Evangelizing Analytics Through Baseball

A great many people are simply allergic to mathematics. Maybe even more so than to public speaking. They just hate the subject and the very thought of it gives them a big headache. In many cases though, I just have to blame their math teachers in their youth for not providing enough appreciation for the subject, as the same people have no trouble understanding baseball stats of their favorite teams and players. What do you think that ERA (Earned Run Average) stands for? It is nothing but an index value made of a numerator and a denominator, multiplied by a factor. If you can paint the quality of a baseball pitcher with a bunch of statistics and indices like that, yes, you do have a basic aptitude to be analytical. Maybe not enough to be a professional analyst, but enough to be a consumer of analytics. In the near future, that kind of basic aptitude may be all we need to navigate through this complex world weaved in numbers and figures originated from humans, machines and networks that connect them all. All the headachy equational problems will be taken care of by the smart machines anyway, right? Maybe. The way this old analyst sees it, the answer could be yes or no. Because there is no way for any machine to provide good answers to illogical questions (like Mr. Spock would point out). And the logical mind comes from mathematical training — with a little help from the DNA with which the subjects were born. Why do I worry about such things now? Simple. I see too many decision-makers who say they must get more into analytics, and their behaviors tell otherwise. It is unfortunate for them, as the verdict is out already on the effectiveness of good analytics. In fact, the question is no longer about whether an organization must embrace more analytics-based decision-making processes, but about how deep and complex they must get into it. The winners and losers in the business world will not solely be determined by the business models, but by the effectiveness of execution, enhanced and measured by analytics. Gut feelings may have worked well for many in the beginning of the last century, but that won’t be enough when competitors are armed with data and analytical toolsets. We are undoubtedly living in the complex world now. The differences between people who freely wield technology and toolsets and people who are afraid of the changes won’t be just income levels; it may even form new social classes. And yet, the way many are dealing with perceived and real challenges isn’t much different from the past era. No, you can’t just work hard and hope that everything will be alright. There are people who see what is coming before anyone else does. Not all may be able to “see” it completely per se, but at least some have better future prediction than others. And those who do properly employ predictive analytics will clearly have an edge over those who don’t. Then, why is it so difficult to “sell” analytics? There are many reasons. The first one, I think, is the fear of unknown (or unfamiliar territory). People hate to spend money and resources on the things that they don’t understand. The majority of the population does not understand how the internal combustion engine works, so the car companies sell coolness and other perceived benefits of their products. Unfortunately, there is absolutely nothing sexy — for the general population, not for the geeks — about algorithms that may increase sales and reduce costs. So, the analysts must try to emphasize the benefits of it all; yet too many fall into the trap of believing that everyone will appreciate the beauty of the solution they agonized over. Well, most people simply don’t care for the details. That is why engineers don’t sell cars, but salespeople do. Analysts must get to the point fast; possibly within a minute, as most don’t have the patience for anyone’s mathematical journey. Another reason why selling the concept of analytics is difficult is collective resistance to change. For most people, change is scary; and even if it's not, it's terribly inconvenient. All organizations and people in them are accustomed to some existing ways of doing their businesses. Analytics inevitably invoke changes in existing behaviors. It may start with a simple request for some extra data collection (“What do you mean I have to enter more data into my Salesforce account?”), and ultimately move onto changes in decision-making processes (“Oh, now I have to look at those model scores while I’m on the phone with the customer?”). At times, even the folks with titles that include words like IT or analytics put up resistance. Maybe their bosses bought into new and improved ideas in dealing with data and analytics, but people who actually touch the data may feel that such change simply poses more work for them, as in, “Great. Now I have to create yet another data stream for this stupid analytics vendor, too?” Analysts who thought that their job was to look at a few sets of reports may have to monitor more and dig deeper into figures, and they won’t see much organizational benefit. But they will see how late they have to leave work now. This is exactly why even an executive must understand the basic mechanics of the data and analytics business. Otherwise, ideas for potential improvements will actually cause more chagrin for his staff by adding more tasks without any common understanding of the benefits of the new analytical practice. Maybe that is why some data professionals call it evangelization of analytics. Without the buy-in, the analytics itself is just a nuisance to many. I hear excuses for not employing better data and analytics practices all the time. To name a few (all these are for real): We don’t have the budget for it; we spent the money elsewhere. Our data sources are so messed up, you wouldn’t know where to start. We already have enough reports around and we are not even making sense out of them. You want to add new analytical stuff? We’ve tried some statistical models, and they didn’t yield good enough results. We are relying on this new software that is supposedly doing all analytics for us. Allow me to argue each point here: Can we at least take a look at your current practice? In fact, budget optimization is an important part of analytics, too. Where did you spend money, anyway? Modern analytics is about making sense out of unclean and unstructured data. And yes, good analysts would know exactly where to start when confronted with messy data (see the title of my home page here). Maybe you are looking at a wrong set of reports. Less is more in the world of good reporting, as no manager has time to look at a 100-page report on a Monday morning. What was the purpose of the statistical model, and how was it used? One shouldn’t dismiss predictive analytics with one bad experience. Good analytics is about incremental improvement by learning from the past exercise. The new world is indeed about Big Data, analytics, and automation of it all. But the machines don’t read human minds yet, and even good toolsets must be tuned properly, based on test results. How does 80 percent automation-20 percent customization sound to you? Analytics in isolation doesn’t not benefit any organization. It of course starts with data and certainly involves lots of number-crunching; but in the end, it must be the essential and consistent part of a decision-making processes on all levels. I’ve met some CDOs and CAOs who have built so-called state-of-the-art data and analytics infrastructures, but still agonize over the fact that not all parts of their company are employing the fruits of their labor (refer to “Putting Data to Use”). At such point, analytics stops being a matter of math and technology, but of organizational commitment, like baseball. And such commitment starts at the top, and also requires buy-in from everyone. Data professionals, for that reason, must continue their efforts to make data smaller and smaller, into easily consumable bite-sizes, in forms of answers to everyday questions. Now, that would be the technical and mathematical part of the job. Still, an even harder job to be done is to convert non-believers into believers. Fans, even. For that, professional analysts must drop the technical pretense and start preaching only using plain language. After all, evangelization is about the audience and the passion for the subject, not about a few practitioners and display of superior technical knowledge. Like I said in the beginning, math is not all that sexy for most, anyway. So always make it short, simple and sweet. Score a home run. Your service as an analyst may no longer be required if no one is buying it.

Data

Evangelizing Analytics Through Baseball

Stephen H. Yu

About

Subscriptions

Events

Publications

WORK SMARTER - LEARN, GROW AND BE INSPIRED.

Subscribe today!

Stephen H. Yu