BI Platform Defined

CFOs should be happy with BI convergence

A BI platform, as Forrester Research defines it, is a rare thing. However, executive and senior management of any company would probably approve of the definition because it appeals to the bottom line.

Forrester says that to keep costs low and value high, a BI platform must come from a single vendor, give you a unified, enterprisewide data layer, provide centralized, thin-client tools, and let you share a universal set of metrics. The vendors that come closest to the ideal, according to a recent report by Forrester analyst Nate L. Root, are Cognos and Oracle. Root rates Cognos stronger on current offerings, but reports that Oracle has the advantage when it comes to strategy and market share.

BI software has traditionally been purchased at the department level, which has led to multiple problems over time. One of those problems is "shelfware." Root states, "Purchasers [at the department level] overestimate which employees will actually use the complicated tools."

Another problem is the cost of IT support and training for the resultant hodgepodge of BI software the company accumulates. Forrester surveyed 25 firms, each with $1 billion or more in annual revenues; 80 percent had multiple BI tools in use.

When BI is deployed departmentally, each department arranges independently for its data sources and defines the metrics it monitors. Of the 40 senior financial and operational managers Forrester interviewed at those 25 firms, 53 percent cited fragmented or inconsistent data as a problem; it was the problem most reported. Root adds, "Even when departments happened to use the same source data and tools [as other departments], differences in their core assumptions about business metrics led them to produce conflicting results." BI platforms provide a metadata layer that includes universally defined, reusable metrics.

Root recommends, uncontroversially, that BI vendors improve their tools and focus on functional depth. But his final recommendation is more provocative: The BI "veterans" (Brio Software Inc., Business Objects, Hyperion, and MicroStrategy) need to "embrace data mining." Users typically argue that, because data mining is done by a small cadre, their best-of-breed preferences can be accommodated. Root rebuts that because CIOs want to "minimize ISV relationships, cut licensing costs, and trim training expenses," they will prefer data mining tools that come from their BI platform vendor.

In this Issue: