‘Data-Driven’ Strategies Can Sometimes Take You to the Wrong Place

Photograph by Cameron Davidson

Like every overused phrase before it, “data-driven” has become the equivalent of the new black, or orange, or whatever color we are supposed to be wearing this year. Data-driven decision-making is now the confidence-inspiring phrase “du jour” in medicine, education, business, and government. It has a nice alliterative, hard-edged sound that lends a muscular certainty to any strategic plan, conference brochure, or presentation title. If the data are driving, it must be true that you will get to the right destination a lot faster.

I think that seductive alliteration may have misled us all into believing that data can actually drive. It is quite clear that people will often take back the wheel. In a poignant column by a young physician with advanced-stage lung cancer, he asks why physicians do not share all the pertinent data with patients about their risk of death so that they can make effective decisions about treatment options. The physicians read all the literature and assessed his chances, which were not good. Yet he chose options that were hopeful and not fully driven by the data. Why? He understood enough about scientific research to see the holes and anomalies in the evidence that doomed him.

There are also times when the data drive us to the wrong place. A “big data” application in which retail establishments use sales figures to set staffing levels daily illustrates the short-sightedness of defining a problem so narrowly that you may cause harm in the long run. Before every work shift, one young sales associate described how he had to call in to see if he was needed. He could be called in when he was not scheduled to work as well. With no guarantee of a set schedule or any certainty in his income week to week, this experienced worker had to quit. What the in-store sales approach to staffing misses is the real costs of hiring, training, and retaining a work force, undermining the viability of the retail establishments they are designed to assist.

I would like to replace the very masculine “data driven” with the softer and gentler “data guided.” I say this not because I don’t believe in the enormous power of tangible evidence to improve everyday life, but because I am deeply immersed in data and know their limits. I recognize that the evidence produced on almost every issue where it is useful is hindered by error, limited by measurement, and bounded by our current understanding of the problems at hand.

Data should be used to shape and inform decisions made by a human being who understands their vagaries. I want to ride with that human and have the data where they belong—in the GPS.

Before it's here, it's on the Bloomberg Terminal.