S.F.T. Inc.

Improving Sales Forecast Accuracy by
Analyzing Demand Trends

(if you are not using a NETSCAPE-compatible browser, this page may appear poorly formatted)

Table of Contents
  • Introduction
  • An Effective View of Forecast Accuracy
  • Conclusion

  • Introduction

    The entire company must endure the discord of a poor forecast. Financial and Operations management must deal with the impact of an optimistic forecast. Cash is often tied up in slow moving inventory as well as the opportunity costs associated with the production time for items that don't sell. Conversely, a pessimistic forecast causes Marketing and Sales to have shortfalls in revenue due to limited product availability. So why doesn't every company improve the situation by making this effort to improve the quality of their forecasts?

    The answer is most companies do make the effort, but are limited by the accessibility to the data and the tools that can easily manipulate the data into credible information. Therefore, the effectiveness of the effort is limited.

    One of the key problems in developing an accurate forecasting process is the inability to obtain a thorough, real-time view of forecast variations. Most forecasts are developed based upon static information and assumptions, and the usual differences between actual demand and forecast demand are often misinterpreted, resulting in attempts at fixing a non-existent problem. This causes a chain reaction of events that cascade down through the company, often resulting in higher inventories, poor customer delivery performance, longer customer order lead times, and increased overhead costs due to excessive changes to production plans.

    To obtain the highest quality forecast, a thorough understanding of products and customers must be obtained. To complete this analysis, the data must be segmented by a combination of customer, product line, sales region, sales channel, units, revenue, average selling price (ASP), costs, and time-frame. To review a forecast with this degree of detail, it is necessary to manipulate the data in several combinations of these segments to see how well the past trends align with the future predictions. On a regular basis, often weekly and sometimes daily, the task becomes impossible without adequate tools.

    An Effective View of Forecast Accuracy

    The most comprehensive methodology for comparing forecast data and demand data is to graph the cumulative results for a given period which include upper and lower control limits (calculated from historic demand based on a normal distribution curve). The following example shows how easy it is to quickly identify when the demand is in control and when the demand is out of control.

    This graph shows a period of meeting the plan (Note A) and then a period when the actual revenues start to deviate from the forecast (Note B). At this point, the demand is still within the control parameters of "normal" demand and a cautionary watch may be put in place, although no action is required. However, it is clear at Note C the lower control limit has been exceeded and the "normal" expected demand is not being met. This is the time for action and the process to gain a complete understanding of the error should be invoked. Supplemental charts will be necessary to analyze what is causing the deviation. These charts are similar to the one shown above, but with a separate breakdown for units, average selling price, product types, sales channels, customer and sales agent.

    A possible result is finding that the graph represents a normal trend in the business with no corrective actions necessary. It is also possible that the total revenue versus forecast may be in sync (note A), however mismatches may exist in the unit, average selling price, customer or product type mixtures. In each case, this information would not be known unless this type of analysis were available as well as being in place for some time in order to understand the long term trends as well.

    Difficulties in Achieving This View

    A study of several companies reveals that four main issues cause the difficulties in performing the cumulative graph analysis on a real-time basis:
    1. Competing goals between the Sales / Marketing and the Finance / Operations groups.
    2. Inherent difficulties in obtaining a highly accurate forecast.
    3. Loss of forecast data visibility when it is converted to a production build plan.
    4. Inability to obtain a thorough view of the forecast exceptions in a real-time manner.

    Competing Goals

    The sales / marketing function is compensated by commission on revenues. It is a more preferable situation to have a greater supply than actual demand to meet those commission objectives. The Finance group pressures the Operations group to ensure that minimal inventories exist. Further, operations must be poised to react to change in several areas: material procurement, quality issues, build schedules, overtime and managing costs. This situation often puts these groups at odds with each other. Negative feelings build as time goes on. Each group begins to wonder if the other is competent. Sales / Marketing doesn't feel the pain when things go wrong in Operations, and Finance / Operations can't understand why Sales can't provide an accurate forecast.

    Inherent Difficulties in Forecasting

    An examination of the "forecast versus build" situation reveals that a very high emphasis is placed on obtaining an "accurate" sales forecast. However, without an equally important emphasis placed on obtaining a quantifiable assessment of demand trends, the "accurate" forecast is an unlikely outcome.

    An achievable process is one that provides Sales with the ability to perform adequate demand analysis so they can provide their "best" estimated forecast. Most often, the Sales / Marketing organization is in the best position to employ the most current information about the forecasted demand requirements, nevertheless, those requirements can change quickly today's economy.

    In reality, both groups must recognize that the forecast is the best understanding at that time and that there will be errors. The emphasis is to reduce the adverse impact. This is achieved by managing the forecast errors quickly and efficiently by using exception planning and real-time demand trend analysis. The key to success is to empower both groups with meaningful real-time information and business motivations for joining together in the corrective action process.

    Forecast Data versus Production Build Data

    Forecast data is typically shown by sales agent, sales channel, and customer, while the production build schedule is the summation of all individual sales forecast represented only by part numbers and scheduled units to accommodate most MRP systems. This is a significant factor in getting the two mentioned groups together. When Operations tries to inform Sales that there are "x units" of an excess part number, it is not clear as to how the forecast was inaccurate nor if the part number was from one or several individual sales manager forecasts. So who needs to take the action? Clearly, further analysis is needed to make the decision, but who will have time in either group?

    The issue is further compounded by several logistical difficulties in managing and manipulation of the data. Spreadsheets are often used; however they are inadequate for this degree of analysis. MRP, WIP and financial software packages usually do not include such analytical capability as their primary objective is to meet accounting requirements, to control user transaction screens and to integrate with other software modules.

    Recommended Solution

    The most significant obstacle is the lack of an effective process to segment and align forecast data with previous demand data on a real-time basis. Only if this type of comparative data is available will real-time corrective action occur. The Sales and Operations groups each require their information to be suitably broken down, but from a common data source, in order to facilitate mutual understanding and joint problem solving. Established relationships between past forecasts need to be fed back to the forecaster to correct their optimism or pessimism towards forecasting. Further, this degree of detail can provide Finance and Operations the ability to plan for revenues, costs, materials and production schedules.


    The forecast processes most often used by companies today usually result in higher inventories, longer customer order lead times, and poor customer delivery performance. The end result of these problems is an increase in overhead costs, and lower revenue due to poor customer satisfaction. In order to properly run the business, it is very important to generate and maintain an accurate demand forecast, based on a combination of historic data, statistical modeling, and an in-depth knowledge of customers and the products that they order. Further, some means of measuring the performance to plan that only displays error information for actual data that falls outside of the normal statistical variations to the forecast is needed to ensure that the forecast is indeed correct.

    Stewart~Frazier Tools, Inc. (SFT) develops and distributes Business Planning Tools and Decision support software applications. SFT has developed a software package called the Demand Planning Tool (DPT) that uses downloaded data from the Order Management system to assist in analyzing and manipulating information in order to make such detailed assessments. DPT is a Microsoft® Windows(TM) decision support software application with features specifically designed for analyzing and comparing the forecast with demand.

    A unique feature of DPT is the ability to merge multiple forecast inputs from several users. This distributes the process over the individuals who best know the sales situation. It also make the most sense with logistics for remote sales offices and departments.

    Back to S.F.T. Inc. home page

    ©1995 by Stewart~Frazier Tools, Inc. - all rights reserved

    Last Update: 7/17/95