Correlation Does Not Equal Causation

When we make decisions in business, there's often a lot of elements to consider. Whether it be economic climate, organizational culture, financial return, or simply competitive pressure, these decisions can be complicated. So we go to the "big data" to give us the answers. But data itself can't provide you with the right answer, but simply more elements to consider when making the decision.

We have an obsession with data, and often times lean on it to justify decisions, whether it be spending more money on digital advertising or investing in a new acquisition. However, data isn't the magic bullet we often think it is. We need to look deeper, and not only learn how to better understand what data is telling us but more importantly, what questions to ask.

For example, an organization might be looking for ways to build up their sales numbers. They start to examine their online site traffic, buyer behaviors and purchase patterns. In the data, they identify the areas where they're having the strongest success and decide to double-down investment and efforts in those areas.

However, the problem is, the data is only the data. The leaders are not asking the right questions, and therefore, making the assumption that since this area is already successful, simply more of the same will increase sales further. But correlation does not equal causation. Causation explicitly applies to cases where action A causes outcome B. On the other hand, correlation is simply a relationship. 

The leaders are seeing a relationship in the data, i.e. correlation, where "whatever is working here" must be right, and we should do more of the same. But they aren't understanding the causation - what exact actions are causing the outcomes. The cause could be a wide variety and combination of things, from a stellar-performing territory manager, product seasonality, geographical influences, price point, and much more. In short, the question of "how to increase sales" is much more complex than simply examining the surface performance data.

It's not that data is all bad, and that as leaders we should simply 'trust our gut' to make decisions, but rather, take the time to understand what data we need to make a better-informed decision, rather than simply take what is generated by our systems at face value. As leaders, we need to understand that not everything the computer churns out is reality. There are many more dimensions often not considered by the data itself. If you don't fully understand the idea or challenge you want to address, how do you know what kind of data you need, how much, or what other critical questions you should be asking?

So while data has it's place, bigger data or better data will not give us the easy answers to our business challenges. Mindlessly just collecting and processing it can lead to misleading conclusions, building correlation assumptions instead of causation-driven ones. And it definitely doesn't lead to understanding. We must not lean solely on the data as a scapegoat to justify our decisions, but rather examine what data we truly need to help guide us by understanding the challenge better and asking the right questions.

About the Author

Andrea's 22-year, field-tested background provides unique, practical approaches to creating more efficient, more competitive, customer-centric organizations. award-winner, she began her career at a tech start-up and led the strategic sales, marketing and customer engagement efforts at two global industrial manufacturers. She now leads a management and communications consultancy, dedicated to helping organizations transform their organizational cultures from "internally-focused" to "customer-centric".

In addition to writing and consulting, Andrea speaks to leaders and industry organizations around the world on how to craft effective customer-centric organizations. Connect with Andrea to access information on her book, workshops, keynote speeches, training or consulting. More information is also available on www.pragmadik.com or www.thecustomermission.com.