Author's Note: This is another installment in a continuing series on corporate performance management. Parts 1 and 2 were featured in the December 2003 and February 2004 issues of DM Review, respectively.
There are many different steps that can be taken to move a company toward a corporate performance management (CPM) solution. Most companies believe that once they have established and communicated performance objectives, defined key business performance indicators, and chosen to design and deploy a graphical interface for a performance scorecard or digital dashboard, they are well down the road to delivering CPM. This article will examine the idea that many of these early, exploratory efforts into CPM will turn into clear and considerable failures unless what lies beneath the CPM solution is a consistent and integrated data architecture. This article puts forth the premise that CPM's underlying business intelligence (BI) systems must be predictable. Unlike the plot of a recent movie of the same name, the data architecture that lies beneath a CPM initiative will be ineffective if it is full of suspense and surprises. What worked at the movie box office won't work for CPM. Let's examine why.
Most companies are actively seeking an effective CPM strategy. However, many questions present themselves in this quest including: What do we do first? How can we build on what we do today? What is the most effective path to take to where we want to go?
In order to begin to understand how to get the answers to these and other questions about CPM, it is important to first understand the CPM framework. CPM is not a slogan, a cliché or a technology. CPM is the application of fundamentally sound business management practices, enhanced by timely and accurate information, in order to effectively communicate, comprehend and control (C3) the performance of an organization.
Simply put, communication, comprehension and control define the exercise of business management over the resources of the organization in the accomplishment of corporate performance objectives. C3 is performed through the arrangement of people, equipment, facilities and operational processes employed by management to plan, direct, coordinate and control the enterprise in the achievement of its objectives. C3 builds upon the foundation of existing business processes, operational systems and BI applications to:
- Identify key performance objectives.
- Measure progress toward identified performance objectives.
- Provide the information necessary to communicate expectations and overall progress and understand exceptions as necessary.
- Comprehend the meaning of performance trends over time.
- Take action to control the organization in its efforts to achieving performance objectives.
Figure 1 presents a framework for understanding CPM, depicting the C3 capabilities as sitting above the existing business intelligence, operational systems and business processes of the organization. Once performance objectives and standard measurements are defined, the CPM initiative must enable the appropriate collection of data to report on progress, exceptions and anomalies for measurement, analysis and reporting. These measurement, analysis and reporting capabilities are the technology-enabled components of CPM.
Figure 1: Framework for Understanding CPM
Communication, comprehension and control (C3) define the exercise of business management over the resources of the organization in the accomplishment of corporate performance objectives.
What Lies Beneath CPM Really Matters
Most companies have adopted or implemented some early component of a CPM framework by virtue of the fact that the BI layer of the CPM framework is relatively mature, and early adopters have developed scorecards, digital dashboards and the like to present operational results to various segments of the corporate management community. The problem with some of these early forays into CPM is they are not able to deliver the C3 components of CPM - the communication, comprehension and control functions - primarily due to the fact that these performance management dashboards are based on a data architecture that is not fundamentally sound.
For example, the COO of a major energy company recently referred to the company's early approach of delivering scorecard results without the ability to examine underlying information detail as putting "lipstick on a pig." The systems and information infrastructure of this company was highly fragmented as a result of rapid growth through acquisition. As a consequence, the organization was starved for meaningful and integrated information. While the corporate scorecard project was initially a huge success with management, their inability to comprehend the information in the appropriate business context led to its rapid replacement by a project that would extend the supporting data infrastructure appropriately. In this example, the company needed to add capabilities for management to retrieve definitions and algorithms for performance scorecard elements, as well as the functionality to drill down into anomalous results to determine and swiftly act upon the causal elements of business performance outcomes.
The lesson learned at the energy company is that although CPM is not a technology, it relies heavily on technology components to deliver an effective solution. Perhaps the most important consideration for a CPM initiative is how well the company's existing systems (and data) infrastructure can deliver the required information in an accurate and consistent manner, from the appropriate business systems, at the proper points in time, for the measurement and monitoring of progress to corporate performance objectives. If the required information is not available, inaccurate or must be developed or modified in a manual fashion, a CPM solution will be of relatively little value because the performance results being communicated can no longer be understood in their proper business context. Performance results can't be analyzed or audited in a systematic fashion back to the source business activities and/or supporting systems to which the results are attributed. This is a very common mistake made by early adopters of CPM solutions.
Let's look at another example to explain these false starts at CPM in a bit more detail. A retailer wants to communicate, comprehend and control its business performance relative to sales and margin forecasts. This company believes it can implement a quick and effective CPM initiative to help do this. Several years ago, the company, originally comprised of several stores located in the northern region of the U.S., achieved significant growth by acquiring a chain of competitive stores located in the southern region of the U.S.
The northern part of the company extracts information from a sales management system, which is fed by its stores' sales transactions on a weekly basis. The extract data is then copied to a large number of flat files for ease of reporting. In order to forecast sales and margin, the retailer combines actual sales and margin information from several of these flat files, merging the data into a new file that feeds a third-party forecasting system. Manual adjustments are made to the forecasting system to reflect plans for changing product mix in future periods. Business rules to adjust for seasonality and other factors were programmed in the forecasting system five years ago. Unfortunately, there is nobody left from the original forecasting system's deployment team who can remember the rationale for the original business rules.
The output of the system is forecasted product demand at the product SKU/store/week level. Because merchandise planners are aware that certain product categories always seem to generate forecasts that are much lower than actual demand, the planners make manual adjustments to the file that is generated by the forecasting system. The planners also make manual adjustments to remove items they know will no longer be available from suppliers. These adjustments are made using a spreadsheet program. The final forecast for the northern region is summarized on another spreadsheet, at the store level, and sent to the corporate planning department. The overall process is depicted in Figure 2.
Figure 2: Northern Companies' Forecast Process
The retailer's recently acquired chain of stores in the southern region has forecasting capabilities within an integrated sales management system. Output from the forecasting module of the sales management system is transferred to spreadsheet format. The spreadsheet is at the product SKU/store/week level and is sent to merchandise planning. Once again, the planners make manual adjustments to the spreadsheet, summarize the demand forecast at the store level and send the summarized forecast to corporate planning. The process is represented in Figure 3.
Figure 3: Southern Companies' Forecast Process
The corporate planning organization is responsible for merging the spreadsheets for the northern and southern regions. The results of the final demand forecast are placed on the CPM dashboard. The dashboard now contains forecasted sales and margin, by store, by month.
The corporate planning organization also maintains another spreadsheet with the current year's sales and margin objectives, by month, for each store. These are also loaded to the CPM dashboard as depicted in Figure 4.
Figure 4: Example Retailer's CPM Dashboard
The company has now provided visibility for this year's corporate objectives for monthly sales and margin, by store, and is able to compare these objectives to current monthly sales and margin forecasts, by store. The first time these metrics are loaded, the retailer's CPM dashboard shows store forecasts are exceeding annual objectives. The company celebrates (albeit a bit prematurely) because they've now implemented CPM and the results look good.
"Bad data leads to bad business decisions. With corporate performance management, the risk of bad decisions is elevated to a more significant and strategic level than in narrower, more-focused business intelligence initiatives. Enterprises executing a CPM strategy must be diligent in analyzing the quality of key data early in their efforts, as well as retaining an ongoing focus on data quality. This requires involvement by the business units and IS organization to identify significant data quality issues, determine the business rules and processes by which they will be addressed, and implement the necessary controls and process changes to ensure that the data is suitable for CPM initiatives."
But wait! We're about to be reminded of what lies beneath ...
In the second month following implementation of the CPM solution, something seems to have gone terribly wrong. All the northern stores' forecasts are showing large downward trends in sales. As soon as the CPM dashboard is updated, a great deal of consternation is noisily expressed in management meetings and throughout the retailer's hallways.
Managers want to drill into the detail of the northern forecast to begin to comprehend what might be causing the anticipated slump in order to institute management control measures. Unfortunately, it is soon apparent there isn't anything to drill into!
Once management has exhausted the spreadsheet figures that were loaded into the CPM dashboard, there is no easy way to get back to lower levels of detail. Corporate management is having a fit, the northern store managers are telling everyone they never believed the numbers in the first place and the CPM dashboard implementation team has long since moved on to their next project.
Fortunately, a resourceful data manager with a great deal of credibility advises corporate management that this whole CPM dashboard initiative was just a prototype. After all, surely everyone recognized that the company really couldn't be successful at CPM without an underlying, integrated data architecture.
As it turns out, the data manager is absolutely correct. This retailer has not succeeded in deploying an effective CPM solution.
Let's review the C3 capabilities. Our retailer can certainly claim that they have succeeded in communicating sales and margin objectives by store. However, when there is significant variance between the adjusted store forecast and the original objective, the company has no means to comprehend the reason for the variance, and there is certainly no mechanism for management to take action to control performance. Because the forecasts are created using manual processes and detailed level information does not exist in storage formats that can be easily accessed, this is an absolutely ineffective solution.
How do you ensure that this won't happen in your company?
Assessment and Planning for What Lies Beneath CPM
There are two major components of CPM road map development: the assessment and planning cycle, and the deployment and execution cycle. These cycles, as depicted in Figure 5, are strongly connected.
Figure 5: Corporate Performance Management
A company should begin with an assessment and plan for CPM, followed by deployment and execution of measurement and monitoring processes, not the other way around.
There are critical issues that must be resolved in the assessment and planning phase for CPM. Aside from the obvious business issues related to the identification and prioritization of performance objectives and business drivers, there is a fundamental data architecture assessment and plan that must be performed. The company must determine how the CPM solution will be architected to continually measure and monitor the impact of key business processes, and develop a plan to mitigate significant risks posed by the data architecture that lies beneath CPM. If the retailer in our example had performed this initial data architecture assessment, it would have discovered the underlying financial data architecture was rapidly approaching information chaos as depicted in Figure 6.
Figure 6: Information Chaos
In Figure 6, which depicts what lies beneath the CPM initiative for our retailer, each arrow represents the movement of data further away from the operational systems of record. To make matters worse, each data flow arrow also represents the application of business rules, filters and formatting, most of which are manually applied. Each arrow compounds the risk for data integrity and accuracy to be compromised, and introduces the further passage of time as individual processes act to transform the data for their own needs. Each "can" in the data architecture represents another copy of, for the most part, the same data. For many of these data stores, any business "intelligence" resides only in the memory of a real person.
Our retailer started with two forecasting systems. Attributes that existed in both, such as SKU, were not consistent when they represented the same product. (Remember that the northern and southern regions were originally two different companies.) Hierarchies and summarization schemes were different in the two forecasts. Over time, information from flat files and spreadsheets was manually "band-aided" together in order for the company to perform functions for which the data was not originally intended, adding to the confusion and complexity of the underlying information, and resulting in the following problems:
Data is not easily accessible. Much of the data at our retailer resides in spreadsheets. Spreadsheets provide extremely limited flexibility for application access. Reporting is highly customized and difficult to support. There is a great deal of manual intervention throughout the forecasting processes, contributing to problems with accuracy and consistency. Spreadsheet templates are highly customized and difficult to support. Allocations and adjustments involve hard-coded calculations that are not easily adjusted for changing business requirements.
The architecture doesn't support the timely distribution of information for CPM. It would be difficult for our retailer to deploy any timely forecasting adjustments to the CPM dashboard because the current data architecture is very manually intensive. Much of the data is generated and maintained at the desktop. All of the spreadsheets require manual maintenance, a process that is highly prone to error.
The architecture is not integrated. The absence of data integration is apparent in the absence of any link among summarized forecasts, detailed forecasts and the actual sales data from which the forecasts were developed and sourced. This results in the inability to drill down, drill back or drill around for analytical and comprehension purposes.
The architecture is not scalable. For all of the aforementioned reasons, the current data architecture lacks the capacity to scale horizontally or vertically. Further growth by acquisition means the company would be required to integrate more actual and forecast data into the existing environment - which is difficult, at best. Expanding the functionality of the existing data stores would also be nearly impossible.
What Should Lie Beneath?
Incremental deliveries of tangible results for CPM must be supported by a systems infrastructure - applications, tools, technologies and data - that will support long-term results. Under-standing the extent to which these components provide the necessary information for the monitoring and measuring of performance results is an integral and important step in the planning and assessment phase in the road map to CPM.
The optimal data architecture that lies beneath a CPM solution and provides the maximum benefit is a single version of corporate truth. The good news is that if your company has implemented an effective BI infrastructure (see Figure 7), a single version of the truth already exists in the form of a data warehouse and dependent marts.
Figure 7: Effective BI Infrastructure
In a survey conducted by TDWI in the fall of 2003, 49 percent of those who had deployed a CPM solution and 57 percent of those who were exploring CPM solutions stated that a single version of the truth was key among desired benefits and drivers.1 In the same survey, greater visibility into the business was the top reason that organizations are deploying CPM solutions.
In order for the CPM solution to provide the necessary visibility into your business, and to ensure that a single version of the truth lies beneath, the data warehouse infrastructure components must be reliable and predictable.
The benefits provided by such an infrastructure are enumerated in many BI and data warehousing best practices. They include:
- The ability to apply a consistent set of business rules for the extraction, transformation and integration of source system data.
- Access to consistent and accurate information in an environment that promotes a "single version of corporate truth."
- The ability to support the delivery of timely information to business stakeholders.
- The integration of operational data in a manner that delivers meta data to provide an audit trail of source-to-target transformations through the application of documented business rules.
- A scalable architecture that can accommodate new transactional sources of data and increasing demands for information consumption throughout the company.
While the data warehouse may need to be extended to support some of the performance objectives and progress metrics for CPM, these gaps should be identified and addressed as part of the CPM road map process.
One way to ensure this happens is to use the approach outlined in Figure 8 for the CPM road map assessment and planning stages.
Figure 8: Approach to CPM
While there are many different steps that can be taken to move a company toward a CPM solution, what lies beneath the solution in the form of data architecture will have an enormous impact on its ultimate success or failure. CPM represents the application of fundamentally sound business principles toward communicating, comprehending and controlling business activities that contribute to overall performance objectives. While CPM is not a technology, a critical success factor is an integrated data architecture that provides accurate information, on a timely basis, that is required to continually monitor and measure progress made toward achieving business performance objectives. One way to ensure what lies beneath is not suspenseful or surprising is to include the data architecture assessment and plan in early CPM road map activities. Be sure you know what lies beneath your CPM initiative.
1. TDWI Report Series, "Best Practices in Business Performance Management: Business and Technical Strategies," by Wayne Eckerson; research sponsored by ThinkFast Consulting, May 2004.
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
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access