Demand Planning & Forecasting: Short Term Demand Forecasting
to the Forecast Solutions website
Training Courses

Forecasting Software

Improvement Plans

Forecasting Service
Contact us about Demand Forecasting Pricing Models

Weather Effects



Short Term Sales Forecasting Methods

Short term demand forecasting usually needs to be carried out at a detailed level for the purpose of stock planning, sometimes also account planning.  The high volatility of the data at this level means that sophisticated methods for forecasting are unnecessary and may be counter-productive in terms of forecast accuracy.  In the short term it is most practical to assume that external factors such as the prevailing competitive structure and economic background will stay relatively constant.

The emphasis is on tracking short term changes in demand. Simple time series forecasting methods such as moving average and the family of exponential smoothing methods are most commonly used. In a highly seasonal business,  finding the best way of dealing with seasonality is important. If promotions are a feature of the business then effective promotional planning is vital in terms of cleansing the historical data prior to forecasting and building the future promotional plan into the forecast.

Causal modelling (cause and effect analysis) can sometimes be useful for short term forecasting. It aims to quantify the effect on sales of causal factors such as unseasonal weather or the often crucial matter of pricing.  Forecast Solutions can provide expert help in this area with weather sensitivity analysis and price modelling.

Small scale systems for short-term forecasting based on time series forecasting and/or causal modelling can often be conveniently set up in Excel - see Forecasting in Excel.

Weekly Demand Forecasting

Forecasting in smaller time buckets does not necessarily make the short term forecast more accurate.  Historical sales data in weekly time buckets is necessarily more volatile than in monthly data and much more likely to be intermittent in nature. Seasonal analysis becomes more difficult for a number of reasons including the greater likelihood that events will fall in different week numbers in different years. It becomes more important to check the validity of the seasonal indices and a variety of methods can be used to reduce the volatility that often remains after an initial analysis of seasonality has been carried out. It is often useful to use seasonal profiles developed at a product group level (the 'group seasonal indices' method) and other methods of smoothing can be employed. See the page on seasonality for more information.

However, practical considerations often to make it desirable to forecast in weekly time buckets rather than months.  For example, it may be necessary to integrate a customer forecast that has been supplied in weekly buckets, and it may be easier to fulfil the needs of promotional analysis and planning.

Daily Sales Forecasting

With daily sales forecasting even lower volumes per time period, high volatility and intermittent sales are usually seen. There is usually a strong day within week cycle that dominates and overwhelms other influences. Statistical forecasting carried out on daily data with multiple cycles such as day within week and week within year seasonality tends not to succeed. There is often a strong case for creating the short range forecast at the weekly level, incorporating weekly seasonality, then splitting weeks to days using daily profiles. 

web stats