You have a well-managed data warehouse, but imagine extending its impact, making information you've gathered for strategic analysis available to front-line decision makers in your enterprise. You can.

Extending real-time intelligence to call center agents, front-line managers and Web sites is the very purpose of active enterprise intelligence.

For instance, it can help your Web customers design custom products, suggesting features based on their profiles, their past buying habits or assessments of next-best offers. It can use real-time information to dynamically plan inventory. Additionally, it can help customers keep abreast of their orders in an automated support center.

How do you, as a project manager, roll out active enterprise intelligence? It takes a mix of new technology and new thinking, changing the way your employees work and how your culture views information. Above all, active enterprise intelligence is a natural, logical extension of your enterprise data warehouse—a way to maximize a current asset.

The brainstorming phase
Where do you start? First, determine where active enterprise intelligence can drive the most value to your front line and your bottom line. Make an inventory of current systems, indexed by IT owners and business owners, and a list of potential projects where back-office insights can add value to front-line users and systems.

Convey the concept. Remember that IT and businesspeople don't always speak the same language or even know each other, since they're often confined to different silos in the enterprise. The front-line people are mostly application-centric, and the back-end people are business intelligence (BI)-centric.

As the project manager, you'll be the de facto diplomat between different cultures, ensuring they share perspectives on pain points and opportunities. For instance, do the call center agents ask for information that was perhaps already provided on the company's Web site? Or when booking a customer's flight, could the call center agent have known what the customer was trying to book on the Web?

Remember that the technical and functional needs of customer-facing systems are different from back-end BI applications. The front-line team may worry about an application's ease of use; the back-office crew may think about bandwidth and transaction metrics.

Identify opportunities. Once you've got process owners on the same page, have them look for gaps in their systems—gaps where active enterprise intelligence can improve your bottom line. Do call center representatives need more information to decide on late-fee forgiveness? Do gate agents have enough data (historical and predictive, with a profitability rating) to decide on upgrading a flier's seat? Can a Web site offer a customer the right cooking utensils for the wok he just bought, based on brand, style and the customer's average purchase amount?

Remember two key points when choosing your pilot project. First, it should be small, from six to eight weeks. That way you can manage the workload and find a quick win, one that cements the value of active enterprise intelligence in your end users' eyes. One way to simplify is to reuse information that's already in the data warehouse, avoiding new data sourcing or quality issues. You can also use BI analysis you've already done—such as next-best offers for customers—that's ready to deploy into front-line systems.

The second point is simple. It should be high-impact and measurable. A little active enterprise intelligence goes a long way, and a small but well-chosen pilot deployment can turn the biggest doubting Thomas into a believer.

Planning for success
Once you've brought everyone together, done your brainstorming and chosen a pilot project, it's time to start the nuts-and-bolts planning.

This is where project managers live and breathe. You'll need a detailed project plan with the standard ingredients, including a mission statement, list of stakeholders by responsibility, timeline, budget, statement of risk and statement of scope.

Your statement of scope is key, and it has to be precise. It should include exact descriptions of the work to be done, down to the technical and data management issues of the system in question. Ask yourself:

	What's new?
	What reports do we need?
	Which are modified and how will that look to the front-line user?
	What queries need to be written?
	How do we alter the interface (or add new interfaces)?
	Does the work affect the logical data model?
	What new data sources do we need to support this project?

You'll also need a roadmap with milestones that offer proof of progress. The best way to do so—and to give your project a firm footing—is a kickoff meeting with your stakeholders. Be sure to set up a governance system with status reports and go/no-go gates. Flesh out a pragmatic communications plan: Will you use an intranet? E-mail? Regular meetings? Decide now, when everyone's in the room.

Last, don't forget about politics. You have to have an executive ally that sits above the front-line and back-end people. Because an active enterprise intelligence project entails more than technological change—it demands that your enterprise learn to share data in new ways—you may find that department heads won't yield control of their information, or that teams won't share data. It's human nature, of course, and it need not be feared—but it must be fixed. A strong sponsor can use the weight of his or her position to end fights among fiefdoms.

Move forward
Now it's time to act. Active enterprise intelligence must be tailored to the needs of each company, so no two active enterprise intelligence projects look alike. But, by and large, your project will go through the process of initial design, development, testing, training, feedback, refinement and sign-offs.

The training phase is vital, in part because the shift from purely strategic thinking to using strategic historical and predictive data for better front-line decisions is seismic—a new terrain for end users. Remember that most of your ticket agents and cashiers won't know about data warehousing, much less about the specifics and power of the technology. You'll have to understand what's on the mind of a front-line person who comes to work and runs the call center, because he or she has other metrics in mind. Front-line employees don't think "How can I exploit the enterprise data warehouse?"

Because active enterprise intelligence is only as good as the data you feed it, training your company's front-line employees to understand information gaps, up-selling and cross-selling, and new efficiencies should be an integral part of your training phase.

Measure success
Last, you'll measure the results of your active enterprise intelligence project and report its business impact.

Of course, you'll need metrics that could be different from those you're used to. Standard call center metrics, for instance, include average call duration and hold times or number of transfers. But active enterprise intelligence demands metrics that measure not only business impact but how decision making and customer experience improved. These can include metrics that determine whether a call center solved the customer's problem on the first call or whether the agent converted a complaint into a sale.

Active enterprise intelligence metrics should measure customer opinions and impressions. They should take a holistic view of what matters to customers—across all interaction channels. In the end, that's the real value of active enterprise intelligence: smarter, faster decisions that build the bottom line, with customers whose loyalty is not only improved, but assured. T

David Garrett is a former IT director who now writes about the nexus of business strategies and strategic technologies.

Teradata Magazine-June 2007