Demand forecasting is a method to predict future customer demand of goods to optimize supply decisions by business organizations. There are basically two types of demand forecasting methods. Qualitative and quantitative methods. Qualitative methods basically use expert opinion and other field information to forecast demand while the quantitative methods are mainly based on historical sales data. Demand forecasting is useful in production planning, inventory management, assessing future capacity requirements and making business decisions for entering new markets, etc.
Why Forecast is Important?
Forecasting is used for several reasons, some of which are given as below:
To Estimate Demand
Forecast is a useful tool to estimate demand of existing products and new products to be introduced. Businesses always search for new and lucrative avenues to generate income streams. If the demand of an existing product will be fairly stable in future, then it will make sense for the organization to invest in the production of that particular product. Similarly, for developing new product portfolio, the organizations need demand forecast for new products or services. The impact of product lifecycle on the demand is also considered. The demand of the product or service varies in each stage of its lifecycle, right form product development, to introduction, to growth and to their maturity stage after which the product or service is discontinued. Therefore, organizations need demand forecast during each stage of a product lifecycle to formulate their strategy.
To Predict Prices
The finished products use different raw materials for their manufacturing. The price of these products is dependent on these raw materials like crude oil, metals, etc. Therefore, the forecast is used to predict the prices of raw materials to fix the cost of finished products. This type of forecast is also useful for the organizations who are the suppliers to another industry, especially the one that is changing quickly. The organizations can take decision on inventory management on the basis of predicted prices of the raw materials.
To Determine if Supply Can Meet Demand
The organizations have their limited capacities to produce and store the products. The production planning and inventory planning will be easy if the organizations will be able to predict demand. Therefore, forecast is helpful to determine if the supply can meet demand. The organizations may accordingly enhance their production capacity or stop taking new orders based on the demand forecast.
To Predict Technology Trends
Organizations also need to forecast technology trends that can impact demand of products and services. In some industries, technological advances have significantly cut product life cycle times, for e.g. computer and mobile phone industries. If an organization is not ready for these changes and is not able to modify their products and service as per the technological changes, then their competitors will take advantage of that. Therefore, senior management must be aware of any technology gaps that may exist within their organization and the potential threat to revenue or market share if their competitors adopt the new technology. Introduction of Reliance Jio services by Reliance Jio Infocomm Limited, is an example of how the technological disruptions can change the market scenario. Jio Infocomm Ltd. Used provided cheap high speed data to the customers with the help of 4G mobile technology and snatched a large chunk of market share from existing players like Vodafone and Idea who were using 3G mobile technology. Vodafone and Idea later on merged and became Vodafone Idea Ltd. (VI) so that they can compete with Jio.
To Predict Dependent Demand
The demand of components and raw materials from which the finished products will be made is also determined with the help of forecast techniques. For e.g., a garment manufacturing company predicts the garments to be sold in summers to estimate the dependent demand for yarns, colors and other raw material used in the cloth manufacturing.
Characteristics of Forecasts
(1) Forecasts are always inaccurate and therefore a measure of forecast error is important to make the forecasts accurate as much as possible. Let us understand this by an example. Suppose a cellphone dealer expects sales of a particular cellphone in the range between 50 to 250, and another dealer expects the sales between 10 to 290 units. Even though the average expected sale of both the dealers is 150 units, the sourcing policies and planning of both the dealers will be different because of the difference in forecast accuracy. Thus, determination of forecast error is very important in most of the sourcing decisions.
(2) Long term forecasts are less accurate then short term forecasts because long term forecasts have a larger standard deviation of error relative to the mean as compared to the short term forecasts. It is evident from the fact that if you have to forecast for one-week period then it will be more accurate than one month forecast because the short lead time allows you to take into account current information that could affect product sale, such as weather.
(3) Aggregate forecasts are more accurate then isolated forecasts, for e.g. it is easy to forecast total sales of cement as whole of cement industry for a particular year then to determine the sales forecast of a particular company due to aggregation of data in case of whole of cement industry sales forecast.
Factors that Affect Demand Forecasts
There are several factor that may affect demand forecast such as lead time, labor market, material shortage, shifts in technology, whether condition, etc. Lead time is the time taken between placement of an order until receipt of that order. Any customer organization plans to receive the order in a planned time as stipulated in the contract but if there are significant variations in the lead time then it will adversely affect the forecast.
The shortage in labor will make it difficult to meet the forecast. Similarly, the material shortage can impede the ability of and organization to meet demand. As manufacturers move to just-in-time and lean production, the chances of supply shortages increase and must be managed effectively.
Shifts in technology also affects the forecast but in long term. For e.g. the change in mobile phone technology may affect the demand of conventional products and services adversely. Climatic conditions also have impact on the demand forecast. Adverse climatic conditions like heave rainfall, snowfall, etc. may often obstruct smooth flow of products and goods and hence may pose difficulty for manufacturers & distributors to meet demand for a long time.
Types of Demand Forecasting
Passive Demand Forecasting
Passive demand forecasting is the simples form of forecasting used for stable businesses and products which are in continuous demand. The past demand data for such products and services is available and is used for passive demand forecasting. Simple extrapolation of past data is done and no statistical or trend analysis of data is done in this type of forecasting. It is suitable for local and small businesses.
Active Demand Forecasting
This type of forecasting model employs marketing research, market campaigns and other survey methods to forecast demand. It also takes into account the external factors like economic environment and growth projection. It is used for businesses having aggression growth plan and is suitable for start-up businesses having lack of historical data.
Short Term Demand Forecasting
When a demand forecast is done for a shorter period say for 3 to 12 months, then it is called short term demand forecasting. Many businesses now a day use just-in-time approach to manage their supply chains. This type of forecasting model is suitable for such type of businesses. This model also considers seasonal pattern of demand while making the forecast.
Long Term Demand Forecasting
Long term demand models are used for demand forecasting of longer durations such as 12 moths to 24 months or longer durations. It is used as a road map for strategical long term business decisions pertaining to sales & marketing planning, financial planning, etc.
External Demand Forecasting
This model studies external macro forecasting trends and evaluates strategic objectives of the organization such as product portfolio expansion, risk mitigation strategies, technological disruptions, etc. in the light of external factors.
Internal Demand Forecasting
The internal operations within the organization such as product category, manufacturing groups, supply chain operations, etc. have certain limitations. The internal operations and factors may not be efficient enough to keep pace with external demands. The internal demand forecasting model helps removing the internal limitations that might slow the growth of an organization. It is a helpful tool for making realistic projections. It can point out the areas where you need capacity building in order to meet expansion goals of your organization.
Components of a forecasting method
There are several factors that affect the demand of a product and it can be predicted at least with some probability if an organization can determine the relationship between these factors and future demand. Some of the factors that should be taken into account for demand forecasting are as follows:
- Past demand
- Lead time of product replenishment
- Planned advertising or marketing efforts
- Planned price discounts
- State of the economy
- Actions and strategies of competitors
Demand Forecasting Methods
Qualitative forecasts are basically judgmental forecasts because they are developed based on the judgement or opinion of industry experts, estimates from sales persons, etc. These methods are useful when the organization has no quantitative data available with them for analysis. When a decision maker in the top management of the organization has high level of experience and knowledge then this these methods are usually employed. These methods are also valuable when the quantitative methods are not able to give accurate results. Details about some of the qualitative methods are given below:
Sales Force Composite
In this method of sales forecast, each salesperson provides an estimate of sales based on his/ her knowledge and then then the estimate of each salesperson is aggregated by a manager. Based on the field data, this forecast may also be compiled at the district or national level. There are some advantages and some disadvantages of this method. The advantage is that the sales force usually has the most knowledge of customer behavior and expected sales in their region and therefore their knowledge and expertise is utilized to determine the accurate forecasts. The disadvantage is that the salespersons providing the data may have their own biases and can adversely affect the forecast. In some instances, the salesperson may deliberately underestimate or overestimate the sales data for various reasons. To overcome these problems, the sales team should be briefed to provide the realistic data and it should also be monitored and adjusted at the management level also.
This method utilizes the result of surveys conducted among customers to estimate the degree of interest of customers in a particular product or service. There are various survey methods used like by telephone, e-mail, through online portals, google forms and personal interviews. A sample representation of market is selected by the marketing team to conduct the survey and the data is analyzed using statistical tools and base on own judgement as well. This method is best when used for short term forecasts. Like other methods, this method too has certain disadvantages. For e.g., the questionnaire may not be true representation of the feelings of the customers or the customers may be unable to provide their true response without actually seeing/ using the product. Moreover, if people do respond to a survey, there is no guarantee that they will actually use the product or service. Sometimes, it is difficult to get the response of customers through e mails or other web based surveys. To overcome this issue, the marketing teams should provide detailed online information of the product or service including pictures.
Jury of Executive Opinion
In this type of method, the experience, knowledge and opinions of employees within the organization or some external organizations is used to forecast the demand. This type of forecast is generally used when past data is not available, when a forecast is totally out of line with competitors of an executive has some information that is not available to a forecaster. The advantage of this method is that it is simple and easy to use and therefore it is one of the most commonly used forecasting methods. The disadvantage of this forecast is that it is subjected to high degree of judgmental bias.
The judgmental bias in case of Jury of Executive Opinion can be reduced in Delphi method. In this method, opinion is obtained from a panel of experts, usually five to ten. Every participant is free to express his/ her opinion. After initial survey, a coordinator circulates back the results of the survey along with the comments to each participant to review the results. The panelists then are free to adjust their opinions based on the outcome of initial survey. Each participant then resubmits any changes based on the results of the first round and the process continues until consensus is reached. The process can take multiple iterations to reach consensus. This method can be used for sales revenue forecasts including long term, developing projections for new product demand and predicting technological developments.
Quantitative forecast methods are useful when the product or service has been in the market since a long time and a past sales data of the product or service is available. With relatively recent advances in computer capabilities and software, quantitative forecasting is less tedious, allowing forecaster to focus on interpreting the results rather than assembling the information. Some of the quantitative methods are discussed below:
This is one of the simplest forecasting method in which the upcoming forecast is set equal to the most recent period’s demand. For e.g. if Hyndai Motor sells 500 cars in August then the demand for September will also be 500. This method is usually uses as a starting point and then adjustments are made based on the past experience. This is also used to compare results of other methods. This is the least expensive, easiest and most efficient method.
Simple Moving Average
This is also a simple method to forecast demand in which average of most current ‘n’ periods are calculated. In each recalculation, the most current period’s data is added and the oldest data is removed.
Weighted Moving Average
It is similar to ‘simple moving average’ method except weighted average is calculated instead of simple average. If a noticeable demand pattern is observed, the forecaster can apply individual weights to each past demand to place more emphasis in certain periods.
Weighted moving average= ∑ [(weight for period n) (demand for period)]
The advantages of both simple and weighted moving average methods is that they smooth out any unexpected fluctuations in demand so that stable estimates can be made. The disadvantage of these methods is that they cannot predict unprecedented events because they are based on past data. Another limitation is that both methods require a significant amount of past data to be effective.
In case the ‘simple or weighted moving average’ methods require a large amount of past data, the ‘exponential smoothing’ method is used because it requires a relatively small data. In this method, the forecast of previous period is taken as a basis for the current forecast and weightage is given to the demand for the most recent period. This is a fairly accurate method and a few calculations are required in this method. ‘Exponential smoothing’ method is a more sophisticated method then the ‘simple or moving average’ method and a qualitative method such as the Delphi Method or managerial experience is required to set the initial forecast.
A value of α closer to 1 heavily weight the previous period’s demand and the value of α closer to 0 weights the previous period’s demand less heavily.
In certain products, the seasonal pattern is observed, for e.g., if you consider the sales pattern of umbrellas, you will find that the sales of umbrellas increase in rainy seasons while it goes downwards in summers. One method to access the seasonal demand is ‘multiplicative seasonality. In this method, the seasonality is expressed in terms of a percentage of the average demand. This percentage above or below the average is known as the seasonal index. For e.g., if the demand for umbrellas is 1.5 times the average demand in rainy season, the rainy season demand is 50% above the average annual demand.
Trend- adjusted Exponential Smoothing
A trend is a consistent upward or downward movement in demand over time. In case a trend is observed and the effect of the trend is to be taken into account in the forecast, then an exponentially smoothed forecast can be adjusted, to a certain extent, by incorporating a trend component. This method is then called as ‘trend- adjusted exponential smoothing’. In this method, two smoothing constants α and β are used. The α value smoothes the initial forecast using the exponential forecast method, while β value smoothes the trend. The value of β must be between 0 and 1, a value closure to 1 mean the forecaster has placed more emphasis on changes in recent trends, while a smaller value (closure to 0) lessens the effect of a current trend.
Sometimes, the demand exhibits a clear linear trend either upwards or downwards. In such cases ‘linear regression’ can also be used for demand forecasting. In ‘linear regression’ one dependent variable is related to one or more independent variables by a linear equation. The dependent variable will be one that will be forecast, such as demand, while the independent variable will be one that has a significant effect on the dependent variable. In ‘linear regression’, a straight line is fitted to past data based on fitting technique known as the least square method.
(1) Demand forecasting is a method to predict future customer demand of goods to optimize supply decisions by business organizations.
(2) There are basically two types of demand forecasting methods qualitative and quantitative methods.
(3) Qualitative methods basically use expert opinion and other field information to forecast demand while the quantitative methods are mainly based on historical sales data.
(4) Forecast is important to predict prices of the products & services, to predict technological changes and to determine if supply can meet demand.
(5) Forecasts are always inaccurate and therefore a measure of forecast error is important to make the forecasts accurate as much as possible.
(6) Long term forecasts are less accurate then short term forecasts and aggregate forecasts are more accurate then isolated forecasts.
(7) Lead time, labor and material shortage, shift in technology & adverse weather conditions are some of the factors affecting demand.
(8) Qualitative methods are useful when the organization has no quantitative data available with them for analysis.
(9) Quantitative forecast methods are useful when the product or service has been in the market since a long time and a past sales data of the product or service is available.