• HOME
• SPEAKERS
• ABSTRACTS
• AGENDA
• TRAINING
• SPONSORS
• HOTEL AND TRAVEL
• INTEREST LIST
• Register Now!

Enter a friend's e-mail address:

Download our brochure (.pdf).

Abstracts

This page will be updated as we receive speaker abstracts, so visit often. The following abstracts have been provided to-date: For years forecasters have been led to believe that their primary purpose is to provide senior management with accurate point estimates based on past sales history. Meanwhile, senior management's primary focus is on growing the profitability of the business, which requires detailed business analysis of the factors that driver consumer and customer demand, not necessarily reflecting past business results. The chances of management or anyone for that matter using forecasts based on past sales history alone is almost zero. At best we can only hope to influence the business plan with our forecasts. The combination of statistical analysis and domain knowledge (not judgment) are required to develop an acceptable sales forecast that will ultimately drive downstream business planning activities. As business analysts, not forecasters, we need to focus on defining and measuring the key business indicators that impact sales and revenue, and provide senior management with "actionable decision support analysis" that drives the business planning process. Only then, will we forecasters, now business analysts, be able to provide value and insights to the forecasting process. This situation and others will be the focus of this presentation. Sales planners have the awesome task of how to appropriately assign and set sales quotas for sales accounts and these planners need to be equipped with the latest and best that forecasting can offer. Our case study shows the business case of what these planners needed here at HP and how we delivered a custom analytic service to forecast unit shipments for two product groups by each account and aggregated by industry vertical. The need was to develop a forecasting model that would

• Find a method to detect seasonality of two product unit purchases in a group of desired accounts
• Find out when the next large purchase might be and to forecast the next future 4 fiscal quarters, and
• The unit forecast should be at the account level, combined by industry vertical and have 90% confidence, and output these forecasts with custom graphics/shading so that they can be clearly understood by the business.
• Deliver account forecasts in Excel using SAS Add-in for Microsoft Office.

This presentation focuses on the role of statistical forecasting in demand planning. The question often is: When can I rely on the statistical forecast, and when do I need other information to develop a demand plan. Some companies rely solely on a statistical forecast, and then struggle with demand plan accuracy. Other companies avoid statistical forecasting and do not take advantage of its strengths to improve demand plan accuracy. Case examples will be provided to illustrate the strengths and weaknesses of both approaches. The United States economy and overall international trade is booming placing the entire North American transportation infrastructure under stress. An important link in the transportation supply chain is railroads.

Operating railroads is an extremely complicated and resource intensive exercise. The Union Pacific Railroad is the largest railroad network in North America providing rail service to a variety of customers in 23 states encompassing the Western two-thirds of the United States running over 32,000 miles of track. Matching railroad resources to customer demand is as challenging as operating the railroad network.

In early 2006, Union Pacific engaged the Power of SAS to help forecast customer demand. All of Union Pacific's six-business teams use SAS to help forecast business demand.

Our presentation highlights the challenges the railroad faces in integrating SAS into the railroad business planning process. Revenue agencies in most states are responsible, to some extent, for forecasts of state revenues. In addition, many revenue departments are responsible for providing estimates of the fiscal impact of proposed legislation (fiscal notes). These forecasting exercises present challenges, including the lag times between the availability of the most current data, data that are subject to large and frequent revisions, frequent changes in tax policy, and occasionally proposed policies that have no historical precedent. Fiscal note estimates represent a special challenge as they often need to be completed very quickly, often in less than 72 hours, and they are often accompanied by dozens of other fiscal note requests. In this presentation, I will discuss these forecasting issues more detail, and outline some of the strategies developed to deal with these challenges. A technically sound forecast goes nowhere if it is not accepted by those in power, whether they are politicians or upper management. Often the forecast gets manipulated to satisfy political ends or meet targets and plans. This presentation tackles some tough issues that organizations face, such as:

• How can we minimize organizational biases that impact forecasting?
• How can we improve collaboration among players with conflicting interests?
• How can we achieve buy-in or acceptance of the forecast by those in power?
• What's the best organizational design for a neutral forecast process?

We will discuss six simple strategies to overcome these problems and allow the technically sound forecast to prevail over politics. Even if you could create a perfect forecast, would your business take the right investment decisions?

What constitutes a quality forecast?

• Identifying the assumptions – creating visibility of the key drivers of the forecast
• Developing a most likely forecast – applying an appropriate forecasting models and validating results
• Estimating uncertainty – identifying a range of possible outcomes

What drives good decision-making?

• Differentiating between forecasting & planning
• Assessing Risk – different business processes carry different risks
• Planning – taking risk-mitigation into account
• Accepting Risk – taking investment decisions, setting targets & expectations
• Monitoring – tracking progress and assessing whether the assumptions are still valid

Company personnel receive little, if any formal educational training on new product forecasting, and literature on how to compose a new product forecast is scarce. Much of this literature tends to portray new product forecasting as a statistically sophisticated approach, whereas most individuals rely almost exclusively on judgment predicated on assumptions. Therefore, new product forecasting should be viewed as a process of assumptions management.

This presentation will discuss what is meant by assumptions management and how a meaningful new product forecast can be derived through systematic assumptions management. Company examples will be presented to illustrate the assumptions management concept and offer directions for how a company can implement a meaningful new product forecasting process. What happens when the usual forecasting techniques are not working? What are the ramifications when historical data no longer represents current and future realities? As Risk Managers, we usually rely upon a basic set of tools to determine the historical relationships that we then project forward to drive forecasting. In the current sub-prime Mortgage business, there are limited historical data that accurately captures the current environment, making forecasting an extreme challenge. I plan to identify some key issues and some options to address forecasting under an environment where we cannot rely solely upon the data. Forecasting is central to strategic growth of an organization. What are the best practices to make it a success? How to ensure this process is efficient, that it really reflects the reality of the business and that it is a credible asset with a great added value?

It is easy to overkill the process and fit a model to everything? How to see that you overkill the process? and conversely, how to see that your model is too rough? Some easy business tests can be performed to ensure the right balance.

Data mining is a fairly recent technique in real life applications. How could it be integrated in the process? How to select the software that best serve your organizational needs? How to make it an asset for your organization? Over the past twenty-five years, much has changed in the practice of business forecasting. Methods have improved, but the use of new forecasting approaches and tools have not become as widespread as one might have expected. The spreadsheet is still the most ubiquitous and widely used tool among forecasters. While microcomputers have greatly streamlined the forecasting process and improved efficiency, forecasting system development and implementations still lag behind capabilities. Hans will discuss the history of these developments from a personal viewpoint gained over the past two decades as a practitioner, teacher, writer and consultant. He will then give a more current perspective on best practices in forecasting and give his own prognosis where he thinks we are heading. There is a tendency to believe that results published in scholarly journals are true. Systematic investigation in the accuracy of published results in economics journals reveals that replicable research is the exception, and not the rule. Examples of non-replicable research from other disciplines will be given. Sources of non-replicability include bad software, sloppy programming, and lax journal editors. Prescriptions for remedying these defects are given. Over the past 25 years, we have benchmarked the state of sales forecasting management in over 500 companies. This presentation will provide insights into effective sales forecasting and demand management (including techniques, systems, and management approaches), the enablers and impediments to accomplishing world class sales forecasting management, and the benefits created by successful implementation. In this session, participants will learn about

• real-world examples of successful sales forecasting and demand management;
• factors that must be in place for both to be successful;
• common problems that impede their success; and
• the benefits that result.

Ever spend hours fishing the web and come up empty? According to Bright Planet, "Two-thirds to three-quarters of all [web] users cite the inability to find the information they seek as one of their primary frustrations". This presentation will help you avoid that frustration by illustrating how to fish the web and quickly catch the information you really want. The first part illustrates how to best use search engines like Google. However, such search engines see no more than 2% of what is on the web, just the surface of a very deep ocean of information. Most of what forecasters want to know is in the deep web, well below the surface. This presentation shows how to find a wealth of information about the past, present, and future for key U.S. or world forces influencing any product's demand and input costs. Topics are illustrated with on-line hits, screen shots and listings of selected URLs. A brief overview of a call center environment will be covered followed by descriptions of the forecasting methods used to support the environment. Although models and data issues will be covered, the presentation will go beyond standard analytical methods employed. Focus will be given to judgemental forecasting stemming from unknown initiative impacts and business partner influences. Finally, methods for successful forecast communication, negotiating political obstacles, and the ideal fit of a forecasting team into a greater organization will be covered. After a long history of subjective, bottom up forecasting on many fronts, purchasing, supply chain, commercial, and finance, Dow is now progressing to statistical forecasting. Implementing statistical forecasting is not just a software play. People, processes and technology, in the form of data and software, all have to interact together to be successful. Infrastructure details and examples of successful forecasting projects will be shared. Forecasting new products or products with very short history often poses a problem for traditional forecasting systems. The predicted demand tends to be either significantly over or under forecast. Furthermore, some of the implicit assumptions behind these commonly applied new product models often miss fundamental business rules and trends.

In this talk, we'll describe typical new and short history product forecasting models, highlighting their shortcomings. We will discuss some methods for more effectively predicting new product demand, capturing inherent business conditions. Additionally, we'll present the challenges with implementing these models in practice. A key component for understanding the competitive landscape is the identification of trends and their impacts. Trends are the underlying phenomena behind the products we use, the media we experience, and the strategies that we employ. If trends are misread or not well understood, then paradoxes of business strategy can occur. In extreme cases, everything that a business knows can be wrong. In other cases, business strategy that no one believes will work somehow does. This session will present key methods and modes of strategic planning that identify trends as well as emerging paradoxes in business models. Through examples, the session will explore how firms that missed changing trends have floundered and how innovative firms that embraced unconventional trends have prospered. Many organizations prepare forecasts using time series methods exclusively. These methods perform well when there is a great deal of continuity between the past and the future; however, they do not perform well in the presence of promotions, business interruptions and other aperiodic events. In these instances, many forecasters resort to adding extensive judgmental overrides to the time series forecasts or to "rewriting" the demand history to remove the impact of the events.

Substantial improvements in forecast accuracy and efficiency can often be obtained by applying quantitative approaches that model the impact of these events directly. This session will overview two approaches to this problem - dynamic regression and event models. We will discuss their strengths and weaknesses and emphasize best practices, using real data from a variety of different corporations to illustrate how the two approaches are applied. Forecasting, as a distinctive field of research, had its infancy in the early 1980s. Between 1980 and 1985, new associations, new journals and new conferences were created to give forecasters dedicated forums for their work and opportunities to network with others in the field.

Now 25 years later, we may ask about the results of these initiatives. What lessons have we learned that should guide us in improving our forecasting methods and processes?

I will talk about the lessons learned in these three categories:

• principles for generating forecasts (i.e. modeling)
• principles for managing and improving the forecasting process
• principles for measuring forecasting accuracy and setting accuracy goals.

I conclude the talk by discussing how we can more effectively present new forecasting research to those who would like to implement good practices. Executive S&OP is a powerful tool that enables a company's Executive Team to establish -- in advance -- targets for customer service, inventory levels, and customer order backlogs and then to manage the business proactively to achieve those goals. It helps companies to balance demand and supply, to integrate operational and financial planning, and to make demand/supply decisions proactively. Done properly, it provides an invaluable "window into the future" and can have a transformational effect upon a company.

Tom Wallace has been involved in Sales & Operations Planning for the past 20 years and has written three books on the topic. He will share his knowledge of:

1. what the process is
2. where it fits
3. how it works
4. the benefits that accrue to companies who implement it successfully, and
5. how to make it happen.

Tom worked with dozens of companies that have implemented the process successfully, and will share some of these experiences — and also some examples of companies who failed to make it work. Maybe your company has only dabbled in forecasting and you want to build a real operational demand planning process. Maybe you have operational forecasting up and running, but you've reached the limits of what you can do without collaborating with your customers. Maybe you believe that S&OP would finally close the gaps between planning and execution. Somehow, you've reached the limits of what slow evolution and gradual improvements can do to enhance your forecasting capabilities. To significantly improve performance, it's time for a big step forward, a dramatic improvement, an investment in the future.

In this presentation, we will share lessons learned from companies that have taken on the big project:

• How was the case for change made to management? How was resistance overcome?
• What technical and process prerequisites were, or should have been in place?
• How did forecasting software fit into the process?
• How were expectations managed? How should they have been?

To identify the root causes of sales forecast variance relative to actual sales, and to aid in the design of revenue stimulation strategies, a leading U.S. financial services company developed a highly segmented "bottom-up" sales forecast model to augment their existing revenue forecasts. A distinctive attribute of this sales forecast model is the linkage to customer lifecycles; the acquisition-growth-migration-attrition cycle common to consumer-facing industries was explicitly designed into this model. This case study presentation will include a description of the econometric and time series techniques utilized for each component of this model, including a discussion of the challenges related to the segment-driven data structures (and required software functionality) to deliver tuned and calibrated forecasts on a repeatable basis.