About Face — Returning BI to Its Roots

In an article entitled “The Changing Face of Business Intelligence,” last month Dave Wells eloquently described how the business intelligence industry has strayed from its original vision and how it is now changing to recover what’s been lost. A longtime veteran of data warehousing and business intelligence, Wells is one of the leaders in the industry who have shaken free of the technology-centric perspective that holds the industry back.

Wells begins by reminding us of Howard Dresner’s original vision when he coined the term “business intelligence” (BI) in the early 1990s. Dresner defined BI as “a set of concepts and methodologies to improve decision making in business through use of facts and fact-based systems.” Over the years, the industry that took hold of Dresner’s visionary term (mostly data warehousing vendors at the time) buried the goal of decision making in an emphasis on technology. As Wells says: “The troubling thing is that all of the definitions are IT-centric” and “too much of today’s business analytics has little connection with real business analysis.”

He goes on to offer a new definition of business intelligence, which recaptures the essence of the original and enhances it to further clarify the goals. I don’t want to give too much away by quoting his definition here; you should read Wells’ words directly. I do want to include one more quote, however, which is central to Well’s vision of BI’s transformation:

It is analysts – the people who perform analysis – who find meaning in the data. These are the people who explore cause-effect relationships and who guide decision-making processes. It is they who will lead the charge to reshape decision making in business.

To recover the original vision, the business intelligence industry must shift from an emphasis on technology to an emphasis on the people who use the technology. Only then will it begin to fulfill its original promise.

(Note: While I consider Wells’ argument brilliant, I believe that some of the software products that he lists as examples of “next generation of analytics” don’t belong there. In fact, I believe that some of the products on the list exemplify little understanding of and support for data analysis. This difference of opinion suggests that our common vision must become informed by clear definitions of data analysis and analytics and clear criteria for assessing products’ ability to deliver. All in good time.)

5 Comments on “About Face — Returning BI to Its Roots”


By Stephen Few. December 27th, 2008 at 12:36 pm

I received an email from Seth Grimes today, who pointed out that the term “business intelligence” actually predates Howard Dresner’s use of it in the 1990s. Back in the 1950s, Hans Peter Luhn wrote an article that was published in IBM Journal entitled “A Business Intelligence System”. In this article, Luhn envisioned many of the goals that the business intelligence industry is only now beginning to address. According to Seth, Luhn’s vision featured an “emphasis on discovering and communicating relationships (and not just data values).” Luhn’s understanding of “intelligence” as “the ability to apprehend the interrelationships of presented facts in such a way as to guide action towards a desired goal,” represents a view of business intelligence that still eludes many vendors today. For more on this topic, take a look at Seth’s article, entitled “BI at 50 Turns Back to the Future.”

By Neil Raden. January 4th, 2009 at 10:34 am

I’ve had this conversation with Seth, and I don’t agree with him. At the time Luhn used this term, there was no concept at all of people using computers as analytical tools. As for Luhn’s goals, I have a few too, but 50 years from now, when computers can actually perform them, like learn what I want to do instead of vice versa, I don’t think anyone will attribute this to me. When Howard first used the term, we were already able to do some BI, albeit far far slower and with far far less data. If I have occasion to address this issue again, I will continue to attribute the term to Howard, not Luhn.

-NR

By Byron Igoe. January 5th, 2009 at 9:10 am

I agree with Neil and Seth, because they’re not really disagreeing. In Seth’s article he says, “The reality is that BI, first described by Luhn before business operations were computerized, developed in directions he did not foresee.”

Despite the fact that Luhn defined Business Intelligence differently, his ideas are good, and companies (especially Google) have implemented many of them. See: http://bionbi.blogspot.com/2008/10/happy-birthday-business-intelligence.html

Neither Luhn nor Dresner envisioned all the things we currently include in Business Intelligence. Dresner’s definition is closer to what we use today (both temporally and in accuracy), but it is vague and abstract; Luhn described a practical system. Also, he published first.

By the way, right here and now I am the first to publish the term Hippopotamus Hegemony!

By Michael E. Driscoll. January 14th, 2009 at 6:46 am

The same kind of broadening of the field (and its attendant deficits) occurred in computational biology in the 1990s. The good news is: a decade later, the talent is still standing, but others have been washed out. I suspect the same may occur in the BI space.

Even better, a new generation of data-aware and sofware-savvy business analysts are arriving on the scene: this alone will have a great impact in driving BI towards to the vision that Dave Wells and Howard Dresner had decades ago.

By Bill Smith. February 5th, 2009 at 11:53 pm

I wrote a piece about Luhn’s 1958 article in the IBM Systems Journal last year. His premise was “Efficient communication is a key to progress in all fields of human endeavor and automation offers the only viable way to manage” what he described, even in 1958, as “increasing amounts of information being created at an ever-increasing rate”. Seems like he had premonition.

The key functions of Luhn’s BIS model and the ones that are key to enabling people to efficiently engage in the dynamics of modern decision-making are:
1. “Know” the information requirements of people
2. Focus technology on tasks that can be automated to free people to focus on tasks, like interacting discussing and developing consensus, that cannot
3. Source data and get it “ready for analysis”
4. Manufacture answers to people’s questions and “deliver” the right information to the right people at the right time
5. Continuously iterate new versions of information to reduce “decision latency”, enable consensus and action

Sure sounds like he had it figured-out.

Regards to all.

Bill Smith