Inventory Forecasting Methods: How to Predict Demand

Inventory forecasting is the process of estimating future product demand so a business can decide what to buy, when to buy it, and how much stock to hold.

A forecast does not need to predict demand perfectly. Its job is to improve purchasing decisions compared with guesswork or blindly repeating the last order.

Why inventory forecasting matters

Forecasting supports decisions around reorder timing, purchase quantities, safety stock, supplier planning, cash flow, and location-level inventory. Weak forecasts can create both stockouts and excess inventory.

Common inventory forecasting methods

Moving average

A moving average uses demand from a fixed number of recent periods and averages them. It is simple and useful when demand is stable, but it can react slowly to major trend changes.

3-period moving average = (Period 1 + Period 2 + Period 3) ÷ 3

Weighted moving average

This method gives more importance to recent periods. It can respond faster to changing demand while still smoothing short-term noise.

Exponential smoothing

Exponential smoothing updates the forecast continuously and places more weight on recent observations. It is useful for ongoing operational forecasting when demand is relatively regular.

Seasonal forecasting

Seasonal models account for recurring patterns such as weekends, holidays, summer demand, winter demand, or annual events. Retailers and food businesses often need seasonality because a flat average can hide predictable peaks.

Trend forecasting

If demand is steadily increasing or decreasing, a trend-based method can be more useful than assuming the recent average will continue unchanged.

Qualitative forecasting

New products or businesses with little history may need judgment-based forecasts using manager experience, supplier information, market knowledge, promotions, and comparable products.

Which forecasting method should you use?

There is no single best method for every SKU. Stable products may work well with a simple moving average. Seasonal items need seasonal adjustments. New products may require judgment until enough data accumulates. High-impact SKUs deserve more careful modeling than low-value items.

Forecasting by product and location

Multi-location businesses should avoid combining unlike stores into one demand average. The same SKU can have different sales patterns at each location. Forecasting at the item-location level can improve replenishment and reduce unnecessary transfers or emergency orders.

Forecast demand vs. actual demand

Forecast accuracy should be measured. Track the difference between forecast and actual demand, identify where errors are largest, and adjust the method or assumptions. Forecasting is a feedback loop, not a one-time setup.

How forecasting connects to safety stock

Forecasts estimate expected demand. Safety stock protects against uncertainty around that expectation. A stronger forecast may reduce some uncertainty, but a buffer can still be necessary when demand or supplier timing varies.

How forecasting connects to reorder points

Traditional reorder points often use average demand. For fast-changing products, forecast demand during lead time can be more useful than a stale historical average. See the reorder point guide.

Restaurant and hospitality inventory forecasting

Restaurant forecasting should account for day-of-week patterns, reservations or expected traffic, menu changes, promotions, weather, events, and ingredient shelf life. A monthly product average can be too blunt for perishables. Forecast at the ingredient and location level where the purchasing decision depends on it, then compare predicted usage with actual depletion and waste.

Coffee shops and hospitality groups should also separate recurring patterns from one-off events. A festival weekend may justify a temporary uplift without becoming the baseline for future orders.

Forecasting seasonal products

For seasonal products, compare the same selling window across prior years, then adjust for current trends, promotions, store openings, price changes, and supplier constraints. Plan the exit as well as the peak: decide when to reduce orders, transfer stock, mark down items, or stop replenishing.

How to measure forecast accuracy

Choose a consistent error measure and review it by SKU and location. Mean absolute error is easy to interpret in units; mean absolute percentage error can help compare differently sized products but behaves poorly when actual demand is zero or very small. Bias is also important because a forecast that repeatedly runs high creates overstock while one that runs low creates stockouts.

The useful question is not whether the forecast was perfect. It is whether the error is understood and the next purchasing decision improves.

Demand forecasting vs. demand planning

Forecasting estimates what customers are likely to demand. Demand planning decides how the business will respond using inventory, suppliers, lead times, budgets, promotions, service targets, and constraints. The forecast is an input to the plan, not the finished purchasing decision.

Common forecasting mistakes

• Using one method for every SKU
• Ignoring seasonality
• Forecasting company-wide instead of by location
• Treating promotions as normal demand
• Failing to measure forecast error
• Using poor inventory or sales data
• Assuming forecasting removes all uncertainty

How Stash fits

Stash combines inventory visibility, forecasting, stock alerts, purchasing, suppliers, and multi-location operations. That helps teams move from historical sales data toward more structured replenishment decisions.

Related inventory guides

• demand planning vs. demand forecasting

Frequently asked questions

How much sales history do I need?

More history can help reveal patterns, but the right amount depends on the business. Recent data may matter more when demand is changing quickly, while a full year or more can help identify seasonality.

Is forecasting the same as replenishment?

No. Forecasting estimates future demand. Replenishment turns that estimate into inventory and purchasing decisions.

What is the best forecasting method for restaurants?

No single method fits every item. Use day-of-week and seasonal patterns for established products, adjust for known events and promotions, and apply judgment to new menu items with little history.

How often should an inventory forecast be updated?

Update it often enough to affect the next purchasing decision. Fast-moving perishables may need frequent review, while stable long-lead-time items can use a slower cadence.

Want to operationalize this process? See inventory forecasting software from Stash for demand, lead-time, stock, and purchasing review workflows.