Forecasting: A Critical Link in Business Performance Management

Article excerpt

Today's rapidly changing business environment demands that organizations have a firm grasp of their internal operations while maintaining the responsiveness and flexibility to exploit market trends and opportunities as they present themselves. Demonstrating an understanding of these issues and an ability and willingness to respond to changes are critical indicators to Wall Street that the company's long-term prospects are in capable hands. The ability to quickly and accurately align and realign resources with corporate strategies has become a necessity. With this in mind, business is starting to awaken to the growing need for Business Performance Management, both from a process standpoint as well as a technology perspective.

Integral to Business Performance Management success is a robust approach to forecasting. Forecasts--both revenue and cost--are the starting point for managing expectations both internal and external to the organization. Recognizing this, the organization's continued or expanded adoption of statistical forecasting and alerting technologies will drive more efficient planning and budgeting processes while keeping the resulting work products realistic, based on underlying trends not immediately obvious to the organization. As accuracy of forecasts improves, so does the likelihood that strategy implementation improves because businesses are acting on numbers and trends that better reflect the likely outcomes of the actions they have undertaken.

Specific benefits of accurate forecasting include:

* Reduced likelihood of earnings surprises

* Market change responsiveness

* Improved decision-making

* Reduced time spent on non-value-added activities

* Organizational "reality checks" on revenue projections and cost budgets

* Proactive identification of future consequences of current trends

Wall Street Worries--Earnings Surprises

"Improving the quality of the original earnings expectation or forecast is an obvious first step in attempting to avoid earnings surprises." (1)

This "improvement" is obviously easier said than done. The starting point, however, is to have a basic understanding of how you will improve the earnings expectation. In short, if you had a forecasting tool and an alerting technology, how and where would you deploy them? Warning signs of possible earnings surprises exist both inside and outside the organization. The decision to deploy technology to help uncover these warning signs must be based in part on some sort of cost benefit analysis. There exists a vast array of symptoms one could scan for in the larger environment, such as natural disasters, disruptions in raw material flow, or unexpected economic activity. In most instances, trying to build monitoring and forecasting technologies aimed at these external environmental factors into your organization's Business Performance Management application will yield questionable returns when compared to simple "paying attention" by people within your organization.

The reason for this is that there are generally no readily available data feeds capable of getting the appropriate data into an application in a timely enough manner to enable meaningful automated data analysis. In simply deciding what data to feed in and scouring the environment for the data itself, you will generally find the answers you are looking for. Further analysis may provide a certain elegance, but it will probably not yield significant additional value.

The more appropriate uses of forecasting and alerting technologies are to point them inwards, towards operations and specific financial indicators wherein trends may not be immediately obvious to management. In these instances, statistical forecasting can add value in excess of the cost of gathering information; therefore, application of the technology is a wise investment. Taking into account the need for raw data from which to generate business intelligence (that is, knowledge), the two most productive places in which to deploy forecasting and alerting technologies are the areas of profitability and liquidity.

From a profitability standpoint, gross profit margin decline should be the number one area under examination. The reason for this is, first and foremost, the sheer amount of data that is available for analysis. As was described earlier, there are a myriad of possible indicators for which to search, but in the deployment of technology we have to come back to the cost/benefit analysis of obtaining the information. In analyzing your internal trends for gross margin decline, you should pay attention to the following factors:

* Selling price trends

* Product mix changes

* Excess capacity growth due to volume slow down

* General cost increases

* Currency movements

When using liquidity measures as early warning indicators, the primary focus of your technology solution should be on the following factors:

* Inventory days

* Receivable days

* Payable days

Information about these factors is generally available in most large organizations and the loading of this data into a Business Performance Management application should be a fairly straightforward proposition. This being said, we still need to pass our measures through "qualification screens" to make sure we get knowledge and not noise.

Early Warning Metrics--Not All Are Created Equal Metrics abound, and not all are created equal. At a minimum, your metrics should meet the requirements of timeliness and length of response interval.

Timeliness, by definition, requires that the measure must occur with sufficient time to respond. Furthermore, you need to know how you plan to respond once you receive a timely alert. Measures that coincide directly with an earnings decline are essentially useless. In many ways, real time is as useless as hindsight because it is too late to do anything other than react in the heat of the moment. What is needed instead is information with which to plan. The timeliness hurdle will vary with the company and the industry so there is no one correct answer. The key is to monitor this aspect of the warnings and work towards improvement. You may well start with 20 to 30 early warnings, which is fine, but the real payoffs will come by studying those warnings with the intent of zeroing in on the 5 to 10 that apply to your organization.

The second consideration when developing your early warning system is response time interval. As mentioned above, receiving warnings as the situation happens gives you a zero response time. After you pass your warnings through your timeliness screen, you will probably not yet have identified the 5 to 10 most useful warnings. To get to your final "warning set," monitor the length of time prior to expected major problems and focus further on developing analysis methodologies and action plans around those warnings with the longest warning interval. Not every warning will have a preplanned response but you should strive for a preplanned response process. Those warnings with sufficient warning intervals and developed response processes can be delegated out into the organization. The remaining timely warnings with shorter response time intervals can then be left to the responsibility of upper management. In this manner, you allow upper management to perform its real charge, which is making critical decisions for the company.

Finally, with the metrics identified, the clam feeds automated, and the forecasting and alerting technologies implemented, management can now scan the details automatically, looking for trends that could cause trouble weeks, months, or even quarters out. With this insight, they will have adequate time to respond to troubles with corrective actions that get the organization back on track towards achieving its stated strategies.

Planning and Budgeting--Forecasting Yields Operational Advantages

Too often the budget process is kicked off with arbitrary seeding based on some fixed percentage of last year's budget or actuals (provided they are available). Arguably, this is one way to get a quick start on the process. The problem is that this quick seeding neglects trends within the actuals and the correlation between revenue drivers and last year's budget. A superior approach is the use of statistical forecasts as a starting point for seeding the budget.

Recognizing that a significant number of costs within the organization are driven by revenue projections, the first place to apply statistical forecasting would be to those periodic revenue projections. Statistical forecasting allows us to combine recent revenue results with knowledge of natural product/service life cycles to create bottom-up revenue projections that factor in these cycles, which may or may not be apparent to the end user. This approach is vastly superior to traditional top-line revenue projections that ignore the potentially hundreds and maybe thousands of detailed projections of which they are composed. Consider the following example (Figure 1) of dealing with only two products at different phases of their life cycle.

If just the top line (orange) were used to predict ahead, the revenue projection would be a straight line going into the future. However, if the details (red and green lines) were projected separately and then combined, the result would be a different overall projection. The true projection would be one sloping upward as the "red" product reached the end of its life cycle while the "green" product continued to grow. As this example illustrates, an objective, impartial method is needed that can detect whether each detailed forecast is reliable, whether it fits the seasonality and adoption life cycle of the product. Only when all of the detail is reliable--and considered--can the consolidated result give any degree of confidence: confidence that the revenue projections are realistic, and confidence that the resulting budgets and forecasts built from these projections are accurate. Deviations, both positive and negative, can then be researched to ensure they are supported by material changes in the business assumptions used in planning. Coupling statistical forecasting with human judgment in this manner increases the likelihood that the resulting budgets and forecasts realistically reflect business expectations.

On the expense side of this equation, forecasting takes on an equally important role. Specifically, statistical forecasting helps ensure that overhead budgets--those not specifically driven by revenue projections--are realistic by guarding against "sand-bagging" in the expenses. Applying statistical forecasting to historical actuals and budgets that exist in the database allows us to quickly and objectively project out expenses for the coming period. These projected expenses can then be compared directly to the bottom-up budgets prepared by the budget holders to help uncover anomalies in the data that need to be addressed in the budgeting process. Deviations from the statistical forecasts are not necessarily problems, but the explanations of the deviations should tie into material changes in operating procedures and/or strategic plans to make sure the changes are warranted and appropriate.

Deploying a statistical forecasting tool in the above manner leads directly to a more efficient and accurate planning process. It is efficient because technology is substituted as the first means of routine plan population, thus allowing staff to focus on the exceptions and understand the business implications they present. It is accurate in that we start with just the facts and sidestep organizational sand-bagging and chest-beating in determining the baseline forecasts and budget targets. Additionally, getting closer to the real plan the first time through can reduce the need for later rebudgeting because of unrealistic plans.

Reporting and Analysis--Statistical Forecasting For Added insight

Integral to plan achievement is the ongoing monitoring of results as time passes to confirm the plan is achieving its goals and to alert us to opportunities or problems that have arisen since the plan was put into place. As part of this monitoring process, organizations often perform periodic forecasting exercises focusing on future revenue and cost expectations based on trends they are experiencing in the marketplace. Inherent in this process are three potential flaws:

* Unwillingness to acknowledge problems until it is too late to take corrective action.

* Inability to accurately recognize positive or negative trends until considerable time has passed, resulting in the failure to capitalize on opportunities or the unnecessary expansion of problems.

* Overly optimistic trending, resulting in over- or under-responding by the organization.

Each of these three phenomena provides an excellent opportunity to deploy statistical forecasting techniques combined with advanced alerting technologies to help the organization minimize negative results and/or capitalize on opportunities in a timely manner.

Take, for example, the first issue, wherein a manager is reluctant to call out a problem because it is felt the situation is under control. While an admirable thought, executive management needs to be alerted as soon as possible so that they can concur with or counsel the operational manager in their response. With the use of statistical forecasting, executive management can set early warning profiles for acceptable and unacceptable deviations from revenue and cost projections. Independent of the personal judgment of operational staff, alerts can be set to notify all interested parties when future forecasts, based on actual results, materially deviate from expected plan results. In this manner, results can he forecasted two or more months out and compared against expected results. Substantial deviations can then be called out to management with enough forewarning of potential problems or opportunities so that there is sufficient time to respond and thus avoid under-or overreacting.

This same scenario could play out for the second and third potential flaws listed previously. Set to run automatically on a weekly, monthly, or other regular basis, statistical forecasting coupled with advanced alerting technologies helps us overcome common organizational shortcomings, giving us a better chance at recognizing the need for adjustment sooner and increasing our odds of successfully dealing with the changing marketplace. We get critical information to senior management in a timelier manner, thus enabling them to formulate specific tactical plans to achieve the strategic goals not currently being met, or that, according to the forecast, will likely not be met in the future without intervention.

Role Of Technology--Evaluating Forecasting Solutions

By combining modern statistical forecasting techniques with alerting technologies, complete management systems can be built that will improve the implementation of strategy. Towards this end, there exists a multitude of forecasting solutions in the marketplace today. In deciding what is best for your organization you should consider the following issues:

* Ability to automate the generation of plans and forecasts. Without automation you are tied to manual data entry, which leads to keying errors and additional non-value-generating activities.

* Integration with alerting technologies. Alerting technologies allow you to focus on issues instead of searching for them. Forecast automation coupled with alerts enables "predictive alerting" which greatly increases process efficiency. An example of this is the ability to, following month's end, automatically generate a forecast, compare it to future plans, and then alert management to potential problems that may occur in the future without timely intervention.

* Selective rebudgeting. Systems should be able to guide management to specific accounts that need to be rebudgeted. This saves time and effort, as only those accounts or units at risk need to be replanned.

* User-friendly interface. The most likely users of a forecasting application are your front-line personnel. If the product lacks an intuitive and friendly interface you will have to add specialized staff, thus placing a resource constraint on your ability to easily access forecast results.

* Too many features. The 80/20 rule applies here. Eighty percent of the value of forecasting is found in a few very basic statistical forecasting functions. While it is enjoyable to watch advanced forecasting techniques at work, the added complexity in their execution makes them a business non-reality for most organizations. In short, don't pay for what you don't need and don't get so complex in your solution that the very people you wish to empower are too intimidated to use the tool.

* Too few features. There exists a happy medium between too few and too many features. Look for a tool that utilizes a variety of statistical forecasting techniques and easily guides the end user to the method best suited for their situation. At a minimum, the tool should give simple statistics as to fit, spread, and confidence levels.

* Overly complex analysis. The goal is to provide actionable information that people trust. Have the system evaluated by people without advanced statistical degrees. The business people in the organization have to use the system to get real value from it. Overly complex calculations that intelligent people have trouble understanding will lead to system dormancy, and the system adoption rate may end up being too low to give you your expected returns.

Conclusion And Next Steps

To improve the implementation of strategy, organizations must become more effective and efficient in their operations. They must be effective in that strategies are better linked to operational activities, and problems are sighted earlier and addressed more quickly. They must be efficient in that the activities related to implementing strategy take less time and have fewer iterations, thus allowing the organization to get on with their business of generating shareholder value.

By embedding modern statistical forecasting techniques and alerting technologies into the new breed of Business Performance Management applications, organizations can create management systems to maximize strategic goals and provide early warning of exceptions or problems. The benefit is that organizations will improve the implementation of strategy by:

Having More Effective Plans

* Budgets and forecasts are based on natural product/service life cycles which may not be apparent to the end user.

* The effect of budget game playing is reduced by challenging exceptional variances.

* Unbiased reality checks are provided for forecast submissions.

Having A More Efficient Planning Process

* The generation of plans and forecasts is automated based on statistically accurate models, which greatly reduces manual data entry tasks

* The number of budget iterations is reduced by getting accurate submissions the first time

* More accurate budgets reduce the need for later rebudgets caused by unrealistic plans, and ensure budget preparation time is not wasted

Avoiding/Minimizing The Impact On Strategy Of Future Problems

* Early warning of problems or opportunities makes you better able to control the future by taking preventative measures before problems hit hard, or by developing action plans to maximize opportunities.

For corporate survival, organizations must continually seek to improve the accuracy of their forecasts. In moving forward, follow these steps to create value and gain competitive advantage:

1. Model your own historical data to assess the predictability of your results.

2. Check long-term and short-term predictability.

3. Identify early warning metrics and then screen for timeliness.

4. Prioritize and delegate early warning responses based on the expected response interval.

5. Assess your own budgeting process. How much time would be saved if the first pass could be generated statistically? How many budget iterations could be eliminated?

6. Privately have an honest discussion of the time spent playing games during the planning and budgeting process.

7. Armed with a better understanding of your organization's needs, investigate Business Performance Management systems that provide statistical forecasting and alerting capabilities.

Geac is a global enterprise software company for Business Performance Management. To find out more about Geac products or services, visit www.performance.geac.com, or call 1.800.922.7979

(1.) Mulford & Comiskey, Financial Warnings (Hoboken, NJ: John Wiley & Sons, 1996), p. 2.

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.