Information technology innovations in commodity 64-bit servers, open source platforms and million transaction per minute databases are redefining the limits of what is possible with data warehousing applications. Collectively, these drivers are catalyzing ongoing and emerging trends such data mart consolidation, active data warehousing and integrated business intelligence.
The Forrester Data Warehousing Web Survey of April 15, 2004 shows that active data warehousing is getting traction (see Figure 1). An active data warehouse drives operational systems, closing the automated loop in both directions between the data warehouse and the transactional system. Approximately 17 percent of respondents claim to have deployed an active data warehouse and another 13 percent have one in design for a total of 30 percent pursuing active data warehousing in some form. Still, the vast majority (70 percent) of respondents do not have an active data warehouse; however, 30 percent are considering it. Therefore, there is a potential pipeline.
Figure 1:What is the status of your "active data warehousing" solution?
Source: Forrester Data Warehousing Web Survey February - April 15, 2004. Respondents: 43.
There is evidence of market traction; but in this case, the numbers include departmental, prototype and tactical deployments. If this sounds like debunking one's own research, it must be acknowledged that the level of understanding of active data warehousing in the market is still emerging. Survey respondents sincerely think they have an active warehouse when the loop is closed somewhere, sometime and not necessarily in an automated, optimized way. In the long run, managed, sensible expectations will most benefit end-user enterprises (as well as the vendors that serve them).
Applications of integrated business intelligence (BI) based on active data warehousing include:
- Sourcing the forecast. For those enterprises that have physical inventory, reducing inventory through a demand planning or forecasting data warehouse results in significant cost reductions. That is a powerful data warehouse application, but it is not yet active data warehousing. When the forecasting system generates a plan or schedule for producing the products that will be used to satisfy the demand, then the loop has been closed back to the operational system. Then the process is bidirectional and the data warehouse is actively optimizing the operational system - from operational system to data warehousing and back from the data warehouse to optimize the sourcing of the product. The business intelligence forecast is integrated back into the operational system, producing integrated BI.
- Automated capture of the marketing promotion result. In marketing automation, the goal is to have a recommendation to offer the customer across a variety of customer contact points such as phone, e-mail, Web, store and snail mail. Based on a customer hub or data warehouse that represents customer revenue, profitability and buying behavior, product recommendations are prepared in advance and offered at the real-time point of contact. The result - customer bought or didn't buy - is then captured and, as the data warehouse becomes active, fed back from the profitability warehouse to optimize and refine the recommendation process. The bidirectional tool also prevents such pathological processes as over-soliciting customers with too many fatiguing offers because the result is captured in an automated way. Similar logic applies to customer churn analysis. This approach has been catalyzed by vendors such as E.piphany, SAS, Data Distilleries (SPSS) and competing solutions from Siebel, PeopleSoft and SAP. One of the main challenges has been to build and maintain a single, unified view of the customer. This is an area where those firms that operate a data warehouse have succeeded in breaking out the perspective of single departments and what might charitably be described as "active data marts."
- Real-time fraud detection. One of the methods of detecting fraud, for example in payment card transactions, is to build a profile of the user of the card in terms of products, stores, services and related buying behavior based on the data forwarded to the data warehouse from the transactional system. As the stream of purchases continues to flow in, it is monitored for conformance to the profile stored and updated in the data warehouse. So far, the data warehouse is passive. I will buy a pair of running shoes, but not 12 pairs at 12 consecutive stores in two hours - someone else obviously has my card. Items that are outside the profile are tagged as suspect. Active data warehousing requires an automated alert or intervention to be sent back from the data warehouse to the operational system to take action to stop the loss. That might also involve initiating a phone call, e-mail or page to the customer to contact the applicable call center for further service. Because the phone number of the store where the purchase is occurring is available online, the call can be directed to the store itself to check if the customer is still physically present. The loop is closed in both directions.
Information technology innovations are enabling new possibilities in data warehousing design. Active data warehousing executes mixed workloads of tactical and strategic inquiries against concurrent updates on multiterabyte volume points with thousands of users. Active data warehousing is an innovation in design, not a specific product. In turn, active data warehousing enables integrated business intelligence to overcome stovepipe data marts, bring transparency to the information supply chain and apply decision support data to optimize transactional processing.
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
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access