Benchmarking Software Packages and Systems

Article excerpt

Given the demands on professional forecasters with regard to the number of items to be forecasted and the continual quest for improvements, it is virtually impossible to do them without the aid of computer or a proper forecasting software/system. Today's practitioners must prepare forecasts covering thousands of Stock Keeping Units (SKUs), which are broken down by regions, distribution centers, marketing channels such as food and drug stores, and customers. The increasing globalization of markets has further added to the responsibility of a forecaster. In some companies, the forecaster is required to prepare forecasts for both domestic and foreign markets. In recent years, a large number of forecasting software packages and systems have emerged, enabling forecasters to do their job more effectively and in a timely manner. Still, a pertinent question surfaces - which is the right forecasting software/system for me? In this article, we will discuss the market share of different forecasting software/systems, as well as which one dominates and in which market.

Before proceeding further, it is important to know the difference between a forecasting software package and a forecasting system. Forecasting software is a stand-alone package that generates forecasts either by the model selected by a forecaster or by an automatic feature (often called expert system) that is built into the software package. Software with an automatic feature tests a number of forecasting models with the historical data, and then prepares forecasts for the next period with a model that, on the average, gives the lowest error.

The forecasting system, on the other hand, does more than forecasting. It basically automates the entire forecasting process. The system follows set procedures for acquiring data/information from various sources and, after the data are acquired, it cleanses them. If needed, it slices and dices data (breaking up monthly data into weekly or daily buckets) or aggregates them (aggregating SKU data into categories and category data into overall totals). It also looks for problems in the data, e.g., outliers (unusual values) and structural change (a permanent change in the data pattern because of merger and acquisition or for any other reason). Once problems are identified, it treats them as specified in the system, prepares forecasts either by the model specified by the forecaster or by using the automatic feature, and prints out the forecast report in a format specified by the forecaster in the system. The forecasts, for example, may have to be expressed in units and dollars; by SKUs, categories and aggregate (company as a whole); and by regions, distribution centers, channels, and customers. Furthermore, forecasts may have to be produced by week, month, and/or quarter. If needed, the system allows overriding of forecasts only by those who have the authority. Once forecasts are finalized, the system transmits them to the end-users. Then begins the monitoring, updating, and revising process, which occurs automatically and routinely. The forecasting system also connects the final forecasts to various modules (e.g., to the production module for preparing production plans, to the procurement module for making plans for purchasing raw materials, and to the logistic module for making plans for transportation). Depending on the system, forecasts may also be linked to the financial module to generate a financial plan, to the sales module to generate a sales plan, and to the marketing module to generate a marketing plan.


Although in principle there is a distinct difference between a forecasting software package and forecasting system, in practice, it is not that clear cut. In some cases, it is difficult to identify whether you are using a forecasting software package or a forecasting system. So, some judgment is necessary. The separation, shown here, was made by the author, based on the author's own knowledge and experience. …