Is business intelligence still possible?
Silly question. Of course, it is. Just keep the computer people away. My friend Hugh, who is a senior partner at an actuarial consulting firm, was making a blunt, but funny, joke. Or was he?
Trying to find a common definition for business intelligence is a challenging task. What I have seen is wide ranging, confusing, and inaccurate. While reviewing the trade journals, I found that some definitions of business intelligence begin with: Business intelligence (BI) is a software solution that
Lets stop. Right here. We have our first problem outlined.
Business intelligence certainly is not a software solution nor any other inanimate process or thing. Business intelligence first and foremost involves people. Moreover, it involves people making decisions. It is manifested best when people seeking solutions have their moments of aha. The role of business intelligence is to help them to get to those moments.
Simple? A lot of empirical evidence tells us it is not. Two contributing factors stand out above all others:
Overreliance on technology, and
The information glut.
Why We Abdicate
There are many good reasons people turn to computers to tell them what to do. Computers are faster, they are infinitely more accurate, and they are reliable. What we tend to forget is, a few exceptions aside, they cannot yet think at the level of human intelligence.
Humans absorb information, and ultimately become experts, through a time-consuming learning process. This process progresses through the following phases: First, we understand; then (we become able to) explain; and finally, we internalize. The true moments of aha do not happen until we reach the last stage of learning. Thats the stage in which we finally are capable of providing clarity and insight.
There are no known shortcuts in that process. Yet misled by the raw speed and power of computers, we often believe such machine-generated shortcuts are possible. That is when, perhaps charmed by a vendors magic or techno-promises of our info-gurus, we follow a mirage. This is when we abdicate our responsibilities to engage our brains.
It is hard to blame people. They already are overburdened by the complexities of contemporary life and systematically robbed of the already scarcest of the important and non-renewable commodities: time.
Enter the Info Glut
The connection between the scarcity of time and the problems with managing information is evident. It makes abdication of the time-consuming information management process so attractive. Nobel prize-winning economist Herbert Simon said it best: What information consumes is rather obviousit consumes the attention of its recipients. Hence, a wealth of information creates a poverty of attention.
Very noble of Simon to say wealth. Im less inclined to be so polite. The sheer volume and garbage-like quality of information we are exposed to every day already impairs our performance and adds to stress. We all know it, and lest we forget, we are reminded by new terms such as Data Smog (David Shenk) or Information Anxiety (Richard S. Wurman). And yet we continue to devise more ways to foil and obscure.
For example, we produce stuff that pretends to be information, yet not only does it not inform, it means nothing.
I once read a statistic that claimed 69 percent of IT execs felt they could respond to change faster than competitors. Im not sure what that means, and regardless of whether you decide to call it disinformation, misinformation, or noninformation, the point is the same. No one can and no one will do anything useful with it. Yet it was conceived, processed, and published. And guess whatit was read.
Data Warehouse to The Rescue
This is where the two factors we discussedtechnological overreliance and the info glutconverge so logically. The promise is irresistible. The recipe goes like this: (a) Take all corporate data; (b) organize it neatly into well-defined bins; (c) add powerful engines capable of extracting any and all combinations of the data; and (d) give the decision-makers easy-to-use tools that make their requests for intelligence a snap. A great recipe? Absolutely! Can you make a meal with it? Not in your kitchen!
OK, Im being melodramatic, but I want to make clear why so many data warehouse projects tank (by some accounts over 80 percent fail to meet expectations) and, more importantly, set the stage to suggest a better way.
Why Data Warehouse Projects Fail
To understand this, lets walk through our recipe:
For (a), you need clean data. How many of you have processes and systems through which you can capture clean, accurate data? Not too many hands raised? Too many legacy applications still in place?
For (b), you need top-notch talent and dedication. In my experience, only the best data analysts are up to this task; moreover, unless they are teamed with the top-quality business minds, the results will be compromised. Remember, this is an intense, time-consuming exercise. Unless you are prepared to dedicate your best multidisciplinary team for a minimum of three to four months, dont even start.
Item (c) is straightforward: Sign a big check. In this domain, the technology is really good and plentiful. And expensive. But, honestly, thats the least of your problems. Once you do make it work, it will pay back handsomely.
As for (d), theres a problem. Such tools dont exist. At least, not as we defined them. The advertised ease of use does not match the habits and capabilities of even an exceptionally computer-adept insurance business executive. And so you have two strategies to choose from: Either your decision-makers will need continuous help (extra cost) or, somehow, they will have to climb the technology mountain (doubtful).
To make matters worse, another set of hazards awaits: How to interpret and fit in the historical data, which almost certainly comes in a variety of formats and meanings? How to deal with data brought in through mergers and acquisitions? What about new divisions created through restructuring? Even your most talented design team cannot accommodate every possible twist nor anticipate every possible change.
A Better Solution
Lets make a couple of changes to such a high-risk, high-tech, all-or-nothing approach. Lets put human experts back in the center of it all and move incrementally.
Create a Business Intelligence Team. This is how it works.
In phase one, you start with a small group of experts, probably no more than three: a business analyst, a database analyst, and a business domain expert. The choice of domain will be driven by your priorities. It could be related to your customers, distribution channels, underwriting for personal property, or commercial lines product development.
Whatever domain you choose, make sure the business executive responsible for it co-sponsors the team, and the BI objectives for the team are well defined. Before you put this interdisciplinary group to work, equip it with the best set of tools you can afford for the task at hand.
Three to six months later, you should be able to demonstrate your first success. You will have created a BI kernel: two experts who rapidly (as long as you didnt skimp on the tools), intelligently, and interactively can respond to complex requests for information and analysis. As only humans are capable, they also will learn a lot about the data and the people. They will learn about the quirks, the habits, and the patterns that do and dont work. From such insights, in due time, the effective paths to further automation will emerge.
The BI kernel gives you a springboard to phase two. Add another domain expert or two, and repeat the process. As the number of domains grows, there will be a need for a well-designed, coordinated view of where the domain boundaries are and how to exchange data among them. That requires an information architect role. Dont introduce it too early, though. Wait until the approach is tested, the capabilities proven, and the track record of the BI team is well established.
In 18 to 24 months, you will have many new intelligence-delivery capabilities built around a team of business intelligence experts. You will have a number of defined information domains, stronger business ownership of these domains, and a number of standardized services for accessing data and building interfaces. You will have a highly effective equivalent of an enterprise data warehouse.
Make Better Decisions
In closing, a few personal words of wisdom. Developing defenses against the info glut is not easy. We already have seen some solutions are well camouflaged. Others are simply good entertainment. However, in the context of running a business, you must put the information to a simple test:
1. Is it something I dont know (and need to know)?
2. Will it help me make better decisions?
The first is a rather trivial point, yet if properly applied, it would eliminate many time-consuming debates. The second test is harder to employ. Not only must your sources be credible, but the intelligence provided must be relevant and objectivetied specifically to your goal, evaluated against all available data, adjusted to your specifics, and delivered so that it produces maximum impact.
Does your intelligence meet these tests, or are you still drowning in it?
Marek Jakubik, a former CIO of Zurich Financial and Pitney Bowes, is a co-founder and managing director of the Insurance Technology Group (www.insurancetg.com).He can be reached at 416-214-3445 or marek.jakubik@insurancetg.com.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.