In-Depth

In ROI we trust

• By Stephen Swoyer
• September 1, 2005

Can one realistically expect to profit from a technology, such as BI, that some folks say isn’t even a competitive differentiator anymore?

The answer to that question is yes—although as with any IT investment, the ROI of BI projects isn’t always immediately intelligible, and often can’t be calculated by means of any standard, tangible metrics. Frequently, BI practitioners say, you just have to trust that it’s there. Call it blind faith, of a sort, in a technology that is, for all intents and purposes, a competitive necessity.

“Usually BI comes down to religion,” says Joe Flynn, assistant VP of application development with regional insurer PMA Insurance Group. “You really have to believe that that’s the right thing to do, and that you will see improvement over time. Any good company is always measuring how they’re doing. It’s just something that you have to do. But it’s very difficult to say if you’ve improved your performance— was it because of BI or 10 or 12 other things?”

That’s a perspective endorsed by Matthew Datillo, CIO of PerkinElmer, who believes it’s difficult to quantify the impact of his company’s BI investments. “We use [our BI technology] to make decisions around product line profitability, and decisions around whether to continue to invest or not invest in specific areas. It’s always difficult to quantify the value of those decisions, but clearly without these tools we probably would not have made as insightful decisions,” he concludes. “The point is that [quantifying the ROI] isn’t something we worry about. I don’t know what we would do if we were trying to run this information out of our ERP systems without [our BI technology]. If you’re a large, complex organization and you want to have a consistent view of your organization, you need to have both [ERP and BI].”

In the end, says PMA’s Flynn, it often comes down to what use you make of your BI investments. Almost all companies use their BI tools to report against ERP and other systems, to be sure, but many—including Graniterock and RuffaloCODY—have also developed customerfacing BI-based offerings to differentiate them from their competitors.

PMA is another example. Notwithstanding Flynn’s claim that organizations must simply believe their BI investments will generate ROI, PMA had decidedly concrete ROI expectations when it invested more than $1 million on its next-generation BI initiative, an enterprise data warehouse with a front-end reporting tool from Hyperion Solutions. “[This project] required a significant study of the benefits before it was approved,” Flynn says. “The project was paid for, not by improving internal intelligence, but by being able to provide services to our customers, basically new services that the customer could subscribe to,” he confirms.

Far from being a slam dunk, Flynn argues, his company took some risks. For starters, PMA, a mainframe shop, had little or no in-house development expertise in the Microsoft Visual Basic.NET and ASP technologies required by the Hyperion Intelligence software. The cultivation of that talent, and of other BI development skills, came at no small expense, he says.

“We had very little experience with that, and we had to develop that experience, so we trained a number of people on Visual Basic along the way. We also needed data warehouse design help, because we wanted to go with star schema [a data warehouse model], and we didn’t have any experience with that,” he explains. “We also had to develop Ascential ETL skills because we needed to have the ability to take data from our legacy systems.”

Nevertheless, Flynn says, PMA’s investment could soon pay off. The company recently unveiled a new customerfacing application, PMA Cinch, that’s based on Hyperion’s Intelligence reporting tool. PMA Cinch exposes new revenue-generating services to customers, says Flynn, but it does so in the context of a low-maintenance, user self-service application. As a result, PMA hasn’t had to augment its in-house administrative and software development teams. “It allows [customers] to analyze data, so there’s preset reports that they can bring up on monthly data, but there’s also the capability to design their own graphs and pivot charts and that sort of thing,” he concludes. “If you’re a relatively unsophisticated user, you can just click on the icons and look at the reports and print them off. But if you’re a more sophisticated user, you can actually build your own reports.”