Are you curious or even excited about active business intelligence (BI) and active data warehousing? Perhaps you are managing a traditional data warehouse that does not allow you to extend your responsibilities into operational applications. Therefore, active BI may seem like a distant dream.

Here is a suggestion on how to experiment with active BI with minimal effort and expense. The intent is to prove to yourself and document for management that active BI has specific tangible benefits for your business.

Step #1:
Identify a critical business process that is on the minds of executives and is covered by the subject areas in your warehouse. Cost: Coffee with a creative colleague.

Step #2:
Identify an abnormal condition that is messing up that business process. Think especially about conditions outside the scope of existing transaction systems (i.e., those having cross-function dependencies). Cost: Included in Step #1.

For example, Ford Motor Company needed to improve the flow of parts to assembly plants. They started with an alerting tool that notified them when a truck was running late per its schedule. This simple tool grew into the Inventory Management and Alerting System (IMAS) as requirements were extended from alerting to analytic metrics and then to a proactive operational application. Currently, IMAS measures the net change in inventory on a nightly basis and then calculates, at the level of individual parts, the days-on-hand inventory based on consumption and demand. Based on the key touchpoints in the supply chain, the flow of material is reprioritized for the next day.1

Step #3:
Identify the individual who cares about this abnormal condition and is able to rectify the situation. Cost: A few phone calls followed by another cup of coffee with a bagel.

This may be the most difficult step! If the person is too high, the alerts may be quickly delegated with little thought. If the person is too low, it is possible that there will not be any action.

Step #4:
Write a simple prototype application that detects the condition, gathers pertinent information and sends an e-mail to the caring soul. Cost: Limit the development effort to two or three person-days.

Most popular query tools have some alerting features. See Computer Associates' Forest & Trees for the needed feature set. The worst case is a few pages of Visual Basic code executing in the background on your desktop. If it is not possible to do so within the time constraint, examine weaknesses in staff skills and middleware infrastructure.

Step #5:
Initiate alerts and carefully observe what happens. Emphasize that this is a limited experiment with no guarantees of service levels. Cost: Limit the operations effort to three or four person-days spread over a specific trial period of two to four weeks.

Observe exactly what happens when the person receives the alert. If possible, ask the person to keep a log of the alerts and responses. After the trial period, schedule an assessment. If you have struck a raw nerve, the discussion should surface a clear idea of a new BI application that responds appropriately. Document the logic for the event response, stressing automation of routine matters and creative intervention of human judgment.

Step #6:
Propose to management a plan for doing it right, stressing the business benefits. Operational applications often have tangible benefits because these abnormal conditions do occur and often damage the business in some manner. Cost: Limit the planning effort to two or three days of your time.

Data latency issues may be critical. In other words, the key data is updated nightly but alerts would be more valuable if initiated from data that changes throughout the workday. Thus, some upgrades to the infrastructure may be required. However, the tangible benefits that you have documented would now justify these upgrades.

Total Cost: Two cups of coffee, one bagel, a few phone calls and 10 person-days.

Expect only a few of these alerting experiments to reveal new high-impact BI applications for your business. Try several experiments, and be patient.

Please share your experiences by e-mailing me. Your stories can be anonymous. I will do a future column on the "alerting experiments."

Reference:
1. Hackathorn, Richard. "Current Practices in Active Data Warehousing," November 2002. Case study whitepaper sponsored by Teradata, a division of NCR. http://www.teradata.com/library/display.asp? dest=4366&type=pd.

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