The article “Make BI Pervasive by Making BI Easy” began by restating the well-known struggle to make business intelligence (BI) pervasive throughout the organization. The comment was made that “the information they need isn’t just an Internet search or mouse click away.”

The value of pervasive BI is well understood. In the paper “The Payoff of Pervasive Performance Management” sponsored by BusinessWeek, it was stated that “organizations exercising world-class enterprise performance management, including widespread dispersal of the tools, enjoy 2.4 times the three-year equity market returns of typical companies in their industry.” This is hugely significant, and it points the way. With potential benefits of that magnitude, it is clear that no organization can walk away from the challenges.

The issue then, is how best to implement pervasive BI. In particular, how best to make BI an intrinsic part of every person’s work day.

The article first focused on a number of important points, such as a familiar user interface, a familiar language, reduced training and user assistance. All of these are valid points. It went on to use the Google search box as a metaphor, and here I felt that the parallel broke down. The article said “staring at a blank search input box can give a novice user stage fright.”

I would have to disagree. This hypothetical novice user is not a novice at her day job. She knows what she has to achieve; she know the mechanisms by which she does that. What she’s looking for from BI is the enhanced information it can provide her. Ease of use is important, yes, but density of relevant, actionable data is paramount.

Here’s an example: she works in Sales and Marketing. With multiple, prebuilt views such as By Sales Rep, By Product, By Geography, or perhaps By Sales Channel, By Product, By Promotion, she can select the most appropriate prebuilt information dense view of the underlying data. It is the job of the BI engineer to understand her job, and build those views for our user to select from.

Put another way, the BI engineer’s job is engineering the business first, and engineering the intelligence second.

We can build our portals in a personalized way. We can ensure that the most relevant views are precatalogued in such a way that they are one-click away from the main portal page, and we can built sophisticated preindexed enterprise search engines into that same portal to direct our users to any available data source.

The article suggests that the way forward is through natural language, backed up with significant semantic understanding of the vocabulary employed. It’s perhaps worthwhile to note that Information Builder’s packaging of English Query Language (EQL) with their PC focus products in the 1980s went nowhere. Theoretical work published by the IEEE from 2001 has not resulted in noteworthy commercial impact. The problem is to imbue the computer system with understanding of what constitutes valuable knowledge, and the effort to do that is not insignificant, nor cheap.

And therein lies the rub. For BI to be pervasive, it must be relatively cheap. At this time, implementation of true natural language processing is an expensive proposition. That’s not because of the cost of software, but because the effort involved in creating a usable knowledgebase.

Finally, for most end users, this is perhaps not the way into the information that is wanted or needed. Going back to the Google metaphor, entry of a search term is easy. Picking the right results, rejecting the dross and analyzing the remainder is very time consuming indeed. There are better ways of fast access to actionable BI.

Donna Kelly

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access