The most talked about analytic application of business intelligence technology is customer relationship management (CRM). But ask a CEO what keeps him or her up at night, and meeting quarterly financial targets will more likely be at the top of the list than the nuances of customer relationships. While CRM is important to a business, the success of the organization is more than just keeping successful customer relationships ­ it's about making sure the overall business is operating efficiently.

How does a CEO make sure the business is running on all cylinders? His likely guidepost is the business plan or forecast against which he can constantly ask if the organization is on track to meet revenue goals. If it's not on track, he asks why it is not. The answer is likely to involve a combination of issues and departments that affect overall performance, not just the customer chain. Thus, calibrating business performance against plans is central to just about every operation in the company ­ procurement, manufacturing, inventory, logistics management and more. It is the communication link that binds sales, marketing, finance, operations and other departments. Updating forecasts helps to calibrate performance, enabling decision-makers to make the business- critical adjustments necessary to meet plan goals.

Ask that same CEO how often performance calibration is required, and the answer is always, "More is better." A monthly forecast is much more accurate than a forecast produced at the beginning of the year. A weekly forecast is even more accurate, so why not produce daily forecasts or updates triggered by important events? With greater forecast frequency, the CEO has an early warning system, allowing quicker response and a better night's sleep.

Business performance calibration (BPC) is the continuous, near real-time forecasting and analysis of related performance metrics to achieve balanced performance ­ i.e., efficient growth and the optimal management of resources. BPC, like CRM, focuses on actionable analytics and relies on the data warehouse architecture to provide the foundation. The extract, transform and load (ETL) process, which is a critical step to cleanse information maintained within the data warehouse architecture, provides users with confidence in a single enterprise-wide system for monitoring performance and avoids the garbage-in/garbage-out law of computing.

The enterprise data warehouse architecture is critical to this performance calibration process because it integrates data from multiple sources to create the most complete record of the enterprise's performance. An important and often overlooked feature of the data warehouse architecture is that it establishes a language standard to facilitate enterprise-wide collaboration. It enforces a consistent definition of products, services, markets and customers, and defines performance measures and key metrics. It also defines business management hierarchies, such as the alignment of sales territories and regions that support the business management process.

As user collaboration becomes common within the enterprise and between trading partners (B2B), a language standard becomes critically important. Because the data warehouse architecture enforces a language standard, it is the strongest foundation on which to build a forecasting analytic application. In environments where marketing, sales, finance and operations calibrate forecasts differently, this language standard helps companies move toward a one-number predictive system that can be used throughout the enterprise. This single-number concept is based on the assumption that numbers are calibrated to meet the needs of different users.

Predictive Analysis for Calibration

One could argue that this language standard/calibration process seems like a nice idea, but is really much ado about very little because most financial goals, which are functions of sales forecasts, do not change very often at the enterprise level. This view, however, is incorrect, as it is at lower levels of business management that decisions are made daily to ensure that corporate goals are met. How are the effects of these decisions calibrated? Generally, users start by producing updated forecasts, usually in spreadsheets, that can be used to calibrate the impact of various assumptions on performance metrics. The challenge, which BPC addresses, is to use the language standard of the data warehouse architecture to create more frequent updates and thereby improve forecast accuracy.

Improving forecast accuracy is a function of three things:

  • Increasing the frequency of forecast updates.
  • Seeking input from those closest to customers, and perhaps customers themselves, to get the most accurate reading of demand.
  • Employing agents, triggers and alerts to focus attention on the critical issues.

With Internet technologies, it is no longer necessary ­ and certainly not desirable ­ to limit forecast updates to a schedule. The forecast should be updated after each event that significantly impacts the business. For example, if a major retailer drops your company's product line, the company shouldn't wait until the next quarter's cycle to update the demand forecast. That forecast needs to be changed immediately to ensure manufacturing knows to adjust its production accordingly. Similarly, forecast accuracy is improved by involving frontline managers who influence the forecasting process. The sales representative who may have an inkling that the retailer may soon drop the company's products can play a key role early in the demand forecasting cycle. Additionally, the forecasting system must be intelligent, alerting users ­ such as the manufacturing manager ­ when there is a need to take action. Alerts can be used to request users to review a forecast or to notify others that the forecast has changed significantly.

Collaboration in Decision Making and Decision Implementation

Who should produce the forecast? Is forecasting a mathematical projection of sales or is it developed with user input? The reality is that frontline managers, customers and suppliers are better at quickly detecting factors that impact the forecast than are statistical models which detect impacts from an analysis of historical trends. While statistical analysis provides an excellent starting point in the business performance calibration process, these users need to add their collective wisdom to create the most accurate forecast.

For forecasts to be useful as a management tool, users must be able to evaluate the impact of forecasts on related performance metrics prior to committing to the forecast. Users must be able to perform the same type of "what if" analysis that they now perform in spreadsheets.

Business performance calibration is much more than a simple forecasting application. It provides spreadsheet-like "what if" analysis capabilities that allow users to interact with the data to determine the impact of their decisions. It allows the CEO to ask what would happen to financial performance if a major retailer dropped his company's product line in advance of that retailer's decision and decide to adjust or not adjust business processes accordingly. BPC frees users from the now ubiquitous practice of copying and pasting data from reports into spreadsheets to continue their analyses.

The example represents only one scenario from the demand direction. Supply problems can also disrupt businesses. For example, when a new product is delayed because of manufacturing problems, too often the call is to "man the spreadsheets," initiating a flurry of "what if" analysis to determine the impact on marketing programs, budgets and logistics. However, each rescheduling decision requires approval before decisions can be implemented. When this analysis is accomplished offline, the data warehouse is relegated to the role of mere data storage and transport.

Business intelligence must integrate the decision process with forecasting. Users must also be able to share the results of their efforts easily when calibrating their decisions, both to gain the necessary approvals and to notify those responsible for implementing decisions. In the future of business intelligence, the data warehouse architecture must become the foundation of a collaborative hub that supports decision making and implementation.

Inter-Enterprise Collaboration

CPFR (collaborative planning, forecasting and replenishment) is an excellent example of inter- enterprise collaboration. Grocery retailers track daily sales (demand) with scanners installed at the point of sale. Daily sales must be calibrated to determine order frequency and quantity ­ how much to order and how often orders should be placed. The re-tailer must balance the cost of inventory with the need to avoid out-of-stock situations. If customer demand is reasonably constant, determining the economic order quantity is simple.

However, manufacturers use marketing efforts and promotions to try to increase demand for their products. Retailers must predict these actions and decide whether to increase orders to meet demand.

The retail grocery industry, recognizing this imbalance between manufacturer and retailer goals, initiated an effort to define a process for manufacturers and retailers to collaborate and reach consensus on demand and order quantities. The result of this effort is the CPFR process specification. CPFR is a forecasting and interactive-analysis application that seeks to balance the goals of the manufacturer and the retailer in reaching consensus on order quantities. CPFR is one example of B2B collaboration that requires continuous business performance calibration.

Decision Latency

Internet technologies have reduced information latency. Many of us remember when it was necessary to wait a day for newspapers to deliver stock prices. Initially, the Internet reduced this information latency to 15 minutes. Now we receive near real-time quotes ­ in some cases delivered to wireless devices. The real breakthrough occurred when online information could be translated into online action ­ empowering users to execute trades online.

Businesses are following a similar path. Much of today's focus is on adapting Internet technologies to reduce information latency within the enterprise and between trading partners (B2B). In most organizations, the flow of information has certainly improved. Every day more information arrives faster. Once again, the important breakthrough will occur when business users can translate this online information into action and reduce decision latency.

Decision latency is measured in the time required to implement a decision. Several things are required to reduce decision latency, but one of the most important is to give users the ability to quickly calibrate the impact of several potential decisions. Decisions are calibrated by comparing forecasts to goals and modeling the financial impact of different scenarios. The most important requirement is to provide this capability near real time so that decisions can be quickly communicated to those who must take action to implement the decision.

Monitoring Performance Against Goals

When the data warehouse architecture is used to support business performance calibration, it becomes both a source of data to facilitate time-critical forecasting and a central place for forecasts to be stored, updated and integrated. The data warehouse becomes the single integrated source of information to monitor and analyze business performance. One problem is that perhaps the most important metrics often are not maintained in many data warehouses. These metrics, against which performance is generally calibrated, are the plans and forecasts used to measure overall company performance.

By managing forecasts and plans within the data warehouse architecture, a wide spectrum of performance- monitoring and predictive-analysis functionality is available for those who rely on the data warehouse. Performance against quarterly or annual goals can be analyzed and predicted. Rather than simply reporting inventory levels, inventory requirements can be predicted. Agent processes can be invoked to alert users on a "need for action" basis by comparing current forecasts to plan. For many people in the industry, this was the original promise of business intelligence ­ actionable analytics.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access