The State of BI

by Vince Kellen, Senior Consultant, Cutter Consortium

For me, the most telling and important statistic in Cutter's recent survey on the state of BI is the percentage of employees that use BI tools: 57% of respondents indicate 0%-9.99% and 70% indicate 0%-14.99% (see Cutter Benchmark Report, "Successful Business Intelligence: Moving Beyond the Obvious," Vol. 7, No. 9). This confirms what I have long since observed -- most employees do not use business intelligence tools. Experience has also shown me that of those that do use BI tools, most use the more mundane features of the tools.

Not to panic, as I don't think this is a bad thing. Many BI projects I have been associated with, especially those since Web-based tools became available, have begun with project leaders hopeful that more data in people's hands would be a good thing, only to come to the realization that the hard effort in deploying BI solutions was appreciated by few. Along the way I have come to learn that many people inside business are often quite disinterested in more information. Or more analysis. Or more statistics. Or anything that takes more than a modicum of mental effort to comprehend.

A few years back, I had the pleasure of totally rewriting a data warehouse for a large movie distributor, where we radically reduced the complexity of the data warehouse. We did something the first implementers did not do: we carefully examined users' real analytical needs. And based on this review, we discovered that a vastly simplified and more automated BI solution was in order. Why were these users not interested in deeper analysis? They needed highly repeatable data models so they would not have to relearn anything and would not have to think deeply about the data they were looking at. They depended on us (and a scant few others) to do their deeper thinking for them. They just didn't have the time. Other recent BI implementations I have been a part of have had a similar user profile. The number of people who can spend the time to deeply analyze data are shockingly few, even in large organizations.

Most businesspeople live in interrupt-driven, instant-gratification, intuitive, "blinklike" decision-making modes. In this world, repetitive routines of human decision making are the norm, with a repeatable process (IT) for delivering data inputs (often in relatively simple form) to decision makers. This form of decision making lies below the attention threshold for the enterprise, especially its executive management. It is business intelligence on autopilot. Most BI solutions are aimed at these kinds of more mundane BI problems.

The Pareto Law of Business Intelligence

In this mode of BI, architects design and/or buy a BI solution, with most of the solution providing line managers with basic data summaries and data analysis. These comprise the employees who actively use BI tools. A much smaller subset of these employees -- in some cases, an astonishingly small number of employees -- actually digs deeper into more complex or abstract models for analyzing data. In this regard, BI is a highly leveraged activity. Very few employees truly understand the complexities of the analytical models, and those few serve as the designers of artifacts that the larger group of BI users will consume. BI solutions should (and do) cater to this group.

Involvement with a BI solution has two Pareto distributions of use, in which a minority of users receive the majority of the value. The first leveraged group is the 20%-30% of employees who use a BI solution. The second group comprises perhaps as few as 1% of the employee base who will use the more advanced features of a BI solution. Those employees who have the intellectual knack and time for abstract analysis are the ones who create objects that the 20%-30% use. BI use within a firm tends to involve three groups: (1) a priestly few who possess the analytical and abstract thinking skills and experience required to build analytical models; (2) a slightly larger group of analysts and decision makers who interact with the tool and extend or modify the models created; and (3) everyone else who either do not interact with a BI tool at all or do so very superficially.

From this, it follows that BI adoption is still occurring along traditional lines. Metaphorically speaking, the market segment is a rather expensive recreation vehicle (RV) parked in a cul-de-sac. Not a lot of movement.

Context Is Worth 80 IQ Points

In general, respondents expressed a neutral or slightly positive response (on a scale of 1 to 5) regarding the context of BI in their organizations. What can we infer from this? On average, respondents feel rather neutral about the following:

• Level of alignment between IT and business

• A continuous improvement culture in their organization

• Consensus as to what is "master data"

• How much the organization values information sharing and transparency

• The importance the organization places on measurement

It looks like the context still is not right for better adoption of BI tools. Despite the past couple of decades of vendors continuing to develop and sell BI tools, consultants trying to help out, and the use of data in virtually every aspect of a business, the best we can report are middling scores in self-assessment of how firms are doing with BI. This is downright depressing, but it is consistent with my own experiences as a practitioner over the past 20 years. A lot still stands between data and a good decision. Most of those obstacles have little to do with technology.

Respondents who answered positively to the statement "My organization has considerable prior experience with decision-processing engineering" were more likely to successfully use and implement BI tools despite experiencing similar levels of aggravation with some of the BI tool problems.

I welcome your comments on this issue of the Cutter Edge and encourage you to send your insights on the market in general to me at vkellen@cutter.com.

-- Vince Kellen, Senior Consultant, Cutter Consortium