Data warehousing has become something of a buzzword, and it's easy to understand why. The volume of data collected at a chain's point of sale, for example, holds invaluable information about products, promotions, customers and cashiers, to name a few.
ehouse system, or too much of its computing power is used chasing far-flung facts.
In addition, because such systems require state-of-the-art technology, initial costs of building a data warehouse remain high. However, information systems executives agree data warehouses are becoming a competitive necessity in today's environment.
SN: How would you characterize the current role of data warehouses in the supermarket industry? What are they collecting, and what are used for?
HOMA: I think many of us are building data warehouses to look at itemized sales data. Nearly everyone has built or is building a data warehouse today. Some of it is driven by the fear that if my competition is doing it, I'd better do it. In our case, I think it promises to fully impact how we manage inventory and items in our stores.
I suspect very few people have started to realize the benefits of the information. We are right at the tip of the iceberg.
SMITH: I'd say there is a high level of interest but a lower adoption rate. The interest is driven by category management and loyalty marketing. Distributors are making some use but are still trying to figure out how to do it.
NICHOLSON: Several new universes of data are coming into the industry, and data warehouses might have some real benefit in some of them. The biggest one is movement data from front-end systems. One of the new applications coming into the industry is for database marketing. All of a sudden we not only have POS data -- how many cans of green beans did we sell -- we have customer-specific data -- what did Mrs. Jones purchase yesterday when she was shopping?
HOMA: The retailer doesn't have to know the name of the customer for the information to be valuable. It could be used in aggregated form to understand what a typical customer would put in his market basket.
SN: What lessons have you learned so far about working with data warehouses?
SMITH: One is that the system performance can be a major factor in the success of this application. There is such a large degree of flexibility for people to come in and ask for information that they can degrade the performance of an entire computer.
In addition, I think distributors need to act as a key collection point for store-level and market information, to aggregate it and pass it on to manufacturers in forms the manufacturer needs.
SN: Do you think the roles of data warehouses will continue to expand?
DRURY: While I can't speak for everyone, I can tell you that the popularity of them speaks for itself. Companies don't invest this kind of money in this type of technology unless they see a payback or they know of a payback.
SN: What makes data warehouses so expensive?
DRURY: When talking about large amounts of data, extremely large engines are needed on the computer side along with very fast storage systems. Any time you push the envelope with the fastest or the biggest, you're always at the high end of the cost curve. Data warehouses are bumping up against the higher-cost hardware.
That being said, the next piece is the tools. These are expensive but not as expensive as the hardware. The hardware is getting less expensive, but the less expensive it gets the more you want to put on the database.
