Taking BI to the next level

CXO talks to Karen Parrish, IBM’s Vice President for Worldwide BI Solutions, about the evolution of business intelligence and why companies need to start thinking in terms of enterprise-wide approaches to data analysis.

CXO. Gartner rates business intelligence as number two in the list of priorities for CIOs. How do you rate BI in terms of its importance for CIOs?
KP. I actually agree with Garner; it could even be the first most important. Six years ago when I took over this position in IBM, the areas of business intelligence and data warehousing were used predominately to cut cost out of the organization. That’s what the C-level executives were focused on. And if you didn’t find a way to take people or technology or whatever were the fixed costs out of the business, they didn’t want to hear anything about it. However, the world has changed considerably since then. C-level executives are now not only focused on cost containment, but also on top-line revenue group and profitability. And they’re finding that information, which is a huge asset in the organization, is grossly under-utilized. They’re starting to realize that information becomes valuable if it becomes insightful – if you can act on it, if you can trust it, if you can drive decisions against it and those decisions give you a competitive advantage.

As a result of that, business intelligence has become incredibly important to C-level executives. In fact, most of the decisions we see today are C-level, if not boardroom decisions, around business intelligence capabilities. And what’s interesting is the dollars have shifted. No longer are all the dollars sitting in IT. No longer is it just considered a technology decision. It is really driven by the business. Most of the dollars associated with business intelligence and data warehouse implementation are driven by the various business lines who see a problem and recognize that having better information, a greater grasp of that information, and the ability to act on it insightfully is a better way to grow the business.

CXO. So it’s not just high on the list of priorities for CIOs, but it’s also high on the list of CEOs and other members of the C-suite?
KP. Absolutely. With the growing interest in compliance and fraud, chief financial officers are particularly interested in new capabilities around business intelligence; similarly, you’re seeing the chief marketing officer step up in really wanting to get a single view of the customer and wanting to understand how they can better maintain customer loyalty across brands. So it’s not just the CIO at all. It really crosses all of the C-level executives in the business.

CXO. There has been much talk about business intelligence 2.0 recently. How does IBM define the next generation of BI products and solutions and how do these differ from previous iterations?
KP. This is a great question and one that fits nicely into IBM’s sweet spot. I would say there are three generations of business intelligence. The first is the traditional query and reporting. All organizations should be doing this in some way, shape or form. You ask a question and you get an answer. You want to know about this year’s sales versus last year’s sales; you deliver a report against it. That is what we call first generation and is the base for any business intelligence system.

The second generation involves more correlation-type information. It’s about technologies that allow you to do analysis of demographics, customer behaviour patterns, product SKUs and identifiers of buying patterns, and it really allows you to take various data and create informational views about a customer, a product, a service, etc. I believe that this is a critical step too, but both of these generations are what I call rear view mirror business intelligence, which means that you’re always looking at what happened in the past and, based on what happened in the past, determine what might occur in the present – or, even better, what could occur in the future. How might one behave if behaviour was this way last year and this way the year prior? I think that’s valuable, but it’s not enough; it doesn’t tell the whole story.

And so our belief is that there is a third generation of business intelligence (and by the way, I think it’s here today). It’s what we call information on-demand, and this is very much focused on optimizing the transaction. So this is not just looking at the past, but acting on what’s occurring in the present. Take insurance fraud, for example. An insurance claim traditionally goes through the claim process, gets paid, and then goes to the fraud department to determine whether or not it is fraudulent. If it’s a fraudulent claim, it then goes to collections and they attempt to recoup the dollars associated with the fraudulent claim. The chances of them getting that money back are next to nothing. Information on-demand, on the other hand, says that during the course of the business process, there should be alerts that identify potential fraudulent activity earlier. If you can identify that during the course of the claims process, the chances of you identifying the fraud activity before you’ve actually paid out the claim are significantly greater. This is incredibly valuable and it’s our belief that optimizing that transaction is critically important in this new generation of business intelligence.

CXO. So how do you see things evolving to meet changing needs?
KP. It’s a good question because I think that IBM has a very different view from what’s been talked about in the industry today. Our view is that business intelligence has to take into consideration all forms of data, not just relational data. Take that claims process, for example. What about all the information that’s been collected by the call center, all the information that’s handwritten in documents, where does that information get analyzed? It gets captured, but in many places it doesn’t get stored in a single repository. How then do we begin to mine that data to get a much broader view of the information that’s in the enterprise?

IBM has some really unique technologies here. We recently acquired a company called Omni Find, which offers search and text analytics capability, and we’re actually incorporating that capability into our broader product suite so that we can allow business intelligence or true analysis on all forms of data, not just on relational data. Our efforts will be focused around text analytics and our ability to search by key search words (defined by the business) so we know exactly what the business is looking for, whatever it happens to be thinking about. And we can define the business process against the data, to capture information and analyze all forms.

CXO. So do you see business intelligence, enterprise search and content management all coming together to offer a more holistic approach to data management?
KP. Yes, absolutely. It’s about search capabilities, it’s about content or unstructured capabilities and it’s also about business process capabilities. Content will be key in providing what we refer to as contextual information. Information needs to be represented in the way the business user wants to see it, wants to manipulate it, and needs to understand it. So I think that’s another phase of the third generation of business intelligence that can’t be overlooked: putting information in an industry format or a contextual format that’s really recognizable to the business user. Unstructured information helps you do that because it says a lot about how the business user communicates with the various constituents, whether they’re a supplier or a vendor or a customer. So I think we’re going to learn a lot about how to deliver information in a contextually valuable format through both the structured and unstructured data that’s available.

CXO. Should firms be overly concerned with the ROI of implementation or are there other less tangible benefits that maybe don’t show up in an ROI analysis? And maybe, how can IT departments or senior management establish whether a BI solution is providing value?
KP. I think ROI will continue to be important, but we actually do something called business value assessment, where we sit with the business users and try to understand what they’re attempting to accomplish and what information and/or data capabilities can help them resolve that. We find that getting a return on an investment is going to be critically important, but we often find that business users see other intangible benefits from these new business intelligence capabilities that they hadn’t seen before, for instance, improvements on business process.

Let’s go back to the claims analysis. If you can identify fraudulent activities before they occur, then your need to have a fraud department at the end of your business process may change. You may want to have those fraud-skilled people become part of the business process, not the tail end of it. Your department that actually does the accounts receivable might actually shrink, or you might want to redeploy those people to other types of roles because you may not need as many people focused on accounts receivable. So you find that business processes change.

You also find that the culture changes, and the roles that people play might change – for instance, changing the role of call centre employees from helpdesk into sales just through being able to offer better solutions to problems by having better information available. This provides enhancements to revenue, new career paths and new cultural environments. So there are all sorts of by-products of business intelligence that go beyond ROI that could be very healthy for an organization and make them far more competitive.

I think many companies underestimate the cultural impact. I had a conversation about enterprise-wide information with a customer just the other day; they looked at me and said, “But our lines of business don’t want to give up their data. They believe that it’s competitive to them, and they don’t believe there is value in an enterprise-wide view.”

My answer is that the company hasn’t allowed them to realize the benefits they can get from learning about what’s happening in other brands – cross-brand selling, up-selling, the ability to be able to identify that three of the 10 brands are dealing with the same customer – and how this can really add value to the brand. It’s all in how you represent it to your lines of business. So there are huge cultural aspects that are really underestimated. Everybody wants to focus on the technology, but I have to tell you that the technology is the easiest part of this. We have to get the organizations ready to accept change in the business, change in roles and accept the value that cross enterprise-wide data can deliver to a business.

CXO. Let’s talk a bit more about enterprise-wide deployment. What potential is there for these larger-scale BI deployments and what benefits could these bring?
KP. I think there is huge potential. Whether you’re talking about a Fortune 1000 company or a company in the SMB category, they’re all struggling with the same thing: differentiation in the marketplace and the ability to be competitive. And what they’re finding is that unless they get an enterprise-wide view of their data, they really have no way to communicate effectively with their customers and their suppliers – and at the end of the day, those are their most critical constituencies.

As we work with our customers that are looking at enterprise-wide data, there are a couple of things that they struggle with. One is this lack of desire amongst the various lines or brands within the business to give up data because they believe that it’s theirs to own. The other is that when you start building enterprise-wide data views, you have issues with governance. IBM’s view around information on-demand is that you provide information as a service. You provide it to all applications, to all business processes, independent of where those business processes or those applications lie in the enterprise. That means that you’re eliminating the silos or the one-to-one correlation between the data and the application. But then you have the issue of how do you govern the data? Who owns the data? Who’s responsible for data stewardship – for keeping it clean, for keeping it updated, and for ensuring that updated information can then be made available to larger communities? So security and access become major issues, but those issues are nothing in comparison to the value that you get of the enterprise-wide view.

I think there are two major drivers towards enterprise-wide views of the data, which oddly enough have very little to do with the customers and the suppliers. Compliance was a big driver a couple of years ago and remains a big driver. How can you be compliant on all facets of your data? You have to see it all, including the unstructured data, and you have to feel confident that the data is something that you’re willing to risk your career on. So that’s a big driver of an enterprise-wide approach.

The second one is fraud. When you start looking at fraudulent activity, most fraudsters rely on the fact that there isn’t an enterprise-wide view and that one hand doesn’t know what the other hand is doing. So those companies that are beginning to look at enterprise-wide views of all their data have a real leg up on fraudulent activity. I would say those are the two major drivers.