Define Your Path

Don’t Overdo Your Analytics

There is no question analytics are popular. After all, Davenport & Harris say that analytics are the best way to gain competitive advantage (2007). Kaplan and Norton see the data, information and insights that we use in analytics as a key part of translating our business strategies into actions (1996). With all this popularity, however, we have to ask is it possible to overdue analytics? While there are some who may disagree, my answer is an unequivocal yes.

Before we begin this discussion, I have to admit that I am a numbers guy and find the kinds of statistics we use for these efforts to be fun (as strange as that sounds). I am fascinated by the idea that changing X thing about our business processes can result in Y return. When I first learned that there was actual data showing that we could quantify the value of employee satisfaction to the bottom line, I thought it was one of the neatest things in the world (Rucci, Kim, & Quinn, 1998). Even though I personally find such ideas captivating, I have learned the hard way to recognize that there are times where the amount of effort we put into analysis can outweigh its utility. Managing that balance is one of the most important parts of our analytics efforts.

Many years ago I worked for a consulting firm that provided outsourced analytics services. We had a senior statistician who was a fairly nice guy and seemed to be well versed in advanced analysis of consumer behavior data He came across as so brilliant that many people just assumed that whatever he said was right because he used a lot of big words and half the time no one knew what he was talking about. The company liked having him around because he did very complex analyses that they could bill many hours for. Some clients like having someone so knowledgeable working for them, even if they weren’t always sure what he had done or how to apply it. There were others, however, who were not as happy for the exact same reasons.

All of this changed, however, when I took over the team. One of the reasons I was assigned to this role was due to my background in advanced statistics as part of my masters’ and doctoral level training. The first thing I did was sit down with each statistician to get up to speed on what they had been doing. Most of these discussions were pretty routine checkpoints. When I came to this particular senior statistician, however, it turned out that just about every effort he was doing was way over the top. Without getting into technicalities, it would be kind of like if the client had asked for a Honda Accord and he decided to give them a gold plated Mercedes SUV with a diesel engine. Moreover, he was taking so long to do some of these analyses that some customers had complained about missed deadlines, and two had dropped us.

As we discussed the matter I realized this individual had fallen into the trap of overdoing the analytics. He was running statistical analyses for their own sake, and not worrying too much about the insights they provided. As I coached him on how to do this better, we came up with a few simple rules that should always be applied to any analytics effort to make sure that you are not falling into this trap:

  • Rule 1: Always ask, “What is the business question we are trying to answer?” If you can’t define that business question in a clear and coherent way, go work with the stakeholders until you can. Nothing drives unnecessary analyses like not knowing what you’re looking for.
  • Rule 2: Always ask, “Is the current approach the most efficient way to answer the business question?” Analytics types (e.g., statisticians) like robust analyses because that is how we are trained. Sometimes, however, a complex analysis such as an intercostals correlation does not deliver any more meaningful value that simple correlation. In these cases, go with the simpler solution.
  • Rule 3: Always ask, “Can I explain what I am doing to someone who is not a statistician?” The simple truth is analytics are done not for other statisticians but for business executives who may only have a basic understanding of statistics. Even if they totally trust us, we should be able to explain to them what we did and how they can use it. If what we said is so complex that we can’t explain it to them without going over the unique nuances of multivariate regression theory, we probably need to tone it down a bit.

Interestingly, we soon started applying these rules for the entire group, not just this one senior statistician. As we did, we found that the entire team’s quality of work went up and the clients were more satisfied because they were getting their results faster. More importantly, the clients better understood what they were getting. Most importantly, they were better able to apply these insights to make better business decisions.


Davenport, T.H. & Harris, J. G. (2007). Competing on analytics: The new science of winning. Boston: Harvard Business School Press.

Kaplan, R.S. & Norton, D.P. (1996). The balanced scorecard: Translating strategy into action. Boston: Harvard Business School Press.

Rucci, A. J., Kirn, S. P., & Quinn, R. T. (1998). The employee-customer-profit chain at Sears. Harvard Business Review, 76, 82-97.

Related Benedictine Programs

If you’re interested in learning more about analytics, Benedictine’s MBA program provides a course on Analytical Tools for Management Decisions and prepares you to successfully lead organizations through the challenges of 21st-century business management. Benedictine University also offers Bachelors degree programs. To learn how an online degree from Benedictine can help you hone your leadership skills talk to a Program Manager today.

About the Author

Jimmy Brown, Ph.D. is a senior level management consultant with eighteen years of experience leading efforts to develop and implement practical strategies for business performance improvement. Dr. Brown has held senior level consulting positions at leading firms such as Booz-Allen & Hamilton, Accenture and Hewlett-Packard.

He can be reached at or via Twitter @jimmybrownphd