Demand forecasting is the process of predicting the demand of a stock item over an upcoming defined period of time. Forecasting is usually done by reviewing historical data e.g past consumption or sales data. This is then used along with knowledge of seasonality, market trends, and events such as promotions, to forecast the future values.
A combination of known and forecasted sales or production orders is usually used to represent demand, and the further ahead you look, the less certainty there often is over the accuracy of the forecast.
Whether you’re a manufacturer, wholesaler or retailer, demand forecasting plays a critical role in effective stock management. If you can accurately forecast demand, you can then take action to ensure you’re holding the correct amounts of stock to deliver high fulfilment rates, or service levels. Get your demand forecasting right and you’ll reduce the risk of stockouts, production disruption and/or lost sales.
Accurate inventory forecasting allows you to efficiently meet consumption requirements or fulfil customer orders, without investing in large amounts of unnecessary stock. This effectively helps you lower overall operational costs. Great forecasting enables you to hold the right amount of inventory without over or under-stocking, for optimum inventory control.
However, accurate inventory forecasting is no mean feat. In this post we uncover eight top demand forecasting techniques you can’t be without.
Demand forecasting techniques can be as basic or as complex as you make them. And there are many forecasting models a business can implement which can use both quantitative forecasting (using historical demand data) and qualitative forecasting (based on more subjective opinions and insights).
One of the simplest and most common inventory forecasting techniques is to calculate moving average forecasts. This is when you take a previous period’s demand data (e.g four week’s of sales data) and calculate the average demand over that period (average sales per week), then use this average as the forecast amount for the coming period.
Whilst this method is suitable for inventory items that have stable demand, and where previous actual demand is a good indicator of the future forecast period – in today’s disrupted marketplaces, this is often not the case.
Many businesses face issues with their simple demand forecasting techniques because they are not accounting for the wide range of demand patterns being experienced by their stock items. They also fail to account for external market factors that lead to demand volatility.
Let’s look at how you can include them in your forecasts!
Here are our eight top demand forecasting techniques to help you improve how you manage your inventory:
Let’s take a closer look at each in more detail:
If you analysed the historical sales/consumption data of every stock item in your warehouse, you’d find that the demand for different products varies considerably. Some will have consistently high or low demand over time, while others could have sporadic or erratic demand.
In addition, as stock items move through their product lifecycle, from market entry, to maturity and decline, their demand types will keep changing:
To forecast your base demand effectively you should therefore identify then use an item’s demand type to determine the type of calculation (or algorithm) you use to produce the most accurate forecast. It makes statistical sense to use a different algorithm to calculate the demand of a product with an erratic demand type, to one with slow demand.
The demand for your inventory items will eb and flow as fashions change, new technologies replace old and social, economic and legal factors influence demand.
Items will also follow demand trends as they move through the product cycle. For example, in the growth phase, the trend in demand will be upwards, whilst in the decline phase, the trend will reverse.
Make sure you look out for such trends in your demand data and adjust your inventory forecasts accordingly. There’s no point creating a forecast based solely on your base demand if items are following a specific trend.
Almost every manufacturer, distributor or retailer can expect to see seasonal demand fluctuations for some of their product lines. Seasonal weather patterns, school holidays and annual traditions all have a seasonal influence on demand.
Understanding how these seasonal factors affect your production levels, or your customers’ purchasing habits, will help you take advantage of peaks in demand and plan for the troughs.
Best practice is to keep seasonal demand factors separate from your base demand calculations. This keeps the data clean and easier to use for forecasting going forward.
Qualitative demand forecasting includes accounting for future events and external market factors, such as sales promotions and competitor activity. It essentially involves asking for ‘human input’. This could be asking for opinions from your sales team, or speaking to customers and suppliers about their thoughts on upcoming orders. Qualitative forecasting methods are especially useful when demand is erratic or unstable, or when you have no historical data to use, such as with new market entrants.
Make sure you input any human insights you have into your inventory forecasts to make them as accurate as possible.
Unusual demand outliers can be the result of known actions (sales promotions, large one-time orders, employee strikes etc) or unpredictable events (a competitor going out of business, natural disasters etc).
Take the time to analyse your inventory forecasting data to detect demand outliers, as they can significantly skew the accuracy of your forecasts. Any demand data – high or low – outside of the reasonable standard deviation of average demand needs to be identified. You then need to make a judgement call on whether it should be included in your demand forecasting calculations (if it’s part of a trend) or not (if it’s an anomaly in demand).
This also includes periods of stock-outs. Make sure you exclude these from your forecasts, or they will incorrectly bring them down overall.
For example, if you have a period where you only sold/used 10 of one item because that’s all you had in stock, whereas you could have actually sold/used 200 with the right availability, make sure you don’t reorder based on a forecast that’s looking at the lower number. Flag periods for exclusion, or even better, make an assumption about the sales/consumption you lost and add this number into the forecast.
Your demand forecasts are very unlikely to be 100% accurate. So, if you can calculate the level of error in your previous demand forecasts, you can factor this into future forecasts. If you can determine how uncertain a forecast is for a given business period you can make the necessary adjustments to your inventory management rules, such as increasing safety stock levels to cover uncertain periods of demand.
There are many formulas to help you measure demand forecast accuracy, or forecast error. The Mean Absolute Percent Error (MAPE) will calculate the mean percentage difference between your actual and forecasted demand over a given period. Whilst the Mean Absolute Deviation (MAD) shows the deviation of forecasted demand from actual demand in units. You can learn more about forecasting error here.
The time period you choose for your demand forecasting has a direct impact on the accuracy of your forecast. For example, a forecast looking at your inventory’s demand over the next two weeks is much more likely to be accurate than a forecast that looks 12 months out.
In addition, if markets are volatile, or an item’s demand pattern is erratic, you’ll need to review your forecasts on a much more regular basis than in slow markets or for slow moving products. If you begin to experience stock-outs or see cases of excess stock, then you may need to adjust your forecasting intervals.
Accurate demand forecasting is not a simple task. Especially if you want to track each stock item and you have a large portfolio. Inventory forecasting also requires an accurate picture of the stock levels in your warehouse and your future orders or sales across each channel.
Demand forecasting software offers a fast and accurate means of forecasting, no matter how complex or varying the demand. Whilst enterprise resource planning systems (ERP), warehouse management systems (WMS) and ecommerce platforms can offer a certain level of functionality, investing in a demand forecasting system will support more complex demand forecasting requirements.
Statistical demand forecasting systems, such as EazyStock, will ensure you have a tool to swiftly and accurately complete your complex demand forecasting requirements in order to reduce stock-outs, decrease cash tied-up in inventory and, most importantly, meet customer requirements.