a centralized data strategy, and it isn't valuable if the data is dirty.

Data issues are a problem for a number of companies throughout manufacturing, hospitality, and the other service industries. It's not just retail companies that have this issue: Data at many companies is decentralized, it's fragmented in different locations.

CIO: How can organizations fix this problem?

Anand: First off, the mandate has to come from the top-level management, it must have top-level executive support. Top-level executives in companies that are doing business intelligence well—Walmart, Best Buy, Staples, Target and other top retailers—make clean data a priority.

Once a data centralization strategy is in place, organizations can start with data cleansing projects to improve the data quality. It's a slow process but it's a crucial one. So for example, you eradicate duplicate data for location, products and customer data such as having one person's name in the system in different ways, such as Robert C. Parker and Bob Parker.

Customer information duplication is a especially a problem since the consumer is interacting with different faces of the organization—Web, store and so on—so information in captured in different ways.

The other thing is to limit the fields that you typically look at when you're analyzing customer data and product location data. There are so many fields that organizations have. Some of them are totally redundant. So you should have some core fields where you have the basic demographics and then you have customer transaction size and market basket size. You can then apply algorithms for cross-relational data analysis and do constructive business intelligence.

Take Best Buy, which does business intelligence very well. They used to have immense disparity of sources and fragmented reports. There was no centralized data repository for them, no clear path for clean information to be correlated for analysis and ad hoc reporting of department or business process performance were the norm. But they cleaned their data, integrated those disparate data sources, and built an excellent national scorecard that measures enterprise performance across different retail processes. They've seen effective results come out of it.

And the scorecard is just one example of successful BI usage by Best Buy. Another area where they've succeeded is looking at the marketing and overall company intelligence to guide how and what to sell to the various segments of their customer base.

Making Predictions with Business Intelligence Tools

CIO: Much of business intelligence looks at historic data. How do you predict the new items customers will want?

Anand: Forecasting the new items your customer will want is crucial for most retailers, especially for the holiday season.

The key is looking at the research on key new product adoption trends, and using predictive analytics to determine just what your customers will want. I can't overstress how important this is. When the iPod wave started, the companies that embraced it immediately saw success and had continued success because the customer got accustomed to seeing that particular merchandise at that particular store.

Companies that did not embrace it were the followers and they lost out. If those followers had looked at the competitive landscape and identified that in advance, they would've been able to fulfill customer need. Now customers will simply go elsewhere.