I have been writing about the ROI of BI before, and it seems to be an ongoing question. Not surprising, and it should be. We should never stop asking about the return on investment, and start taking things for granted. Recently I had a discussion about the return on investment on data warehousing, and I wanted to share three approaches with you.
Approach #1: So what’s the ROI on the local area network?
I have been using this question for years, and no one has ever claimed to know. Wait, once a person raised his hand, but he misunderstood the question. Basically, there is no ROI on a LAN. Yet, I don’t know of a single company who got rid of their LAN. It’s needed to support all the other applications, that each may have their own ROI. The LAN is a cost of doing business. So what is the difference between an integrated communications infrastructure (LAN) and an integration data infrastructure (DW). I don’t think there is a difference. From this point of view, it is useless to try and determine the ROI of the DW. Just focus on the ROI (cost savings, new opportunities).
There may not be many larger enterprises who do not have a data warehouse. Most have gone through multiple generations. Right now, architects are wondering what the impact of SOA on data warehouses could be, and if that should lead to a next generation of the data warehouse. In this discussion, I think the focus should be on what new possibilities this generates for sharing the information with more people, particularly external stakeholders, or feeding back the information to transactional systems to create smarter business processes. There are plenty of solid business cases to be made for both approaches. And the DW is simply the way of architecting it, not a goal with an ROI by itself.
Approach #2: Information Capital
Perhaps a bit more conceptual way of determining the return on investment, is making use of a strategy map, a technique developed as part of the balanced scorecard. The balanced scorecard has four perspectives on performance: financial, customer, process and growth/learning. The growth/learning perspective consists of three topics: human capital, organization capital and information capital. A strategy map describes how the information capital supports critical business processes and customer interaction, and how that leads to financial results. The ROI of the DW simply would be offsetting the cost of building it against the benefits identified in the strategy map. The difference between the strategy map approach and approach #1 is that the strategy map is more generic of nature, it doesn’t focus on just a few business initiatives.
Approach #3: What’s the cost of not doing it?
This may actually be the most straightforward way of doing it. Each data mart or each application may have its own neat ROI. But lots of different applications and data marts managing their own data, meta data and master data will lead to very high integration costs, usually destroying the locally optimized returns of each separate initiative. The research I did at Gartner in this area suggested a break-even point of somewhere between 4 and 8. This means that if you have more than 4-8 different data marts or business initiatives requiring integrated data, it is simply cheaper to set up a data warehouse infrastructure to support all of them.
What approaches have you come across? Let me know!
