Related Links

Web Seminars

BAM (business activity monitoring) is an extension of traditional business intelligence, adding event monitoring to scheduled, batch-based reporting. Early accounts suggesting "zero latency" were exaggerated, since the discrete steps in the decision cycle do not happen all at once. These steps are:

• Track/monitor
• Analyze
• Hypothesize/model
• Decide
• Adjust/act

Businesses must accelerate the flow of information, analysis and decision making in order to be more responsive to fast-moving events. This business requirement will drive the augmentation of schedule-based technologies with event-based technologies - i.e., event-based business intelligence (BI).

Hype or Reality?

Given all the attention to "real-time" BI, one would think that this shift to event monitoring is pervasive. However, surveys of IT buyers tell a different story. Recent survey data suggests that BAM initiatives still have low priority within most organizations, with adoption currently at less than 10 percent (see Figure 1).

Figure 1: Adoption Stage for BAM Technology

It is evident that there is a healthy degree of skepticism in the marketplace on whether event-based technology is required. The reality is that most operational business processes don't require anything approaching real-time monitoring and analysis. In addition, most existing systems are not equipped to handle real-time data feeds.

The level of adoption demonstrated by the survey suggests that we are still in an early stage for event-based BI. The priority to monitor and respond to events will not affect all organizations or all business functions equally.

Benefits by Business Process

A fruitful way to look at the issue is to consider the differing requirements of applying BI (i.e., analytic applications) by business process or application. IDC segments analytic applications (that incorporate BI technologies) into three categories, by groups of business processes:

• Operations/production analytic applications measure and optimize the production and delivery of a business' products and/or services. Examples are demand planning, manufacturing quality analysis and fraud detection.
• Customer-related analytic applications measure and optimize customer relationships. Examples are marketing analysis, Web clickstream analysis and customer profitability analysis.
• Business performance management/financial analytic applications measure and optimize financial performance and/or establish and evaluate an enterprise business strategy. Examples are scorecarding, budgeting/planning and financial consolidation.

Comparing the three segments, operations/ production processes are areas where managers must be able to monitor and respond to events. This drives new investments in event-based BI in support of these processes. However, there are aspects of the other two areas that are also being impacted by this shift as well. Real-time call center interaction is one target. Another example is current Sarbanes-Oxley legislation requiring C-level executives to certify financial results and day-to-day controls, driving the demand for increased visibility of events and information.

Operations/Production Analytic Applications

The primary technical challenge in executing operational analytic applications that support event-based monitoring is the task of data capture and integration. Some of the examples of current products that perform this task include Informatica's PowerCenter RT, Information Builder's iWay and Ascential's DataStage RTI. These and other similar tools enable real-time access to transactions as they are committed within the core transactional systems. In turn, such tools are often used as the data integration components in operations analytic applications.

Other prepackaged operations/production analytic applications include those for quality control, fraud detection, trading monitoring, environmental surveillance and other processes that require true real-time activity monitoring. Application examples include Fair Isaac's Falcon credit card fraud detection, OSI Soft's power plant monitoring, Mantas' trading compliance, Celequest's Activity Suite for quality control, inventory monitoring and other similar applications.

Monitoring detects exceptions or deviations from norms, alerting responsible individuals on the need to take action. Operations/production is the largest analytic applications market segment (representing $1.4 billion in total software license and maintenance revenue in 2003) and the one where real-time, event-based requirements are most evident.

Customer-Related Analytic Applications

Event-based support for customer-related analytics enables companies to optimize live interactions with their customers and prospects. In many cases, the support consists of on-the-fly scoring and decision making that leverages the results of off-line analytics, although it is also possible to use analytics to constantly evolve the constraints that are used to trigger actions on an individual basis. As an example, a retail banking customer who suddenly makes an unusually large deposit may be a prospect for another financial instrument, but the definitions of "suddenly" (the timing) as well as "unusually large" (the amount) can change over time along with other variables to alter the definition of "prospect." If the customer is determined to be a hot prospect, action must be taken very swiftly before the deposit is designated for alternate uses.

The market is still in its early stages, but individual companies are using analytics to set up decisions that can be enacted in real time or near-real time. SPSS' PredictiveCallCenter is one example. SAS Interaction Management identifies when customer behavior becomes unusual. Hosted Web analytics from companies such as NetIQ (WebTrends) and WebSideStory can be configured to offer reports in near-real time.

The live interactions are usually sales or service oriented; the decision-making rules and analytics that identify the rules can emanate from sales, service or marketing. The sales interactions (including pre-sales informational contact) can occur between the customer/prospect and a Web site, live salesperson or automated phone menu. In all cases, the "next step" decisions for guiding the sales process are fed in real time to the site, sales-person or server. Which Web page to show next, what product to cross-sell or up-sell and even the "best" price and terms of the deal can be determined with analytics. Similarly, the offer that a service center agent makes to resolve a customer complaint can be optimized in terms of variables such as the potential value of the customer, likelihood of successful resolution and specific content of the offer.

BPM/Financial Analytic Applications