Marketing IT In-House: Don’t Let BI Failure Statistics Stop You
Whatever the actual BI failure rate is, you don’t need to fear it.
By Max T. Russell, Max and Max Communications
I have scraped around for reliable evidence, and I still don’t know what the BI failure rate is. I don’t think anyone does. It varies according to whom you ask. One estimate I found was 75 percent, another source claimed the figure was closer to 30 percent, and a third says the right figure is 50 percent. Some sources even say 90 percent (or higher).
That kind of figure can be paralyzing. Why start a project if you’re bound to fail? I do know that if your BI team is not behaving in a way described by the statistic, the failure rate is simply irrelevant to your outcome.
When estimates vary so wildly, people are clearly not talking about the same populations or measuring failure in the same way. A statistic cannot describe everyone. It describes a specific population of events or characteristics.
Your BI team can choose behaviors that don’t end in failure. Before looking at some of those, let’s take a closer look at our attitudes toward statistics.
A Careful Use of Statistics
Most people throw statistics around as though their numerical values automatically carry authority. I rarely hear a statistic used in an accurate, meaningful way, and I sometimes upset people by questioning the statistics they quote as interesting or authoritative, because they usually are neither.
For instance, someone might say, “I heard that 95 percent of all marsh hawks live to be 25 years old. Isn’t that interesting?” No, it isn’t, because it’s not even true. Especially in BI, it’s counterproductive to be interested in things that are not true.
One of my measurements professors at Purdue, Ernie McDaniel, made the following statement at his retirement reception: “We say it all the time: correlation is not causation. Unfortunately, people (i.e., researchers) still get it wrong!”
I’m sure his own professors told him the same thing.
Educated people also continue to think that statistics have meaning just because they get reported at conferences or in the media. Let me restate Dr. McDaniel as follows: “We say it all the time – statistics don’t necessarily mean anything. They have to reflect an actual population, and they have to reflect it accurately and meaningfully.”
That kind of ranting isn’t going to fix the problem, either. Statistics are simply too irresistible and handy for most people to ignore them or to investigate their validity. However, even if 90 percent of all BI efforts did end in failure, the statistic would not touch you if your behaviors were consistent with successful BI.
With that in mind, let’s look at several behaviors that BI expert Roger Cogswell says lift you above the fateful statistic. Roger and I have spent several years working together on BI communication issues. He identifies the following sequence of behaviors as a road to BI success. I will use them to illustrate why they make the failure statistic irrelevant to your team.
Success Behavior #1: Identify a Meaningful Business Opportunity
This becomes the driver for everything. You must have access to decision makers and the company’s information in order to see if, in fact, an opportunity exists. If there’s no verifiable opportunity for impact, you need to pack up and go home. You have avoided failure.
If opportunity is visible, you can start aligning everything in your plan. Roger and I strongly believe that IT should not try to sell executive management on an initiative unless IT has the skills to communicate effectively about the business to the businesspeople.
Success Behavior #2: Align Everything
Approach the opportunity as a meaningful experiment -- scientifically. Your laboratory is your data architecture, which must be constructed to run experiments that can help you achieve your business objective.
For example, if you intend to optimize the procurement for medical research facilities, you will have to capture costs and descriptions of a wide variety of medical equipment and animal subjects. You will need to set up controls to ensure that you are taking measurements that will provide reliable analysis over the experiment’s duration. You will also need to make sure that your calculations are meaningful and understandable to users.
Do not waste time doing anything that doesn’t take advantage of the opportunity.Do not take time to be interested in anything that is not true.
A sloppy project will aggravate users and fail to impress user management. I am currently watching a large BI project torture users by making them suffer through the same design problems for the third year in a row. Rather than aligning the users’ needs with the stated overall goal, IT is spending its time doing what interests it most. Users have a sense of BI failure every day. Alignment would turn them into friends of the program.
Success Behavior #3: BI Team Members Know What to Expect
All team members must remain aware of what to expect while running the experiment so they can support the decisions the team makes as individuals and as a group. Too often, nobody is measuring the decisions, and impact is unknown.
When everyone sees the potential impact and behaves with transparency (no secrets), they are in line with success.
Team members need to be accustomed to transparency. They cannot operate on their own feelings, they must proceed scientifically, and they must hold themselves accountable. As Roger says, that’s scary to people who want predictability instead of experimentation.
Success Behavior #4: Embrace Change
There must be a culture of change if your enterprise is to keep up with the market. If the competition knows how to target customers better, serve them better, and run more efficiently than you do, then how will you compete?
Mergers are good examples of the importance of a culture of change because they join people of different organizations and expose levels of tolerance. Successful BI turns away from failure by leading and building on a readiness to embrace change.
Success Behavior #5: Use the Same Language
When IT and users use the same terminology, departments can talk to each other about their BI experience. They’re able to see the same things and to evaluate and appreciate the same things. Roger notes that when it comes to discussing data, different departments often use different language, which causes confusion. Successful BI plans are characterized by a common language.
Success Behavior #6: Know What Success Looks Like
You have to see the target in order to hit it. To see it, you have to know the business. In a successful BI plan, users want it to work because they see the mutual benefits. You often have to create that picture in their minds until they see it in operation. The better they see it, the more easily you can make it happen. You are defying the failure statistic.
A Final Word
So there you have it -- six behaviors that make the failure rate absolutely irrelevant to your BI effort. Go forth boldly. Change your enterprise one department at a time or one set of operations at a time with failure-defying behaviors.
You will have many failures along the way, but they are part of the road to success when you evaluate them and make smart adjustments. That is the most important difference between BI failure and success.
Max T. Russell is the owner of Max and Max Communications. He works behind the scenes to promote individuals and projects in a variety of industries. He and his identical twin, Max S., have been discussing and dissecting the challenges of IT in the workplace for the past 18 years. You can reach him firstname.lastname@example.org.