Sales Forecasting, an Introduction

     Sales forecasts are becoming more important to businesses.  Objectivity is a key ingredient.  Understanding forecasting techniques is a necessity to insure that a proper base is established for planning.

Sales forecasting, an Introduction:

     Forecasts are the raw material of planning.  For most businesses, the sales forecast is the first step.  It can be the basis for developing plans to run the business and plan future strategy.  If sales can be reasonably anticipated, costs and inventory can be controlled, and customer service enhanced.  Sales estimates provide this help.

 

    Even if much of the content in this paper is familiar, a review will probably provide additional information that is helpful in learning about forecasting strategy and tactics.  In recent years sales forecasting has become an important business endeavor.  It is essential that company personnel and small business owners understand forecasting methodology.

 

     Forecasts can be classified in several ways:  by purpose, as either operational or strategic; by type, as quantitative (statistical) or qualitative (judgment); or by method, that is, by forecasting technique.

 

     Operational forecasts are those sales (business) estimates that have an immediate impact upon near- and mid-term operations, whether the business is manufacturing, retail, importing, or distribution.  Sales estimates support sales/marketing campaigns, scheduling end-item or materials purchases, manufacturing (if applicable), customer service, budgeting, and short term financial goals.  Operational forecasts extend at least to that future time period in which end-items need to be ready for sale.  Usually, however, forecasts are made to one year in the future in support of marketing and financial objectives.  Forecasts are revised frequently, typically monthly, but for some businesses weekly.

 

      It is important that as much objectivity as possible be incorporated into the forecasts.  This implies that a statistical or mathematical technique be used to forecast each end-item.  This does not imply that the method has to be complex.  There are many books that explain many different forecasting techniques.  Unless you are a mathematician, suggest that the literature selected be comprehensive (describes a multitude of models) and is written for the practitioner; that is, in a step-by-step format.  

 

       It is not unheard of that each department has its own sales projections.  However, forecasting is a centralized function with all consumers a part of the process.  In larger companies, there should be one person or group responsible for forecasting, but with participation from each of the functional activities, the users.  In smaller firms, it is likely that executives wear more than one hat, but still there is one person in charge of forecasting.  The point is that managers participate and that there is only one sales estimate.  Operating plans and budgets issue from the estimates. 

 

     Strategic forecasts estimate the longer range, perhaps as much as five to ten years in the future.  They are a different breed than operational forecasts and likely will employ different forecasting techniques. 

 

     Strategy is the plan by which stated long-term objectives are achieved.  These objectives are usually concerned with growing the business or maintaining market share during uncertainty.  Strategy formation is dependent upon long-range forecasts, probably related to economic factors that affect the business, but also, social, political, and technological factors. 

 

     

     Quantitative forecasts are those that apply mathematical techniques to arrive at the sales estimates.  In most cases, highly complex methods will be unnecessary for operational estimates.  There are exceptions: if the pattern contains seasonal or cyclical data or is nonlinear, the technique may be more complicated, but not beyond average ability.   One note: naiveté must be avoided, such as next month’s sales will be the same as this month’s sales or this month’s sales will be the same as sales last year in the same month. Regardless of the size of the business, objective forecasting is best.

 

     Qualitative forecasts are those that rely upon judgment in one form or another.   The simplest form is executive judgment (a decision by an individual).  The most comprehensive are Delphi and probability trees. 

     There are three conditions in which judgment is appropriate:  for new businesses with very little sales history, when a forecast needs to be modified for a one-time or unusual situation (one-time promotion, etc.), and when there is a sudden, dramatic shift in sales, as occurs during a severe, unexpected recession.

 

     The following paragraphs briefly describe the several categories of the most applicable forecasting techniques.

 

     Time-series models are statistical in nature.  They focus upon the historical pattern of the demand (sales) itself.  The forecast (sales estimate) is a projection of the past into the future.  Although the pattern of past demand may be changing, the assumption is that the change is orderly.  Time-series analysis is concerned with trend and the rate at which the demand is changing, but also considers cyclical and seasonal variations.  The main categories are:

 

·         Moving averages are arithmetic or weighted averages of a number of past demands.  The estimates are based upon the pattern or trend of the historical values.

·         Exponential smoothing is a method similar to moving averages except that more recent demands are given more weight.  The equation itself assigns the weights (how much each period of demand will be considered by the model).  The forecaster assigns a smoothing constant (a weighting factor) to the equation.

·         Decomposition is more complex.  Its main function is to divide the historical time series into its components for evaluation.  The components that may be present are: trend, seasonal, cyclical (the business cycle), and randomness (the unknown variations in a time-series).  It may also be a forecasting vehicle. 

·         Nonlinear trend models are applicable when the pattern follows a curve rather than a relatively straight line.  There are several useful models.

·         Other time-series methods are least squares trend, which takes a number of past observations and calculates a straight line; percent of sales, which calculates percentages for each period and projects those percentages into the future; erratic demand, which are techniques available when there does not appear to be a definite pattern; and Box-Jenkins, which is complex.  

     There are about eighteen models all told.  Many of the less difficult methods will frequently be suitable for estimating end-item or product group sales.

 

     Regression (causal) models relate independent data (independent variables) to the item being forecast.  The forecast is based upon the explicit relationship between the independent variable(s) and the demand being forecast, the dependent variable.   Most often (but not always) regression is used in the longer term in support of strategy formation.

 

     An independent variable may be any data series that moves in relation to the dependent variable, that is, when one moves in a given direction, so does the other.  Independent variables may be economic data, such as gross domestic product or its subdivisions, total industry sales, demographic data, housing starts, the size of the advertising budget, etc.  If there is one independent variable, it is termed simple regression.  If more than one variable, it is multiple regression.

 

     Econometric models involve a system of interdependent equations designed to estimate an economy or an economic unit.  However, there are much less complex portions of the discipline that can be used by companies in the forecasting process, such as diffusion indices and rates of change (growth rate).

 

     Forecasting is not complete without measuring the average accuracy of the estimates.  As important may be calculating a range; that is, a high, low, and probable estimate.  There are several methods available, but a range derived from the standard deviation is a very good choice.  A range allows the planning staff or management to engage in a “what-if� scenario (what-if the low range is the most probable, what affect will it have on inventory cost, customer service, etc?).  Anticipating a range is important both to operational and strategic forecasting.  Also, in strategy formation various scenarios about the future economy need to be considered.

 

     A good practice in forecasting is to continually track the error rate, the difference between estimated sales and actual sales.  The objective is to see if the error rate is holding steady or increasing.  If forecasts are drifting further from reality, review of the forecast model is appropriate with the idea of corrective action.

 

     Forecasting question?  I’ll try to answer it for you.  Also, check out my book (Business Forecasting: A Practical, Comprehensive Resource for Managers and Practitioners) on Amazon.

 

 

 

   

 

     

 

More Recent Articles

• Minority Business Loan
• Business Loan Interest Rates
• Business Loans For Poor Credit
• Is Your Timing Keeping You From Being Success...
• Interest Rate For Business Loan
• VA Small Business Loan Qualifications
• Federal Business Loan
• Insurance For Cleaning Business