Continue in 2 seconds

Platforms for Next Generation Tactical Analysis Systems

Published
  • October 01 1998, 1:00am EDT

In the fast-paced world of computing paradigms in which technology life cycles often resemble the Doppler effect of a train rushing by, data marts and warehouses are no longer bleeding-edge novelties. Nearly every major organization has at least one data mart or warehouse in operation, with ever- increasing numbers of executives, business analysts, strategic planners, marketers and sales managers firing off increasing numbers of increasingly complex, ad hoc queries. True, the late adopters are still navigating the issues of how to build and maintain warehouses quickly and cost-effectively, but the major players have moved on to tackle larger issues. As they envision the potential of their warehousing systems, they focus almost exclusively on data explosion and user scalability issues related to the original decision support requirements that led to the development of their existing warehouses.

Their vision of the future of warehousing is one of more users, running more queries more quickly to spot trends and anomalies, leading to more targeted, one-to-one marketing programs.

These programs, targeting prospects for credit cards, telecommunication services and various other products, have undoubtedly enjoyed some success. In one particularly amusing, and often cited example, a supermarket chain discovered through its warehousing system an unexpected correlation between sales of disposable diapers and beer. Once the company spotted this trend, it instructed its store managers to put the beer next to the diapers and found that beer sales increased significantly.

But impressive as this is, is this sort of decision support all there is to data warehousing?

The Next Generation of DSS: Tactical Analysis Systems

The current breed of back-end, after-the-fact, decision support systems will not be enough to win the hearts, minds and loyalty of customers as we enter the next century. What's needed are systems that spot and seize opportunities in the moment. Systems that reward customers for their repeat patronage. Systems that spot trends and potentially dangerous situations as they're occurring, rather than weeks or months later.


Figure 1: Tactical Analysis System

Next generation systems will be tactical analysis systems. These systems-- marriages of current DSS technology with artificial intelligence (AI) and automated data mining tools--will make decisions and identify decisions that need to be made expeditiously, in real time to produce positive, "moment- of-truth" outcomes for a broad range of organizations including credit card companies, banks, brokerage firms, retail stores, law enforcement agencies and more.

Tailored Business Decisions

Imagine, for instance, the supermarket scenario mentioned earlier. A tactical analysis system would enable the supermarket to deliver customized pricing to patrons based on the frequency of their visits, the volume of business they do with the store, specific shopping patterns that coincide with manufacturers' cooperative programs, usage of electronic Web-based coupons that the system accesses at the point of purchase--the variables are limited only by the ingenuity of the marketers.

Or picture yourself at your local video store. When you go to check out your couple of tapes for the weekend, the system will recognize that you've rented a lot of action tapes in the past, prompt the cashier to tell you that the latest Arnold Schwarzenegger tape just came in and create an instant, customized discount or incentive to persuade you to rent that tape too.

An example of this type of information use can be experienced now when you're traveling on business. At some hotels as you check in, the desk clerk surprises you by welcoming you back to the hotel as though he or she actually knows you. "Welcome back, Mr. Jones. It's been a few months since we've seen you. We're glad you'll be staying with us again."

The next step for this hotel chain is to remember your habits and preferences in order to truly personalize your stay and ensure your ongoing patronage. Things like what food you ordered in the hotel restaurants (or affiliated restaurants--a whole new opportunity for partnering), that in past stays you've spent an average of $20 in the hotel gift shop on magazines, which magazines in particular (so that complimentary copies can be sent to your room on your first night in town), that you don't like your bed sheets turned down (but do want the complimentary mints), and that you usually request a 6:00 a.m. wake-up call. This type of personalized service is coming and will provide a significant competitive advantage to the first organizations to adopt it!

In an entirely different context, imagine you work for a government agency charged with the identification of suspicious cargo coming into the country. To do this, you have a very limited window of time. A traditional DSS system might be able to detect anomalous patterns indicating irregularities, but certainly not in real time.

What's needed in each of these circumstances is a tactical analysis system: a system trained to instantly compare current transactions against historical data and to produce immediate business decisions, be they tailored, point-of- purchase sales incentives, fraud detection, security alerts or hundreds of other types of opportunistic decisions.

Building on Existing Infrastructures

The good news is that the beginning of the infrastructure needed to build and deploy these systems is already in place at many organizations. Most organizations already provide front-line employees with access to corporate data either via direct links to local database servers or to centralized database servers via wide area networks. Data marts and centralized warehouses built on powerful SMP machines running high performing databases, can and will be adapted and optimized for tactical analysis. The business requirements of the organization, of course, will dictate the specific data and data structures and indicators of anomalies and/or customer opportunities to be noted and analyzed by the tactical analysis system.

The bad news is that the complexity of these types of systems is often not recognized until development is well underway. A tactical analysis system is to a decision support system what a motorcycle is to a bicycle. The core DSS still exists, but in a tactical analysis system it is carefully tuned to allow the AI- driven, automated analysis to occur for every allowable question in a sub- second time frame. No commercial off-the-shelf (COTS) tool can be purchased yet to address the average tactical analysis system requirements. So at the present time, each must be custom developed. Tool vendors who are currently rushing from building query tools to data mining tools, however, will soon be rushing to develop tactical analysis tools.

One government agency is currently operating such a system utilizing distributed SMP boxes running Oracle and Visual Basic and C++ interfaces on the front end. It also employs expert rules and a neural network which has been trained to automatically compare current transactions against historical activity and to recognize anomalies and unusual activity in real time. Over the past few years the system has been in operation, it has been extremely accurate and effective.

Breakthrough Performance and AI-Driven Data Mining are Key

A key to the effectiveness of this and any tactical analysis system is a tight integration with corporate information stores to enable real-time data updates to the system, reflecting each new transaction within seconds of its occurrence while simultaneously using the information store to evaluate the new transaction. This requires servers that can provide breakthrough performance, high availability and scalability such as servers based on the NUMA architecture (an expansion of the SMP model). It also requires the development of more industrial-strength data mining tools to search for patterns, trends and anomalies in an ocean of information that is anything but static. Currently, only a few tools are being developed to do this. Initially, they will require extensive modification for each deployment, but they will soon be on the market.

The Future is Now

As competition grows more fierce, organizations must think differently about how they run their businesses. The winners will be those who are able to act and react in the moment to spot opportunities, trends and threats, and to respond immediately and appropriately to these in accordance with the organization's objectives. The delays inherent in current DSS systems will prove to be a significant liability in many situations. By taking these DSS systems to the next level, organizations can increase their effectiveness and competitive edge by several orders of magnitude.

The reality and importance of these new tactical analysis systems gives new meaning to the old saying that the future is now.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access