For more information on Cutter Consortium's
Intelligence advisory service, please contact
Dennis Crowley at +1 781 641 5125 or e-mail
ENTERPRISE DECISION MANAGEMENT
by Curt Hall, Senior Consultant, Cutter Consortium
A new concept has entered the BI space: "enterprise decision management" (EDM). EDM refers to the application of rule-based systems -- in conjunction with analytic models -- to automate, improve, and distribute decision-making capabilities across an organization.
Currently, EDM is most commonly used to automate customer- interaction decisions associated with revenue-generating activities -- such as marketing, product recommendation (i.e., personalization), up-sell/cross-sell offers, and pricing. However, decision management is also seeing increasing use for automating other types of high-volume decisions, such as guidance and employee support, forms and data management, and work-flow and process control.
Overview of EDM and Rule-Based Systems
When I first began to examine EDM, I was immediately struck by how its underlying concepts resembled those championed by expert systems proponents in the 1980s and 1990s. This is because I began my computing career in the mid-1980s covering the commercial application of artificial intelligence (AI), which, at that time, primarily involved the application of rule-based expert systems technology. The similarities are evident: both concepts rely on rule-based systems to automate and improve an organization's decision-making capabilities. In addition, both use the term "rule-based" to describe systems that employ an engine that can process rules that are stated explicitly and stored separately from database and application logic in a "rule base." Rules are processed dynamically (typically by an inference engine) in response to specific events -- for example, a customer clicking on a Web page advertisement or inputting facts (e.g., automobile model and driving history) into an insurance-processing application.
In an effort to distance their technology from any negative connotations associated with AI and expert systems, EDM proponents and vendors typically refer to their rules engines as "business rules servers" or "business rules engines," as opposed to inference engines.
Where the two concepts mainly differ is in the degree to which they distribute their decision-making functionality. Expert systems are primarily used to apply complex logic to automate a process or activity for a single application or channel. One of the most popular uses of expert systems today is in the financial industry for loan approval. Another is in the airline industry for aircraft landing and gate assignment at airports, and determining the optimum time to schedule aircraft maintenance. In contrast, EDM applications tend to apply rule-based systems to automate an organization's decision-making across multiple customer channels or lines of business. EDM also appears to make greater use of BI and analytic applications.
Why Rule-Based Systems?
Rule-based systems have become the technology of choice for implementing EDM applications. This is because they offer a number of advantages when used for automated decision-making. First, because rules are stored separately from application and database logic, the system's decision-making logic can be changed fairly easily because there is no need to recompile the application every time new rules are added or existing rules are changed. Second, rules provide an excellent means for representing and simplifying complex business logic. Rules are expressed as IF-THEN statements using English-like syntax, for example:
IF CUSTOMER_INCOME = $75,000 - $100,000 AND
CUSTOMER_HOMEOWNER_CODE = 3 THEN
CUSTOMER_LIFETIMEVALUE = 9
Another major benefit of using rule-based systems is because the expressive nature of rules makes it easier for less-technical people (e.g., marketing managers and merchandisers) to update and maintain rulebases. In addition, rule-based systems can provide a straightforward explanation for their output, as opposed to neural networks and other "black box" technologies. This is especially important for financial and insurance applications in which regulatory requirements require a clear explanation for refusing to approve a transaction (e.g., loan approval).
The major drawback to using rule-based systems is that initial rulebase development requires someone to write the rules that will be applied to respond to specific events. And developing an initial rulebase is very difficult because it demands a developer who is experienced with implementing rule-based systems and who possesses an in-depth knowledge of the industry and application domain(s) in which the system will be used. Such developers are rare in most corporate IT environments.
Status of EDM
EDM is currently enjoying its greatest use for automating customer-interaction decisions, because it enables the deployment of applications that:
Offer near real-time response to events occurring over multiple channels by monitoring transaction-level data
Leverage the content of existing customer information repositories to provide personalization of Web content, call center, and customer service, etc.
Coordinate sales and service across multiple channels based on customer interactions across all these channels and comprehensive profiles generated from these interactions
In some sense, you might think of EDM used for automating customer interaction decision-making as personalization on steroids. This is because EDM provides the ability to analyze customer interactions and determine their significance for the whole customer relationship, and to allow the proper offer or customer service response to be sent out over the appropriate channels. By monitoring customer interactions as they take place, and by applying predefined customer relationship rules, it is possible to identify marketing and service opportunities and offer an automated, cohesive, and tailored customer response over the appropriate channel in near real time.
Sounds Great, But...
EDM is more talked about today than it is actually practiced. This is because implementing EDM requires more than just deploying a rule-based system. It requires an infrastructure that allows the technology to be applied across multiple channels. And this means integrating different business processes as well as integrating data from various operational systems and analytic applications. Consequently, EDM applications must combine rule-based systems with enterprise application integration (EAI), data warehousing, BI, CRM, and messaging and middleware.
-- Curt Hall, Senior Consultant, Cutter ConsortiumEnterprise Decision Management