A Realistic Business Perspective on the BI/DW Decision

by Robert J. Benson, Senior Consultant, Cutter Consortium

My objective in this article is to look at the discussion of BI with or without DW from a business rather than a technical or operational perspective. As we know, it is all too easy to let technical issues become the primary basis for IT decisions, and this isn''t necessarily the only important perspective -- not least because it can bias the decision process from the perspective of business management ("Oh boy, there go the technocrats again!"). Frankly speaking, issues of "data management" and "data asset management" are, in the minds of many business executives, fundamentally technical matters.

Let me be clear on one starting point: I am in the BI-with-DW camp. To my mind, the long-term implications for the business of data chaos and "spider web" operational implementations trump the (also) important short-term individual business unit (BU) issues. However -- and this is the point of this article -- this all may be irrelevant. That is, worrying about BI with DW or siloed BI ignores the fundamental and pragmatic point that it's the larger business considerations that determine the success or failure of BI and DW.

Thinking Effectively About BI

This discussion is reminiscent of enterprise architecture (EA) decision making. My colleagues and I recently worked with a large financial services enterprise in the BI and DW areas. In this IT-driven business, it proved very difficult to make a successful business case for significant investment in either area. This was simply because the benefits were obscure to the management team, and the IT folks couched their case in technical terms. Worse, both EA and DW issues needed to be taken into account in their business-strategy decision making, and this made things even more complicated. It seems the enterprise business management got the idea in its head that both EA and DW added only complexity and served merely to make the IT jobs more interesting and perhaps easier. Business management did not consider the potential benefit of significant new and powerful customer information in its business strategy. As a consequence, the team did not support the EA and DW investment and lost a competitive opportunity.

So how can we think effectively about the BI and EA questions such that business management understands the issues, supports the directions, and, most important, achieves success in the use of BI/DW? As the BI/DW debate has been framed, the key points on either side seem to be:

• BI without DW. This approach makes early success (i.e., useful applications in the business) possible without waiting for the DW or, for that matter, waiting for others in the enterprise. However, it leads to IT devolution, spider web implementations, and the continuation of data chaos throughout the enterprise.

• BI with DW. This approach makes data management, consistency, and a "single version of the truth" possible. On the other hand, it takes a long time and might not even be responsive to each BU's needs.

My point will be that the tradeoffs here are exactly what business-IT governance should determine and what business strategy will shape: deciding which factors are more important to the business -- and who exactly should be making this decision.

These issues are not the same for every business. I understand that some enterprises have the luxury of IT-sympathetic and IT-strategic business executives at the helm. For those of you in this situation, please make a sound technical decision; the business side will pretty much take care of itself. For everyone else, the business issues are much more critical. The BI/DW discussion is not a "one size fits all" matter -- although it is sure to turn into one if we treat it as a technical decision.

Note that these issues are particularly difficult and important for enterprises that have more than one line of business (LOB). Companies with a single LOB have only to deal with the functional areas of the business. This makes a "single view of the customer" and the related data mining much easier to address as desirable objectives, since the functional areas already have to, in effect, deal with the same customer base. In contrast, multiple-LOB companies have much more difficulty with BI/DW because the customer base is mostly different, or at least has different relationships with the enterprise. Even financial services companies, which one would think are simple, have enormous difficulties among their various wealth management, credit card, loan/deposit, commercial lending, and community-based businesses. (These are, in our experience, the ultimate silos.)

My fundamental point is this: ultimately, the limiting factor to any BI and/or DW success is the ability and willingness of the business to use the resulting capabilities effectively. As my colleagues and I have observed, this is not a "build it and they will come" situation. There's so much involved on the business side that the business-side considerations have to dominate -- and therefore take the lead in the BI-with-or-without-DW decisions.

Reality is hard. In the absence of concrete senior executive support for BI and DW, there's seldom real agreement on the benefits of an enterprise-wide approach. Various parts of the business jealously horde their customer relationships and, worse, believe it's vital to their success to have differences maintained. When making the BI/DW decision, some of the possible business factors to consider include:

• Maturity. Both business and technical maturity in the application of BI to the business are paramount. The problem, of course, is that the successful use of BI requires a learning curve. This reality tends to suggest that starting simple is important for making progress. A plan for a complete enterprise-wide implementation is so much harder if the maturity level is low.

• Holistic perspective. Much of the value of both DW and BI lies in the integration of data across otherwise siloed business units. Unfortunately, this need to work across isolated BUs almost assures problems of definition, data flow, ownership, and all the other modest issues of such integration. Again, the simple approach may be better for going up the learning curve.

• Data versus business process. This is a double-barreled issue. The business process up front deals with securing, reconciling, and controlling the data. The business process in back deals with how the data and BI will actually be used in discovery, in strategy development, and in integrated customer services. Does data or process dominate? Yes. But the degree to which data or process dominates depends on the specific business strategy and where the business stands in its maturity in using BI.

• Differing expectations. In the financial services organization example above, there are three separate views of the importance of the DW and/or BI activities. It's sometimes hard to distinguish between them, but in fact they are quite different:

  1. Simplify the spider web that currently ships data hither and yon within and between siloed application systems.

  2. Provide field managers with a 360-degree view of customers, in real time.

  3. Use BI to provide marketing analysis for product and customer planning.

With such varied objectives in the business, the DW and/or BI activity is likely to satisfy no one. In practical terms, without clear governance, the activity founders.

All these factors are interesting and important. Furthermore, we have observed that these are the things our clients and other companies are concerned about with regard to BI/DW. But in the end, they are merely symptoms of the larger issues, which all come down to strategy and governance.

Making the BI/DW Decision Based on Business Issues

This all just gets us to the really important question: how should a CIO go about making the BI/DW decision? In thinking about this article, I noted five key questions about the BI/DW decision for a given enterprise:

1. How and where do business and IT executives entertain the "why worry about it" discussions about BI/DW?

2. How and where do business executives express how they intend to use the BI results (understanding that this may be a journey sometimes), and how should this impact the BI/DW implementation questions?

3. How do IT executives frame their IT strategic imperatives in addressing these issues?

4. How do the above questions frame the governance issues?

5. How do change process matters get handled? That is, how does governance prevent the encroachment of additional complexity and spider web implementations?

Clearly, these are all basically governance and planning questions. This leads to two perspectives: one strategic and one governance-oriented. They are related, and ultimately they suggest the same conclusion about BI/DW.

I welcome your comments on this issue of the Cutter Edge and encourage you to send your insights on the market in general to me at bbenson@cutter.com.

-- Robert J. Benson, Senior Consultant, Cutter Consortium