It's a problem for nearly every business—decision makers can't react quickly enough to channel the flood of data surging through their organizations from all sources. Examples of this can be seen across industries:

	Large health insurance companies process millions of claims each year using simple rules-based and largely manual methods. As a result, an alarming rate of fraud takes place within claims, but by the time any fraud is detected, checks have already been mailed. Recovering these funds requires considerable resources and litigation.
	Growing manufacturing operations need to apply constantly shifting quality assurance rules against product changes. If production volume increases, so does the required diagnostics. Employees are stretched thin trying to keep up with new products and rules.
	Call centers at financial institutions handle complaints of disputed charges from customers daily. Representatives often spend inordinate amounts of time to verify that these claims are legitimate and not simply record-keeping errors or other mistakes. As a result, lines are tied up and wait times increase.
	Retailers aggressively extend seasonal and targeted discounts to shoppers through online channels. However, they often have limited insight into how effectively they are reaching targeted consumer segments or whether the increased traffic generated by these discounts increased revenue.

While executive brainpower will always be required for large, sweeping decisions, such as moving into new markets or acquiring a competitor, an organization's success depends largely on the countless tactical or operational decisions that are made on the front lines. Such decisions may include whether to flag a suspicious claim, extend credit to a client, cross-sell a new product to a customer on the phone or approve a transaction. Software already on the market can play a key supporting role in this lower-level decision making by looking at data from past transactions and making recommendations for moving forward on pending transactions. Or, in some cases, the software itself can complete the transaction.

High-volume, operational decisions with repeatable outcomes are perfect candidates for automated decision making, in which business rules are applied to incoming data and responses—or recommended responses—are delivered via applications. Automated decisioning, a new breed of actionable intelligence systems, can be pre-programmed and regularly updated with expert judgment based on experiences from past successes and failures.

More than 95% of the decisions made across organizations can potentially be automated, saving countless employee hours and reducing time to market, according to James Taylor, co-founder of Smart (Enough) Systems LLC, a consulting and services firm that specializes in enterprise decision management (EDM). Taylor says that a strong case can be made for more automated decisioning and that it doesn't take an extensive technology investment to make this capability possible.

Enter EDM
Taylor refers to this next evolution as EDM, in which day-to-day operational decisions are made systemically rather than on a random or ad hoc basis. Taylor urges organizations to "turn operational decision making into a corporate asset you can measure, control and improve." In fact, organizations that have a robust enterprise data warehouse in place, along with business intelligence (BI) capabilities, are well on their way to automated decision making.

EDM takes BI—both real time and historical—and incorporates the disciplines of business rules and modeling to help apply automation to BI solutions, which often have been hampered by cost, complexity and under-use by most of the enterprise. In addition, BI typically has provided support only for hindsight analysis. Migrating from traditional BI to EDM calls into play "trying to influence decisions that are more operational, more fine-grained," Taylor says.

Such decision making leaves it up to rules-based systems to move processes to the next step. Linking or integrating systems may move transactions more quickly and seamlessly through the enterprise, but in most of those systems, "transactions must wait for people," Taylor says. The key is the ability to make intelligent decisions in the absence of human intervention. For that small percentage of decisions that have exceptions or special circumstances, a manager can intervene.

The impact on transaction speeds could be enormous, Taylor says, but there's more to EDM than simply speeding up the pace of decision making. In fact, speedier decisions have long been the pitch associated with each new generation of BI tools. "At some point you're going to hit the wall," he says, "because there is either no one there to look at something, or there is absolutely no time in which to do it."

Making sense of data
Many data reporting and analysis systems are overwhelmed, and decision makers don't have the time or resources to make sense of all of this data. EDM helps companies navigate through and make sense of the huge amount of data flooding organizations and systems. Digital information is coming in from every corner of the enterprise—from operations, customer interactions and transactions, and Web sites. New initiatives, such as supply chain tracking through RFID and GPS devices, add even more bulk to this continuous stream of data.

In addition, it's inherently difficult to build actionable intelligence into most existing applications. "Enterprise applications don't focus coherently on decisions and don't enable the decisions they do automate to be managed effectively," Taylor says.

EDM provides a mechanism for helping employees and other end users make informed decisions that are consistent across the business. Plus, automated decisions ensure that processes are carried out in a much more rapid and reliable fashion, thereby dramatically decreasing time to market. By handling many well-established, day-to-day decisions with software, employees are freed up to better focus on higher-level decisions in which human intervention is still needed.

Taylor says an EDM approach can show a positive return on investment (ROI) by reducing costs in acquiring, building, operating or maintaining information systems. Many of today's systems cannot address current demands for competitive analytics and cause organizations to squander an inordinate amount of money on unnecessary data, wasteful activities, lost opportunities, fraud and fines.

EDM also helps enhance revenue growth, Taylor says, because automated decisioning helps target customers better, respond more quickly to market demands and deliver better cross-channel customer experiences. EDM bolsters strategic control over information systems as well.

Building EDM
EDM can be built on top of the resources and methodologies associated with a service-oriented architecture (SOA). With SOA, organizations can create decision services as formal Web services that can be managed and stored in a single place within the enterprise.

Technology areas making this a reality also include business process management and business rules management. Because SOA involves the assembling of applications or interfaces from components or services with different properties, this paves the way for the assembly and invoking of decision services.

"Automated decision making is most likely to be seen within processes that are already fairly well-automated," Taylor says. "The business rules engines, data mining and predictive analytics that have worked well for many years are in a position to be piped into a lot of mainstream business processes."

The focus of EDM is to systemize lower-level operational decisions, typically by "people who don't have any sort of executive influence, such as call center reps or drivers," Taylor says. "Or, the decisions are programmed into a Web site. While the value of each of these decisions is small, organizations make millions of them. And little decisions add up. Risk-centric industries such as insurance companies or banks have thought this way for a while. They know that lots of individual decisions about credit or risk will add up to determine the overall profitability of their portfolio."

Half of the decisions
The role of EDM is often to provide guidance to operational decision makers—not issue a single hard-and-fast decision, Taylor adds. This is an advance from BI systems that, up until now, essentially dumped a lot of information in users' laps, saying, "Here's the information, go figure out what to do," according to Taylor. An effective EDM solution will offer viable alternatives based on available data and mapped against business rules. "It may say to a rep, 'Here are the three things that are OK for you to do. Do any of them—it doesn't matter which one you do, but do one of them.'�"

It's not necessary for every decision within a process to be automated. Typically, an organization deploying EDM may choose to automate about half of the decisions in a certain process. "You may just automate the really obvious decisions," Taylor says. "Then the manual decisions are reviewed, each day or each week, and gradually you start to understand: 'Whenever these things are true, we always do this. So let's add those rules.' Gradually, the percentage of automated decisions goes up."

An organization may choose to provide automated decisioning support to a select group of customer service representatives to measure results. "You're not committing to change everybody," Taylor explains. "Perhaps you deliver the automated cross-sell approach to one or several of your call center reps, or even put it up for everybody with the option to use it or not. You then track who used it and who didn't, and compare the results." Once you've shown that the reps with automated decisioning support do better than the reps using manual decision processes, he says, you can challenge them with more sophisticated automated approaches. T

Joe McKendrick is an author and independent analyst who tracks the impact of information technology on management and markets.

Teradata Magazine-December 2008