Sales volume forecasting. Criteria and factors for choosing certain methods when making forecasts

1. Development of technical condition for the forecast of an object and subject

2. Set tasks

4. Indicators

2. stage Collection of objective information

3. stage Sampling and its compilation

4. Drawing up a search forecast

5. Regulatory forecast

6. We give it to specialists for examination

7. Collective export assessment

8. We make adjustments to our forecast

9. We hand over to the customer

10. We implement

11. We make adjustments to the forecast

12. Summing up

13. Criteria for the truth of social forecasts and projects.

The value of social forecasts is determined not only by the social significance of the predicted phenomenon, but also by the degree of its accuracy and reliability. The level of reliability of forecasts determines both the outcome of future events and the timeliness of impacts on the processes of the present, from which the future is born. The present represents the only period of time flow in which people influence the historical process. The degree of this impact depends on the expected outcome of future events, on the degree of probable implementation of the forecasts. Making decisions in the present, making plans for the future, determining immediate tasks and more distant goals both in individual human behavior and in the activities of individual teams, social groups and the entire society as a whole - all this puts the problem of the truth or reliability of social forecasts at the center scientific research.

Forecasting is a necessary link in the process of cognition; it is organically included in it and is subject to all its laws and principles. The most important of these principles is a correct understanding of the essence of social practice, its role in the process of cognition. Being the basis, goal and criterion of the truth and reliability of human knowledge, practice plays a similar role in relation to the predictive function of the cognition process. Here, as in the entire process of cognition, practice appears in dialectical unity, that is, as the basis, goal and criterion for the reliability of our knowledge about the future.

However, social forecasting has its own specifics, arising from the peculiarities of the manifestation and implementation of the laws of social life through the conscious, purposeful activities of people. The study of these laws and, on this basis, a scientific assumption of the direction and nature of their action in the future constitute, in essence, the basis of social forecasting. Consequently, the role of practice in social forecasting must inevitably be characterized by the same specificity that distinguishes social forecasting from the process of cognition as a whole and from forecasting in general. Of course, any forecasting as a function of social consciousness is social, meaning a forecast of directly social processes, i.e. a forecast of the development and further functioning of the social organism itself, social forecasting V in the proper sense of the word.

Reliability of the forecast. The reliability of a forecast is understood as the necessary or at least sufficient degree of probability of the justification of the developed forecasts. The reliability of forecasts depends on a number of factors: the depth and clarity of our knowledge of objective patterns and laws that determine the development of various social systems; the ability to correctly display and “model” social systems. An important role is played by the degree of knowledge of specific social processes, the completeness of information about the conditions and cause-and-effect relationships that determine the development of a particular social process, as well as the timeliness and speed of processing the flow of information data about specific social processes and relationships. The quality, completeness, and reliability of social information, as a rule, are directly proportional to the reliability of the social forecast.

The above allows us to draw some general theoretical and methodological conclusions that are fundamental for any prognostic activity. Firstly, It becomes obvious how important the theory is for prognostication. Without theoretical insight into the essence of ongoing processes and without practical experience, without the creative application of scientific theory, scientific forecasting is generally unthinkable. Forecasting activity is primarily a scientific and theoretical activity. In any area of public life it must be understood and implemented as a continuous constant creative process. The point is not to hastily draw up some picture of the future, to imagine the future in accordance with some arbitrary requirements, but to organize responsible, purposeful and continuous scientific activity, which must be free from any assault and incompetent supervision.

Secondly, The decisive condition for any forecasting is the completeness of information regarding the laws that determine the course of the predicted process; without this, creating a serious forecast is generally impossible. Of course, it is impossible to obtain complete information about all applicable laws. In many cases you will have to deal with hypotheses about patterns. However, you need to know in advance that the reliability of the forecast, as a rule, decreases the more you have to resort to hypothetical statements about the law, so that ultimately the forecast may lose value. Therefore, analysis of the law of development (and this is an important theoretical condition) is the “alpha and omega” of any forecast.

Third, Crucial for the forecast is the most realistic and accurate understanding of the determining conditions under which the predicted processes begin and which influence them. This includes the anticipation of the implementation of the conditions for the operation of a significant (relevant) law for the forecast, and data on other possible conditions and factors that could affect the object of the forecast. As a rule, information about the initial and accompanying conditions of the predicted process is as difficult to obtain as information about the law of development. It is not uncommon to first have to deal with a large number of interacting factors, which can then be included in larger groups. Therefore (especially when forecasting in the field of economic management), the strictest objectivity, scientific reliability, honesty, exclusion of subjectivity and “embellishment” in the analysis and assessment of initial and accompanying conditions are indispensable prerequisites for successful forecasting.

Forecasting- activities aimed at identifying and studying possible alternatives for the future development of the company. The main role here is given to forecasting product sales. The main purpose of the forecast is to determine trends in factors affecting market conditions.

When forecasting, short-term forecasts are usually distinguished - for 1 - 1.5 years, medium-term - for 4-6 years and long-term - for 10-15 years.

The main emphasis when short-term forecasting is done on the quantitative and qualitative assessment of changes in production volume, supply and demand, price levels and indices, currency ratios and credit conditions. Temporary, random factors are also taken into account.

Medium term And long-term forecasting is based on a system of forecasts - the relationship between supply and demand, restrictions on environmental protection, and international trade.

Formalized quantitative methods (factorial, statistical analysis, mathematical modeling), methods of expert assessments based on the experience and intuition of specialists in a given product and market are used as forecasting tools.

The most important forecasts in the activities of companies are sales forecasts, in the development of which the following basic methods can be used:

survey of a group of managers of various services and departments of the company, and generalization of assessments of individual sales agents of the enterprise and heads of its sales divisions - the forecast is the average of their opinions. The method is used for new firms that do not have experience in using other methods, and also when there is no detailed information about market trends. Within the framework of this method, it is possible to take into account regional characteristics of demand and conditions for selling the company’s products;
forecasting based on past turnover - the growth rate of sales volume in the reporting year is determined in comparison with the previous one and the assumption is made that the achieved growth rates will continue next year:
Next year's turnover = Reporting year's turnover x (Current year's turnover: Last year's turnover).
The method is used for markets with stable conditions, slightly changing assortment, minor fluctuations in turnover and sluggish scientific and technical progress;
analysis of trends, cycles and factors influencing sales volume. The most significant factors include: long-term growth trends of the company, cyclical fluctuations in business activity, seasonal changes in sales, technical changes, the emergence of new competitors, etc. The method is used for long-term forecasts for a period of at least 3-5 years and is most applicable in capital-intensive activities;
correlation analysis - complements the previous method, but is based on the use of more complex methods of statistical analysis. The close relationship between the level of sales and various factors influencing it is revealed, on the basis of which the factors are ranked in order of importance. The method requires large expenses associated with in-depth market research, and produces the most accurate results in markets with stable conditions;
forecasting based on the “market share” of a company’s sales— sales are forecast as a certain percentage of the firm's market share in a given industry. A calculation is made of the company's share in total sales in the market. When using the method, it is important to be confident in the accuracy of the sales forecast for the market as a whole and not to take into account non-price competition;
end use analysis— the forecast is based on the expected volumes of orders from the company's main customers. Total sales usually exceed this figure by a certain percentage. The method requires conducting research on the main industries that consume the enterprise's products, and is most preferable in the sectors of the raw materials and energy complex and in companies that produce finished products and components;
product range analysis— sales forecasts for individual types of products are brought together and form the company’s planned turnover. The method is suitable for diversified firms; its accuracy depends on detailed market research for each type of product;
test marketing - one of the most accurate approaches to sales forecasting. A new product and the system for its promotion on the market (prices, types of advertising, sales channels, type of packaging) are tested in a small regional market, and then information about the sales volume on it is distributed to the entire sales market of the company;
standard probability distribution methods— three types of sales forecasts are determined by experts: O — optimistic forecast; IN - most likely prognosis; P - pessimistic assessment of the sales forecast. Next, the expected value of the sales forecast (C) is calculated using the formula

C = (O + 4B + P): 6.

Standard deviation (CO) calculated as C0 = (0 − P) : 6. In accordance with the general theory of statistics, the most probable value of the variable - sales volume with a 95% probability will be within C ±2 CO.

The effectiveness of using a particular method depends on the specifics of the company’s activities. It is usually considered that the forecast has been drawn up correctly if the deviation of actual turnover from the planned one is no more than 5%.

The sales forecast is the basis for drawing up a plan for the production and sale of the company's products.

When developing a sales forecast, an integrated approach, the simultaneous use of several forecasting methods and comparison of the results obtained are important. Among these methods, the most common are the following:

1) Method of expert assessments (including the opinion of a group of managers and a combination of opinions of sales employees). This forecasting method is most suitable for new businesses that do not have enough experience in using other methods. This method is also applicable when there are no detailed calculations about the state of the market, there are no complete statistics on sales trends for certain types of products.

2) Extrapolation of trends and cycle. When using this method, errors are inevitable, but it is invariably used in sales forecasting; the low percentage of predicting the consequences of socio-economic phenomena does not contribute to high accuracy of the forecast. The use of this method is possible if the analyst has at his disposal massive amounts of information on various areas of the company’s activities over the past 10 years.

The use of this method is based on the following techniques:

A) Determination of moving averages.

The product sales diagram most often has an abrupt character. Averaging the observation results will allow us to construct a sales curve over time. A suitable number of observation results are averaged. It can use quarters, which means adding the first three results and dividing the sum by three. Then the results of the second, third and fourth observations are added and divided by three, etc. The result is a quarterly moving average. The constructed graph determines the prospective sales values.

B) Smoothing models.

Over time, more and more observations are made and the size of the forecast errors is determined. At the same time, it seems rational to take past mistakes into account when predicting the future. One way is to add a fixed percentage of last month's error to last month's actual sales and use the result to forecast the next month. Using this method, you can get quite good short-term forecasts. Such forecasts are useful for production planning and inventory management, but are practically inapplicable for financial planning.

3) Forecasting based on a portfolio of orders, that is, based on existing or expected orders from potential buyers of products, which is preferable for generating sales volume in high-tech industries. The application of this method requires conducting special research on the main industries consuming the products of a given enterprise, collecting and processing significant statistical and factual material. This method is preferable in the raw materials and energy sectors, as well as in enterprises that produce components and components.

4) Correlation analysis, that is, the identification of statistically significant factors influencing the sales of the company’s products. Using the correlation relationship, the closeness of the connection between the level of sales and various results of the enterprise’s economic activity is determined, the impact on sales of which can be logically proven and justified. Thus, the most significant factors, depending on which the sales volume may change in the future, are identified and ranked (according to the degree of their influence). This method requires special and expensive research. The most accurate results can be obtained in the most stable industries in terms of economic conditions.

The effectiveness of using a particular method depends entirely on competitive conditions and the specifics of the enterprise’s economic activity and can only be determined in a system of general market research activities. In marketing-oriented companies, several forecast options are compiled using various methods (3-4 methods). The resulting estimates are then compared to identify any differences in estimates that may arise. It is usually considered that the forecast is made correctly if the difference between the estimated and actual sales does not exceed 5%. If these discrepancies are significant (the spread of sales forecast indicators using various methods exceeds 10%), then most likely errors were made when drawing up the sales forecast using some method.

1. Survey of a group of managers of various services and departments of the company. These managers must first obtain relevant information regarding market analysis. In this case, the sales forecast is something “average” of the views and outlines of the surveyed group of managers. This forecasting method is most suitable for new businesses that do not have enough experience in using other methods. This method is also applicable when there are no detailed calculations about the state of the market, there are no complete statistics on sales trends for certain types of products.

2. Generalization of assessments of individual sales agents of the company and heads of its sales divisions. In this case, market analysis is supplemented by the opinion of those who directly experience the reaction of consumers and most acutely feel the slightest fluctuations in consumer preferences. The regional aspect is also taken into account here: individual employees or sales managers can provide additional information about the specifics of selling certain products in different regions of the country. Accordingly, the accuracy of estimates with this method is higher than with the first. But organizing such work is associated with large overhead costs (primarily additional costs for remuneration of specialists and analysts, data processing, etc.). And although companies that value their brand (especially leading industrial companies with world-class production or striving to become such) never skimp on them, it often requires the development of special procedures for controlling and budgeting these expenses. Otherwise, the accuracy of the forecast may negatively affect the financial position of the enterprise.

3. Forecasting based on past turnover. In this case, sales data for the past year are taken as the basis for predicting likely sales in the future. It is assumed that next year’s turnover will exceed or be lower than this year’s turnover by a certain amount (usually a percentage increase is taken to the data for the previous year according to the so-called “achieved” principle). This forecasting method is suitable for industries and markets with stable economic conditions, weak a changing range of goods and services, with sluggish scientific and technical progress, where significant fluctuations in trade turnover occur extremely rarely. The most typical example of such an industry is public utilities. Using this method, it is impossible to take into account rapid changes in the nature of commercial activities, in the structure of consumer demand, etc. As for competition, its degree is not taken into account here at all.

4. Analysis of trends and cycles, factors causing changes in sales volume. The sales forecast is based on identifying probabilistic trends and statistically significant factors underlying them using market analysis. Typically, the following main factors are taken into account: long-term growth trends of the company, cyclical fluctuations in business activity, seasonal changes in the company's sales, possible impacts of strikes, technical changes, the emergence of new competitors in the market. This method is most preferable when making long-term forecasts. Statistical patterns, identified trends and dependencies over the course of many years neutralize the effect of random and minor factors. At the same time, using this method it is difficult to predict for a period of less than 3-5 years, the sample, the array of processed statistical information, as well as the period of manifestation of cyclical fluctuations are too small. This method is most suitable in capital-intensive industries.

5. Correlation analysis, i.e. identification of statistically significant factors influencing the sales of the company's products. It logically complements the previous method, but is based on more complex scientific tools for statistical market analysis. Usually, within the framework of special surveys, the closeness of the correlation between the level of sales of an enterprise and various aspects of economic activity is determined, the influence on sales of which can be logically proven or justified. Thus, the most significant factors are identified and ranked (by degree of influence), depending on which the sales volume may change in the future. It should be noted that such a forecast method necessarily requires serious special and comprehensive, and therefore quite expensive, not always economically justified, market research. The most accurate results, however, can be obtained using this method in the most stable industries in terms of economic conditions.

Making a forecast

Evidence of criterion-related validity means that test results can be used to draw conclusions in the form of predictions. Therefore, the basic procedure used to collect this evidence is called forecasting ( predictive design).The correlation coefficient is calculated between the test results and the grades subsequently given by the same subject according to some criterion. This is exactly the procedure used in the example shown graphically in Fig. 3.3: a correlation was revealed between the results of the arithmetic test obtained before admission to training, and grades given by managers after two weeks of training in a vocational training program.

Forecasting has traditionally been considered the preferred way to obtain evidence of criterion validity, but its practical application has certain disadvantages. The main one is associated with the presence of a normal distribution of subjects across the entire scale of test scores. For the correct use of a predictive scheme, it is necessary that the range of results of subjects from the sample used in test validation be complete. Therefore, it is necessary to hire several candidates with low test scores. It is quite difficult to convince employers of the need for this requirement. If the test is used to screen for employment, it is natural to assume that people who score low on the test will not be able to do a good job - so why hire them?

Another possible problem with making criterion-related validity predictions is that there is a time lag between the collection of test data (predictive variable) and the collection of criterion data. When predictions of behavior are extended further into the future, their accuracy decreases significantly (Henry & Hulin, 1987; Hulin, Henry, & Noon, 1990). Supervisor ratings, which seem to be the most commonly used criteria in such studies, may be particularly susceptible to this problem because they are made at a specific point in time and relate to the performance of a specific job. One way to get around this problem is to use parallel (concurrent) schemes for proving criterion validity.