sales forecasting

Sales forecasting Time-series analysisclick for a larger image
Fig. 77 Sales forecasting Time-series analysis.

sales forecasting

The process of predicting future product demand to help in making decisions about marketing expenditure, investment in production capacity and the scheduling of factory output. Various sales forecasting methods can be used to make predictions about both future levels of industry demand and the company's own particular sales (which depend on its market share). These methods vary greatly in terms of their subjectivity, sophistication, data requirements and cost, and a manager's choice of method will depend on the level of accuracy he requires, on the distance into the future to be forecast, and on expense. The main forecasting methods are:
  1. survey methods, the most subjective of the demand-estimating techniques available. Surveys generally involve the use of interviews or mailed QUESTIONNAIRES asking individual consumers about their future buying intentions; alternatively, opinions of the sales force may be obtained. Such data can often be useful for making short-term projections about future spending, and, in the case of new products, surveys may provide one of the few methods available in the absence of historical data. However, their subjectivity can detract from their accuracy Buyers may be unwilling to give correct answers for reasons of commercial secrecy, or be uncertain about their future buying intentions; or interviewer bias may occur in the way in which questions are asked. Sales-force opinions represent ‘second-hand’ data and can be distorted by optimism or pessimism of sales representatives. For these reasons surveys are frequently used to supplement rather than replace other forecasting methods.
  2. Extrapolation methods employ time-series data, using past sales to predict future sales. These techniques assume that the historical relationship between past and future sales will continue to hold. Time-series data usually comprise:
    1. secular trend, which shows the relatively smooth, regular movement of the time series over the long term.
    2. cyclical variation, which consists of medium-term, regular repeating patterns, generally associated with BUSINESS CYCLES. The recurring upswings and downswings in economic activity are superimposed upon the secular trend.
    3. seasonal variation, which consists of short-term regular repeating patterns, generally associated with different seasons of the year. These seasonal variations are superimposed upon the secular trend and cyclical variations.
    4. irregular variations, which are erratic fluctuations in the time series caused by unpredictable, chance events. These irregular variations are superimposed upon the secular trend, cyclical variation and seasonal variation (see Fig. 77).

    Time-series analysis is concerned with isolating the effect of each of these influences upon a time series with a view to using them to project this past experience into the future. In order to identify the underlying secular trend in a time series, the forecaster may ‘fit’ a line by eye to time-series observations depicted on a graph. Alternatively, the forecaster may use a moving average to smooth the time series and help identify the underlying trend. For example, he could use a five-period moving average, replacing each consecutive observation by the average of that observation and the two preceding and two succeeding observations. Exponential smoothing provides yet another technique which can be used to smooth time-series data. It is similar to the moving-average method but gives greater weight to more recent observations in calculating the average.

    In order to identify the effect of seasonal variations, the forecaster can construct a measure of seasonal variation (called the seasonal index) and use this to ‘deseasonalize’ the time-series data and show how the time series would look if there were no seasonal fluctuations. Once the trend has been identified it is possible to extrapolate that trend, and estimate trend values for time periods beyond the present time period. In Fig. 77, for example, the trend for time periods up to and including time t can be extrapolated to time t + 1. Such trend-projection techniques are generally more useful for longer-term forecasting than short-term estimation. Their big drawback is that they make no attempt to represent the factors which causally affect demand, simply assuming that historical relationships involved in the time series will continue into the future. This renders them unable to predict sharp upturns or downturns in sales associated with dramatic changes in demand-influencing variables.

  3. Barometric methods seek to predict the future value of sales from the present values of particular statistical indicators which have a consistent relationship with sales. Such leading indicators as business capital investment plans and new house building starts can be used as a barometer for forecasting product demand, and they can be useful for predicting sharp changes in demand.
  4. Econometric methods predict future sales by examining other variables which are causally related to it. Econometric models link variables in the form of equations which can be estimated statistically and then used as a basis for forecasting. Judgement has to be exercised in identifying the independent variables which causally affect demand. For example, in order to predict future quantity of a product demanded (Qd) we would formulate an equation linking it to product price (P) and disposable income (Y ):

then use past data to estimate the numerical value of the coefficients a, b and c. This method can be expensive in terms of data collection and processing costs, but it can produce reasonably accurate forecasts and offers the opportunity to learn from past forecasting mistakes by amending the forecasting equations, adding new independent variables etc., to improve future forecasts.

Ultimately no forecasting method will generate consistently accurate forecasts. This means that managers need to exercise judgement in choosing which forecasting methods to use and in combining information compiled from several different forecasting methods in arriving at a judgement about future demand conditions. Moreover, in making any forecast the forecaster must allow for a margin of error in the forecast, anticipating that future demand may be higher or lower than the forecast value.