Getting the Most from Your Data Warehouse
Editorial by Bill Bittner, President, BWH Consulting
Many retailers have spent considerable time and effort installing Data Warehouses. When they’re done and begin writing queries against it, it is hard to avoid “getting lost in the trees.” All of a sudden, they have so much detail about each item sold, at what register, at what time, and to whom, that the data loses all meaning. They need a way to extract the “information” from all the data.
This inspires two thoughts.
First, in addition to storing the raw data, the Data Warehouse needs to have benchmarks for comparison. These benchmarks would describe what is “normal” for a particular transaction and serve as the basis for filtering the raw data into exception reports that show where deviations are occurring. These benchmarks will let the user manage by exception. The key is to decide ahead of time what benchmarks are needed, and then have them available in the Data Warehouse during query operations.
The second thought is to approach using the data in the Data Warehouse from a completely different angle. Instead of inquiring against the raw data, first decide what information is needed to drive the various decision processes a retailer goes through. This approach would ask, “What facts are important to take into consideration when making assortment, pricing, promotion, or replenishment decisions?” Then the Data Warehouse would be used to determine these specific facts and they would be loaded into the decision-making applications. The design of the decision-making processes can plan to use these facts when determining what action to take.
Data Warehouse software is designed to answer a lot of “what if” questions typical of a research facility. But retailers don’t ask a thousand different questions. Retailers ask a relatively few questions thousands of times as they evaluate the performance of individual items in various market areas and locations.
Moderator’s Comment: Do you agree that a Data Warehouse can provide too much detail? Can the key “facts” for driving
various decision processes be defined? Are there other ways to get to the “information” that is hidden in the data out of a Data Warehouse?
I really believe the two simple steps I described would improve the return on Data Warehouse investments. Once it is accepted that processes should be driven
by facts extracted from the retailer’s Data Warehouse, it may also be desirable to use facts from other sources such as market research firms. This can provide additional facts
about items the retailer may not even carry. –
Bill Bittner – Moderator