INTRODUCTION TO MEASUREMENT AND REPORTING SYSTEMS
Since 1494 with the appearance of the double-entry accounting system, developed by Pacioli, those involved in business have attempted to measure business performance in an organized manner. As many accounting functions are repetitive in nature (payroll, inventory, etc.), accounting was one of the first business disciplines to which early computing technology was applied. Today we see comprehensive enterprise models that have been incorporated into ISO Standards in an attempt to build quality, capability, and uniformity into business enterprise systems. Supporting these models and systems is an effort to also launch the Extensible Business Reporting Language (XBRL), such that metadata models gain uniformity and make business information more readily accessible across systems and enterprises.
This article addresses a concept for transitioning from an “end of period” reporting model to one based on “push agents” delivering to the pertinent manager information that is key to managing the enterprise in near real time in order to gain substantive competitive advantage. The high-level model in Figure 1 demonstrates the suggested movement from an “end of period” model to an automated push agent model, with an intermediate step already utilized in some enterprises, that is a “dynamic query model.”
Comparing where we are today in business reporting to that of network reporting, Computer Associates, perhaps the first to offer comprehensive, automated agents that forewarn of impending network trouble introduced some time ago neural agents that measure current states as they change against historical databases of past network activity in order to discern conditions that may be reoccurring and are similar to those that caused problems previously. Such detections may involve likely equipment failure to circuits becoming overloaded, with automated warnings and recommendations as to corrective actions being sent to the appropriate manager via the chosen message system (e-mail, voicemail, paging, etc).
Such “heads-up,” automated, near-real-time reporting of impending conditions permits network operators to be proactive in addressing problems before they occur. Such real-time network feedback requires an enterprise to develop “key performance indicators” (KPIs), or key measurements that it wishes to track against to plan and run the enterprise. Here, many products are on the market that address the identification of such KPIs and go by the names of “Cockpit Charts,” “Digital Dashboards,” and “Balanced Scorecards.” The secret of the balanced scorecard and the reason it has gained such wide acceptance is primarily due to the fact that it allows organizations to reach their full potential by putting strategy—the key driver of results today—at the center of the management process in organizations facing uncertain equity markets, an accelerating pace of change, and increased expectations for productivity and results. The comparative characteristics of reporting models are listed in
Table 1.
Instead of waiting for end-of-period reports, critical measures can be monitored in near real time as a function of the KPIs assigned to each measure, as seen in Figure 2.
Figure 1. Migration of reporting models for competitive advantage
Table 1. Comparative characteristics of reporting models
Figure 2. Simplified intelligent agent push model
BENEFITS OF EMPLOYING A BUSINESS MANAGEMENT PROCESS INTELLIGENT AGENT TOPOLOGY
Erik Thomsen, Distinguished Scientist at Hyperion Solutions Corporation, defines the term “agent” as:
“…a solution-oriented ensemble of capabilities including natural language processing, autonomous reasoning, proactive computing, discourse modeling, knowledge representation, action-oriented semantics, multimodal interaction, environmental awareness, self awareness, and distributed architectures.”
He describes the following five areas where the potential impact of intelligent agents on the logical functionality and physical performance of traditional business analytic systems can be positive (Thomsen, 2002):
• Agents should help move business intelligence (BI) from being application centric to being truly process centric, and provide a single point-of-access to distributed information. This is operationalized in intelligent agent solutions by self-description of individual modules that is made available to the system of agents as a whole, and a user’s personal agent being able to query all librarian agents responsible for various data sets in an organization for specific information.
• An active dialog is needed between the software and the user to seek out and learn user wants, and be able to anticipate/predict user wants in the future. Thus in addition to tailoring client layers to individual users and enhancing BI applications with options and preferences, the software/intelligent software agent plays an active role in querying the user.
• As attention shifts to business process as much as data states, the number-centric BI applications will have to provide more integrated text and multimedia handling.
• Intelligent software agents can provide personal analytic “coaching” for higher-level business processes by observing, learning about, and interacting with users. Agent applications within an overall BI/Business Process Management (BPM) framework can be deployed to encode horizontal and domain-specific analytic knowledge.
• The server side of most BI applications is very complex due to the changes in the load patterns/ factors and the wide range of physical tuning options. Intelligent physical optimizers can evaluate their own physical organization and, optionally, interact with an administrator before performing any reorganization.
A survey by Lederer, Mirchandani, and Sims (1998) indicates that businesses adopting Web-based information systems (WISs) responded that the most important benefit of being on the Web was to “enhance competitiveness or create strategic advantage.” Intelligent online agents can help simplify complex Web environments acting as artificial secretaries (Maes, 1994). Nwana et al. (1998) define agents as “software entities that have been given sufficient autonomy and intelligence to enable them to carry out specified tasks with little or no human supervision.” There is no standard definition for intelligent agents, but they are generally described (Decker, Pannu, Sycara & Williamson, 1997; Specter, 1997; Hendler, 1996) as programs that act on behalf of their human users to perform laborious and routine tasks such as locating and accessing necessary information, resolving inconsistencies in the retrieved information, filtering away irrelevant and unwanted information, and integrating information from heterogeneous information sources.
Business intelligence software using intelligent agents and BI technologies such as data marts/warehouses, data mining, Web site analytics, modeling and predictive tools, and data visualization tools can help analyze customer data to make connections that can boost sales, increase efficiency, and retain customers. The BI server connects to the data sources and provides analytical services to clients, which access the BI server through a Web interface or other desktop client. An application server typically provides the Web interface and often runs on the same server as the BI software.
The key benefits of employing a business management process intelligent agent topology are as follows:
• Timeliness of reports
• Automated measurements against planned goals
• Unfiltered information
• Delivery based on the end-user’s preference
The key features that should be analyzed when evaluating BI reporting tools are as follows:
• Analysis: The ease in creating reports and the end-user’s ability to manipulate the report once distributed.
• Formats and data sources supported.
• Data access: Preferably provide data-attribute-level security.
• Price: This can be compared based on a given number of users, on a specific hardware configuration.
To enable enterprises to better track and manage cross-departmental performance, BI vendors are rolling out frameworks designed to help enterprises integrate and leverage multiple existing BI systems and analytic capabilities. Theoretically, this enables a powerful enterprise-wide management tool for optimizing performance and profits. However, in practice, there are challenges associated with these frameworks that range from data acquisition, cleansing, and metadata management to aligning models and delivering performance management.
POTENTIAL DRAWBACKS OF EMPLOYING A BUSINESS MANAGEMENT PROCESS INTELLIGENT AGENT TOPOLOGY
The pressures on the management team are intense. Executives often struggle to integrate multiple data sources that can then be analyzed and synthesized for financial reporting. For example, inventory data may be stored in the data warehouse while sales information is housed elsewhere. In addition, executives may want to close their topics as often as daily, without having to wait until the end of the month to address a problem that could affect revenue or earnings forecasts.
Business intelligence tools are best suited for allowing executives to drill down into the source of financial reports to review their accuracy. For example, if an employee forgets to topic an order in an order management system, but makes the adjustment in the general ledger, a BI tool could start investigating further; if anomalies are found, it is critical to drill into the support data to determine the reason. However, a majority of BI systems would be hard-pressed to catch someone intent on committing fraud if the data was entered correctly in the data capture system.
Enterprise application vendors also are adding analytics to back-end financial systems to help companies better leverage raw transactional data. For example, PeopleSoft has embedded analytics in its financial module designed to report to executives on a daily basis the status of KPIs. Executives have little time to analyze data before senior management inquiries begin, and they need visibility across the internal and extended supply chain to have the control in their organization. With a broader audience, the need for more detailed intelligence will drive analytics that are targeted not only at specific vertical industries, but at specific departments and roles. However the dangers of creating more analytic silos cannot be overemphasized.
It is prudent to remember that pressure to act may result in erroneous actions. Business survival depends on an infrastructure that can adapt to changing market conditions. With companies focused on leveraging existing resources and increasing efficiency, infrastructure is no longer just an operational cost of doing business. Solution providers are paying attention to this growing need for intelligent infrastructure, with companies such as Network Appliance Inc. and Web Methods Inc. partnering with BI vendors to bring new abilities—such as storage analytics and business activity monitoring— to their solutions which can help companies realize the full potential of their resources.
CONCLUSION
This article conceptualizes the transition of intelligent agents utilized in network performance management into the field of business and management. A tiered IA system could be implemented at many levels of management and could be the key for successful, timely knowledge management strategies and successes. Such a system would be timely, unbiased, objective, and should provide significant competitive advantages. Moreover, it could leverage existing assets and provide a single objective measure of employee performance at appraisal time.
Although there has been some progress in increasing the number of users in the area of query and reporting, a lot more progress needs to be made between closing the loop between decision making and operations. IA benefits realized in telecommunications networks, grid computing, and data visualization for exploratory analysis connected to simulations should likewise be achievable in business management processes.
KEY TERMS
Data Mart: Database containing data extracted and often summarized from one or more operational systems or from a data warehouse and optimized to support the business analysis needs of a particular unit.
Data Mining: A component of business intelligence decision-support process in which patterns of information in data are discovered through the use of a smart program that automatically searches the database, finds significant patterns and correlations through the use of statistical algorithms, and infers rules from them.
Data Warehouse: A form of data storage geared towards business intelligence. It integrates data from various parts of the company. The data in a data warehouse is read-only and tends to include historical as well as current data so that users can perform trend analysis.
Enterprise Deployment: Term used in the computer industry to describe hardware and software configurations. They are aimed to address the corporation as a whole as opposed to a single department.
ERP: Enterprise resource planning system. Enables the company to integrate data used throughout the organizations in functions such as finance, operations, human resources, and sales. This system extends the pool of information for business intelligence.
Knowledge Management: Application for any business to create, maintain, and share company knowledge. The challenge of capturing collective experience, core values, and expertise of an organization.
Metadata: The information a database or application stores to describe the university’s business data and the applications that support it. Refers to the information used to describe a set of data.
Portals: Means entry point. Describes the entry point for users to information available throughout the World
Wide Web.
Set-Based Analysis: A recent method that uses groups of sets. It facilitates the examination and comparison of data.
Slice and Dice: Another term for multidimensional analysis. When data has three (or more) dimensions, it can be thought of as being arranged in a cube (or hypercube), with each side representing a dimension. When the data is analyzed, part of the cube can be “sliced” off or “diced” to get to an individual cell.
