We had an assignment in INFS 762, Data Warehousing + Data Mining, to write three quick briefs on industry data warehousing projects. Here’s the second from my paper:

Home Depot launched a data warehousing project in 2002. The company had no data warehouse prior to this project. Before this project, analysts seeking big-picture information about Home Depot’s 1500 stores had to access information from as many as 16 separate mainframe systems. One could get the information if one really wanted to, but it took too long to provide effective support for decision-making (Whiting, 2002).

Home Depot was actually behind competitors like Lowe’s in adopting data warehouse technology, but, to make lemons out of lemonade, there may have been a cost advantage to being last mover instead of first mover. Instead of having to go through conversion of even embryonic prior data warehousing efforts, Home Depot could charge straight into building a new system with the latest technology (Schwartz, 2002). Bob DeRodes, the chief information officer who led Home Depot’s IT overhaul at the time, also wisely followed the phased implementation strategy that our textbook and others (such as Schwartz, 2002) say is essential to successful data warehouse development and deployment. Home Depot launched its data warehouse with applications dedicated to “analyzing human-resource expenses” (Whiting, 2002), a key business goal for the new CEO Bob Nardelli. The intent was to add incrementally to the system, with point-of-sale data to be added the following year.

Home Depot’s data warehousing project was part of a company-wide IT overhaul that cost $2 billion and required one year and one million person-hours to complete (Webster, 2006). Measuring the success of the project requires more than a look at the bottom line—after all, even the best data warehouse could not insulate Home Depot a 66% drop in 2008 Q1 profit due to the recession-launching collapse of the housing market (Clifford, 2008). The data warehouse implementation and the entire IT overhaul were part of a larger plan to remake the culture of the organization. When Bob Nardelli took over as CEO at the turn of the millennium, he found a corporation that was surprisingly decentralized. Store managers had vast autonomy and regularly rejected directives from corporate (Charan, 2006). Imagine the challenge of telling such staff to standardize their data for a new data warehouse. But that was part of the challenge Nardelli tackled, as he recognized that the lack of standardization, centralization, and discipline throughout the company would prevent the company from sustaining the growth it had enjoyed in the 1990s, especially now that competitors like Lowe’s were crowding their market.

One cultural change directly supported by the data warehouse was to shift the basis for decision-making from intuition and anecdote to data. Anecdotes and gut feelings, coupled with isolated, nonstandard bits of data, don’t travel well across over a thousand separate retail operations. The data warehouse provided the transparent data necessary for shared decision-making. It also provided the quantitative reports necessary to produce data- based templates for business review meetings. Previously managers would defend their positions with sketchy yet hard-to-challenge anecdotes from their own stores; the rich data from the data warehouse made such anecdotal escapes much more difficult and forced managers to acknowledge and act to correct failings in their operations (Charan, 2006). This change to data-based decision-making did not please everyone, and some employees did leave the company, but this cultural change, supported by the data warehouse, did take hold, as evidenced in part by improved employee satisfaction scores on in-house surveys (Charan, 2006).