Ann All spoke with Lyndsay Wise, the principal of  WiseAnalytics, which conducts research and provides services to organizations within the areas of business intelligence, unstructured data and master data management. For more than seven years, Wise has assisted clients in business systems analysis, software selection and implementation of enterprise applications. She is a monthly columnist for DMReview and DashboardInsight.

All: We've been hearing for a long time about companies wanting to extend their use of BI to employees other than specialists. But surveys have shown that relatively few companies are doing so. Is this because this idea is more marketing hype than reality, or are other issues/challenges involved?
Wise: In general, I find that a lot of times with BI, there is a lag time in adoption with what we see coming out at the moment. You see the marketing hype and organizations will say, "Yes, this is great, but how are we going to get there?" BI is so data heavy that it's often very difficult in terms of being able to get the information that people want. I used to be a business systems analyst. When we were implementing BI, one of the issues was that each department was responsible for their data, and sharing that data wasn't necessarily something that they wanted to do — even if it wasn’t sensitive data. It wasn't something that they wanted to lose control over.

All: What about data quality? Is that a problem that is hampering broader adoption of BI?
Wise: With any BI deployment, data quality becomes an issue. A lot of times bad data or dirty data is going in, or data is getting keyed in wrong. If there isn't some sort of cleansing effort within the source systems or the operational data stores, then once you bring it into the data warehouse, it's kind of a glaring error.

But I think the bigger issue becomes how you use (the data). I know I have seen situations where data marts have been developed for each business unit, as opposed to having a centralized data store. If you are going for an organization-wide approach, then at least theoretically it makes sense to have a centralized data store where all of the data is the same. If each area updates its data at different times, or the areas might not even have the same data source, the issue of conflicting reports or information might occur. But with a centralized data store, once data quality issues are repaired, they should be repaired for everywhere.

Also, in terms of expanding usage, it's the same problem that customer data integration efforts are having, in terms of identifying things like who is the customer, how your product numbers are derived, etc. All of these definitions and the way business is done need to be identified centrally. For the IT department, for instance, their customers are other employees. When you say "vendor," do you mean a software vendor or someone you are paying for services? It's really determining how all of those entities are identified and maintained. Another issue is data governance. If there is a lot of data, there needs to be collaboration in identifying the owner of the data and who is going to maintain it.

All: You've already touched on the issue of data marts vs. a central data warehouse. Is one approach preferable over the other?
Wise: On one hand, you want to think about what a data warehouse will ultimately look like. But at the same time, if you start huge, you are pretty much bound to fail. If you start small, at least then you can identify the business problems you are trying to solve, and figure out what information you need to do the analysis in order to make better decisions. At the same time, you should try to maintain that view of what the overall picture will look like.

All: I know that with many BI implementations, cultural issues can present a bigger challenge than technical issues. Any advice on dealing with those?
Wise: When I talk to end users, I hear a lot about training. It's not at all a technical aspect, but it ends up becoming very big in terms of whether people actually adopt BI and how they end up using it. For super users and IT, I think they may think (BI is) more intuitive than it actually is. With BI search and different areas in which BI is starting to become more intuitive, by developing interfaces that look and feel the same way a person might surf on the Internet, hopefully in the future it won't be as much of an issue.

Another recommendation is to get the business unit more involved. More super users within the business so that not only are end users not always going back to IT, but also those issues of whether something will work or how well it will work are determined by the business. Once the business adopts BI and likes it, they'll see other uses for it and want to use it more.