Is big data hurting your business? 

Though big data has become the au courant method for decision-making in business, Gerd Gigerenzer, Director of the Max Planck Institute for Human Development and author of Risk Savvy, has argued that too much data-driven decisions could hurt companies. 

We recently had the pleasure of attending one of his talks on heuristics and optimal decision making at the University of Pennsylvania where he discussed his ‘fast-and-frugal’ method, which consists of a set of heuristics, or rules of thumb, designed to reduce the need for big-data-driven decisions and provide decision makers with intuitive and simple ways to make better decisions under conditions of uncertainty. Here’s what we learned.

 Imagine you’re an ER doctor. A man is rushed to the hospital with severe chest pain. You must make a life or death decision: Is this a heart attack? If so, he should be rushed to the coronary care unit, if not sending the patient into coronary care is not only costly, but will divert care from those most in need, as well as increase his risk hospital-transmitted infection.

A defensive decision-maker would send anyone displaying symptoms of chest pain into coronary care, after all nobody enjoys being on the receiving end of a law suit. This is why doctors at a rural Michigan hospital had been sending 90% of patients reporting chest pain straight into coronary care— a practice which compromised quality of care and and patient health.

The problem was complex, so a team of researchers presented an equally complex solution: they tracked some 50 indicators using a logistic regression. Physicians could use this complicated equation to estimate the probability that a patient should be transferred to coronary care. However, few physicians actually understood the math and researchers soon found that they performed just as well once the system was abandoned. Why? Through the equation, doctors actually learned to track a handful of key indicators which more or less reliably predicted the probability of cardiac arrest. 

This lead Green & Mehr (1997) to design a short sequential model, a so-called fast-and-frugal tree, comprised of 3 questions organized into a decision tree to decide whether a patient should be admitted into coronary care or not. Though the tree relied on only a small subset of the available information, it significantly outperformed the previous, complicated model. This short and intuitive method empowered physicians to make on-the-fly decisions that saved patient lives and increased overall quality of care. 

According to Gerd Gigerenzer when a situation is uncertain, lends many possible solutions, but only offers small amounts of on-hand data, simple heuristic-based decisions tend to outperform those which take into account large amounts of data in complex ways. In his view, these fast-and-frugal trees provide a means to filter out noise while concentrating only on the most crucial cues needed to solve a problem. 

This isn’t to say that every problem requires a quick rule of thumb, or heuristic, to solve it. In situations where uncertainty is low, the alternatives are few, and data abounds, complex solutions will likely outperform heuristics— astronomers, for instance, whose field of study is known as being relatively stable, benefit from big data approaches (see Katsikopoulous, 2011 for other potential environments). 

Recently, Gigerenzer has partnered with the Royal Bank of England fast-and-frugal to create a decision tree that could have helped avoid the 2008 financial crisis. Rather than feeding millions of data points into a single, complex model the tree tracks three important indicators in a one-strike-out fashion (Aikman et al. 2014). His lab has also helped the German military provide with a fast-and-frugal tree to decrease the number of accidental civilian deaths at military checkpoints (Keller & Katsikopoulos, 2015)

Gigerenzer argues that the less is more approach applies to companies as well. While executives and managers find comfort in hiding behind big data-driven decisions, these tend to be subpar for business. This defensive decision-making, as Gigerenzer calls it, protects the decision maker from blame, but more often than not, hurts business. When humans, especially experts, base decisions on intuition, that gut feeling, though subconscious, usually arises from years of experience. Nonetheless, we often prefer data-backed decisions to our intuitions, because when things go South, we are able to point to a trail  paper trail as justification. Still Gigerenzer argues, when uncertainty is low, intuitions and smart heuristics outperform data-driven decisions. 


Angelia Buerkin-Salgado is Director of Research at Lausanne Business Solutions, a comprehensive talent management consultancy. Lausanne helps organizations recruit top talent, increase employee engagement, while optimizing performance, and develop individuals and teams. You can like Lausanne Business Solutions on facebook or follow us on twitter @lausanneconnect,  instagram@lausannesolutions, and also follow Angelica on twitter@AngelicaBuerkin.