Getting Past the Silos

The recent theme in books and articles on the subject of business intelligence is that in order to get value from these initiatives, you have to do something with the information: make better decisions, take action, measure your results...lather, rinse, repeat. I'm up on that crowded soapbox and I'll tell you why.

In order for business intelligence to be more than just a smarter way of doing what you're already doing, you have to be able to reach across departmental boundaries (which are imaginary, by the way, as in this humorous Coke commercial where one department tries to sue another) and put the data together in new ways that will give you access to the new insights that can make you more competitive. Everything you're learning about your own department? Your competitors already know all that stuff. You're not getting a leg up that way. But if you can ignore those artificial boundaries and think about things from a customer perspective--like an end-to-end process that cuts across multiple departments, for example--you can find ways to move your business to a new level of competition.

What's difficult about doing this is not necessarily a data issue. Typically each department/division has it's own goals that it needs to meet, it's own incentive programs and each department will focus on their goals which, by the very nature of organizations, will be different than those of another department. There is typically little or no incentive for the managers to allocate resources to activities that will not have a direct impact on their departmental goals.

In previous posts I have stated that data warehousing and business intelligence functions should be centralized with some resources assigned to various internal clients. There's an important reason why the organization needs to be set up this way. I can hear the protests right now: "But, but, but! Our needs are unique!" "Our IT doesn't get it!" "We need to move faster than that!" You know how I know you're saying this? Because that's exactly where I was a few years ago when I started working on our data warehousing project. But once we started working on projects that leveraged the data we had gathered, the light bulb came on as we began to realize the true value of what we had created. Creating a more efficient and elegant way to manage our division's processes with information was our primary aim in the beginning. As we began to accumulate data from various areas, our division executive started coming up with ways to combine data from different systems to get new information. For example, she requested analysis comparing customer satisfaction data with employee satisfaction data to see if there was any correlation. This may not sound groundbreaking, but historically these were two different processes with totally separate data stores and combining the data was something that had never been considered. It doesn't take a rocket scientist to figure out that the two were directly correlated and this information was used to make important management decisions.

Another great idea was connecting customer satisfaction data to health data. What we wanted to know was whether we could predict that someone was going to be unhappy simply based on the types of claims they were filing. The answer was amazing and not at all what we thought. Customers who were dissatisfied tended to fall under the category of either having filed very little or a whole, whole lot. Most of our satisfied customers were those who had filed enough claims to have some experience with our customer service department, but not enough to be buried in the avalanche of paperwork that often comes with a catastrophic illness or injury. What this told us was that we needed to approach two very different populations with different processes in order to bring about significant improvement in satisfaction.

The common thread in both of these scenarios was that these were data sets that had never been connected before and in both cases it took the support of an upper-level executive to make the analysis and subsequent actions take place. So if you're lucky enough to have someone in your upper ranks who "gets it" and asks provocative questions and then does something about the answers, you are on the road from simply faster and more accurate analysis to "competing" on analytics.