus a trend, with no explicit seasonal pattern.
- Results are for general planning and educational use.
Where it stops
- A simple moving average cannot model seasonality, promotions, or external shocks that real forecasting must handle.
- Only three months of history are used, so the forecast reacts slowly and can miss turning points.
- It assumes a constant growth rate, which rarely holds for long in practice.
- For high-stakes planning, use a method that incorporates seasonality and more data.