Portals eNewsletters Web Seminars dataWarehouse.com DM Review Magazine
DM Review | Covering Business Intelligence, Integration & Analytics
   Covering Business Intelligence, Integration & Analytics Advanced Search
advertisement

Resource Portals
Business Intelligence
Business Performance Management
Compliance
CRM
Data Integration
Data Quality
Data Warehousing Basics
EAI
EDM
EII
ETL
More Portals...

Advertisement

Information Center
DM Review Home
Web Seminars & Archives
Newsletters
Current Magazine Issue
Magazine Archives
Online Columnists
Ask the Experts
Industry News
Search DM Review

General Resources
Bookstore
Industry Events Calendar
Vendor Listings
White Paper Library
Glossary
Software Demo Lab
Monthly Product Guides
Buyer's Guide

General Resources
About Us
Press Releases
Awards
Media Kit
Reprints
Magazine Subscriptions
Editorial Calendar
Contact Us
Customer Service

Decision Management Applications

  Article published in DM Direct Special Report
May 24, 2005 Issue
 
  By James Taylor

A decision management application (DMA) isolates the logic behind business decisions from the mechanical operations of application procedural code. Treating decision logic as a manageable enterprise resource in this way makes it reusable by multiple applications in many different operational environments. It makes advanced decision making available as a service to legacy applications, eliminating the time, cost and technical risk of trying to simultaneously reprogram the individual systems to keep up with changing business requirements. It enables developers to swiftly and inexpensively deliver new decisioning applications by flexibly combining business rules, data-driven strategies, thresholds, calculations and other elements. A DMA also makes it possible, for the first time, to use predictive analytics as an integral part of a real-time decision process. As a result, businesses can reap the benefits of analytics - clear forecasts of customer behavior, deep insights into customer needs, the ability to compare many related factors and criteria in a single measurement - to apply operational strategies that give them the greatest return.

The most sophisticated DMAs allow the application of analytics to the development of decision strategies, considering the mathematical relationships between varying business objectives, actions, probable customer reactions, constraints and outcomes. This enables companies to take market and economic uncertainties into account and arrive at optimal action strategies immediately, instead of after months of iteration - thereby sustaining the highest levels of performance for the longest possible duration.

A decision management application is designed to ensure that these decisions are:

  • Precise in that the risk and reward of the decision is known and managed.
  • Consistent so that the decision does not vary by time, channel, customer unless the organization means it to.
  • Agile so that it can be evolved rapidly in the face of new regulations, new policies, new risk or reward calculations.
  • Cost-effective so that no more money is spent on it than absolutely necessary.
  • Fast enough so that it is taken in "right time" or as near real time as adds value for both the organization and its customers.

Indeed these five aspects of a decision can be managed and improved to measure the "decision yield" of the decision.

The Benefits of Real-Time Decision Management

The returns of developing a decision management application come from faster, less costly application development and maintenance as well as from operational cost savings, more accurate risk management, reduced losses due to error and fraud, and increased revenue generation and profitability.

Clearly a decision about whether or not to approve credit involves risk, but so does a decision about whether or not to make a promotional offer to a customer. It's a decision that involves committing limited resources, making them unavailable for other opportunities. Every interaction with a customer also involves the risk that the attempt to target their requirements will miss the mark, not stand up against competitive offers, and so increase the risk of attrition.

The ROI for a DMA generally comes from increased productivity - of both the people who create the decisioning applications and the people who use them. The use of specialized decisioning infrastructure to develop these often complex applications can reduce development time greatly with resulting cost savings. In addition, developers spend less time building and particularly maintaining these applications, so they deliver more projects. DMAs have been estimated to automate 85 percent or more of many business decisions, which can reduce labor costs or free up staff to focus on higher value work such as portfolio analysis.1 Once a DMA has been developed, evolving it over time becomes easier allowing additional improvement when new data can also be leveraged, new analysis techniques applied, new rules of thumb integrated.

Ultimately, when improved decision making in more than one functional area is linked, the benefits expand exponentially. Ultimately, the aim of an organization should complete enterprise decision management - the linking of as many decision areas as possible, so that all decisions yield more value to the enterprise.

Increased Efficiencies and Productivity Benefits

Using traditional techniques, decision logic is hidden deep inside software and is time-consuming and costly to develop. Developers have had to translate business requirements ("If this condition is encountered, then respond in this manner") into abstract representations in programming languages. The process is laborious and full of opportunities for error through misinterpretation.

By separating decision logic from application code, and making it visible and accessible, companies have estimated a cost reduction of developing new decisioning applications ranging from 25 to 80 percent.2 If organizations go further, and give business users the power to make their own rule changes, IT will spend less time supporting deployed applications, and reduce maintenance expense s by as much as 75 percent.2 In addition, IT resources are no longer consumed by trying to support multiple decisioning applications for different channels and operating environments, because the same application can be deployed across them all.

Improvements also come from reducing labor and cycle times through automation, increasing decision quality and consistency, better and earlier detection of various kinds of risk (credit, attrition, fraud), more sophisticated balancing of risk/reward, more granular segmentation enabling more precise targeting of offers and treatments, and deeper insights into customer behavior and preferences.

Reducing the Latencies between the Time Information is Received and Action

Closed-loop decisioning enables organizations to capture results from production systems and immediately put them into useful form for development and refinement of rules, models and decision strategies. These results include decision outcomes (e.g., Did the customer accept the offer?) and other data from the point of decision as well as subsequent performance data (e.g., Has the account proved profitable?). This acceleration of the cycle from design to execution, and back again, increases ROI from DMAs. Companies gain marketplace agility and compress decision learning time down to a fraction of the norm. In fact companies can keep moving "the bar" upward, executing a new or enhanced strategy before competitors have even had the chance to react to their last move - i.e., they can begin to process multiple decision cycles within one decision cycle of the competitor.

Better Business Decisions with Improved Visibility - Business Process Improvements

As companies move to improve, innovate, automate and even offshore their processes they must face the fact that their differentiation increasingly boils down to how they process decisions. Sustainable competitive differentiation can be injected into processes effectively by identifying key decisions and using modern software technologies such as business rules and predictive analytics to take control of these decisions.

"Smart" process techniques implemented this way are easier to change or more agile. Many processes remain fairly static while the risk assessment, value calculations, policies and regulations that determine the outcomes of decisions change all the time. The use of decisioning infrastructure to develop decision management applications for critical decisions - what product to offer, how to price and underwrite a policy, etc. - makes decisions easy to manage and change. The use of a technology designed for these kinds of problems also makes the decisioning logic more visible and accessible to the business experts who set the policy, not just to the programmers who can read the code.

Focusing on these decisions as specific components or services, not just as steps in a process, allows them to be reused across business processes. Thus a DMA that automates a cross-sell decision can be used in the customer account originations process, the client retention process, as part of the standard billing cycle and as part of the normal call center operations.

Automation and the use of analytics can also dramatically improve the ability of an organization to anticipate the impact of changes. Not only can they be sure the change will be made quickly, avoiding unforeseen consequences from delay, but a formal assessment can be done of how previous decisions would have been impacted by the change and this allows true impact analysis.

Steps in Creating a Real-Time Decision Management Application

Real-time decision management applications can be quite complex but they need not be to be useful and can, indeed should, be created incrementally for an ever-increasing level of automation and a better and better decision yield. Three key steps can be taken to develop these applications cost-effectively and with low risk.

Stage One: Automate Procedural Decisions

The first step is to create decision management applications that are focused on compliance, policies and regulations that expose a company to unwanted risk and by being automated can lead to immediate and large impact. A robust business rules management system provides an integrated development and deployment environment for creating and maintaining decision logic, such as policies, thresholds and segmentation trees. Because decision logic is managed independently of operational applications (i.e., not programmed into their procedural code), it can be accessed by multiple applications and executed in just about any operational environment. Such a decision management application may automate a fairly high percentage of all transactions, although it is typically going to automate the most straightforward. For instance, in insurance underwriting, the rules allow for the enforcement of all company policies and legal regulations. This will mean that those policies without a complex risk/reward calculation can be automatically underwritten while the remainder must be manually reviewed. While this can result in automation of up to 80 percent, those decisions that are automated in this way may tend to be more straightforward meaning that there is not an 80 percent reduction in effort.

Such rules-based decisioning not only ensures consistency and compliance, it also increases business control and understanding by enabling managers to see in one place all the rules and policies contributing to a decision. If the decisioning infrastructure allows for deployment of these decisions in multiple formats such as COBOL, .NET and Java then these decision management applications can be deployed to improve the decisioning throughout a typical company's heterogeneous portfolio of information management systems.

Over time, as more decisions are automated, organizations will need to manage their enterprise repository of rules for reuse and will find that each successive decision is easier to automate thanks to this reuse.

Stage Two: Modeling Decisions

When predictive analytics are incorporated into rules-driven applications, decision precision greatly improves. Analytics improve precision by expanding the range of data and depth of analysis that can be brought to bear on complex decisions in very short timeframes. For examples, the output of analytics, such as clear forecasts of customer behavior and deep insights into customer requirements, can then be used by and combined with rules and other elements into complete decisioning processes. When a businessperson takes a decision, they will typically refer to multiple reports and sources of information to do so. Increasingly, Business Intelligence tools give them sophisticated insight into the operation of the business to assist in this decision-making process. Embedding predictive analytic models into a decision management application is the equivalent step - taking insight derived from data and using it to improve a decision. Except where BI improves manual decisioning, predictive analytics improve automated decisioning.

This kind of analysis can improve performance by micro-segmenting markets and portfolios that will provide a more accurate assessment of risk/revenue and better targeting of marketing offers and customer treatments. These techniques allow users to bring massive amounts of data and analytic rigor to bear on difficult decisions that must be made very quickly. By and large this is something organizations can only do once they understand their data and have it organized.  In other words, this might only be possible after a successful data warehouse/BI implementation for the relevant area of the business. This data can then be mined for business rules - techniques like CART (Classification And Regression Trees) allow data to be used to produce decision trees, for example, a standard business rule representation. More sophisticated users of the data will be deriving predictions from this data - how likely is a customer to jump to a competitor, what predicts default on a loan etc. Embedding these kinds of analytics into the decisioning application allows for new rules - rules that could not have been written before - to be added. At this point a critical success factor becomes the lag time between understanding what the data tells you and implementing this understanding in a production application. Rapid development, deployment and self-learning for redeployment of these predictive analytic models, therefore, becomes a must.

Figure 1

Stage Three: Optimizing Decisions

While predictive analytics can be used in decisioning processes, it is possible to model the mathematical relationships between all the elements that go into making a decision - multiple predictive models and data elements that describe customers or prospects, multiple (sometimes opposing) objectives, possible actions, possible customer reactions (forecasted by any number of predictive models), constraints (risk, resources, schedules, etc.) and outcomes. This analysis allows for the identification of the single best, or "optimal," strategy for achieving the performance outcome, given constraints and objectives.

When properly applied to a decision, this approach does not increase the automation percentage but quantifiably improves the value of each decision. This allows an organization to reach the highest levels of performance immediately instead of after time-consuming strategy iterations. It allows them to balance conflicting objectives and constraints and take market and economic uncertainties into account. It also gives them the ability to accurately simulate the outcome of decision strategies before the strategies are implemented. Once again, time to implement is critical. This means both that a tight learning loop must be implemented - feeding back results quickly - and that the results of the optimization must be deployable back into the same decision management application so as to improve the quality of decisioning.

Organizations can use this approach to deliver the decision management applications that are crucial to becoming a real-time enterprise. Considering business decisioning as a separate issue and developing a decisioning infrastructure of linked decision management applications that can take full advantage of integrated customer data and enterprise data warehouses will deliver true enterprise decision management and boost operational results while creating new value from current investments. By increasing the precision, consistency, agility, and speed of the decisions that drive business processes while reducing costs, an increased ROI can be delivered from existing applications, data and process. These benefits can be achieved in many areas of operations - wherever complex decisions must be made rapidly, in high volumes, and under changing conditions. Taking a comprehensive approach that leverages business rules management, predictive analytics and strategy optimization across the enterprise will multiply these returns.

References:

  1. Transforming Underwriting: From Risk Selection to Portfolio Management, Celent Communications, March 2004.
  2. Companies Achieve Significant ROI Using Fair Isaac Blaze Advisor, IDC, June 2003.
...............................................................................

For more information on related topics visit the following related portals...
Business Intelligence, Business Process Management, Business Rules and Strategic Intelligence.

James Taylor is vice president of marketing for Enterprise Decision Management at Fair Isaac where he is responsible for working with clients to identify and bring to market advanced decision management solutions that better solve the demands of business users and IT. You can contact him at jamestaylor@fairisaac.com. For more information, please visit www.fairisaac.com/edm.

Solutions Marketplace
Provided by IndustryBrains

Design Databases with ER/Studio: Free Trial
ER/Studio delivers next-generation data modeling. Multiple, distinct physical models based on a single logical model give you the tools you need to manage complex database environments and critical metadata in an intuitive user interface.

Fair Isaac Advanced Analytic Solutions
Fair Isaac makes decisions smarter. Six of the top 10 worldwide telecom service providers use our solutions to increase bottom lines. Bring together data, analytics and business intelligence software. Complex technology, simple decision: Learn more.

File Replication and Web Publishing - RepliWeb
Cross-platform peer-to-peer file replication, content synchronization and one-to-many file distribution solutions enabling content delivery. Replace site server publishing.

Help Desk Software Co-Winners HelpSTAR and Remedy
Help Desk Technology's HelpSTAR and BMC Remedy have been declared co-winners in Windows IT Pro Readers' Choice Awards for 2004. Discover proven help desk best practices right out of the box.

GDS Solutions for Consumer Products
Get global data synchronization solutions for consumer products to help you meet customer demands quickly & accurately. Improve speed and flexibility, & better get collaboration with SAP.

Click here to advertise in this space


E-mail This Article E-Mail This Article
Printer Friendly Version Printer-Friendly Version
Related Content Related Content
Request Reprints Request Reprints
Advertisement
advertisement
Site Map Terms of Use Privacy Policy
SourceMedia (c) 2005 DM Review and SourceMedia, Inc. All rights reserved.
Use, duplication, or sale of this service, or data contained herein, is strictly prohibited.