Four Pillars of Decision-driven Analytics

In an excerpt from their new book, Wharton’s Stefano Puntoni and co-author Bart De Langhe argue that the power of data can only be realized by leveraging human intelligence.

In a new book titled Decision-Driven Analytics: Leveraging Human Intelligence to Unlock the Power of Data, professors and behavioral scientists Bart De Langhe and Stefano Puntoni challenge the idea that our decisions should be driven by data. Rather, they argue that the power of data can only be realized by putting data in the background.

In this excerpt from their book, De Langhe and Puntoni draw from their own research and teaching to offer four pillars of decision-driven analytics.

In the mid-1850s, astronomers figured that the orbit of the planet Uranus was not like it should be according to the laws of physics. A French astronomer, Alexis Bouvard, thought that perhaps that was because we didn’t know about a planet further out in the solar system that exerted an influence on Uranus’s orbit. People started searching the sky for it. Soon enough, Urbain Le Verrier, another Frenchman, found the missing planet. It was named Neptune.

This was a great victory for the power of observation. Investing in data collection saved the day. It taught astronomers that the key to unraveling the mysteries of the cosmos was more and better data.

The rationale for making decisions with input from analytics rests on similar principles. Without data we navigate blind, while with data we can make decisions rooted in evidence. The implication is that good thinking means thinking with data.

The story doesn’t end here, though. An anomaly was soon observed also in the orbit of another planet: Mercury. The same Urbain Le Verrier who had found Neptune now hypothesized the existence of a missing planet lying between Mercury and the Sun. He called this missing planet Vulcan. Again, people started looking for it, only this time nobody could find it. Astronomers kept looking for Vulcan in the subsequent decades but the missing planet remained missing, and the mystery of Mercury unsolved.

The anomaly in the orbit of Mercury could be explained only half a century later. The explanation had to wait for Albert Einstein’s publication of a new theory of gravitation, called the theory of general relativity. This theory revolutionized our understanding of the universe by placing space and time in a four-dimensional continuum.

Although nobody knew that before Einstein entered the scene, all planetary orbits were in fact not conforming to Isaac Newton’s laws. Nobody knew that because the difference between the predictions of the two theories are smaller and smaller as you move away from the Sun. Only in the case of Mercury, which is the planet closest to the Sun, the curvature in space-time caused by the mass of the Sun was large enough for the divergence between the predictions based on Newton’s and Einstein’s theories to be detected by the telescopes of the time.

The mystery of Mercury was solved in a very different way from the mystery of Uranus. While the latter could be solved with better observations, the former could only be solved with better theory, by thinking without data.

Managers are like astronomers, looking to solve problems and find solutions in a complex world, where data is abundant but often hard to make sense of. The message is clear: Data and algorithms are crucial to making good decisions. But human judgment and intelligence are crucial, too.

The Four Pillars of Decision-driven Analytics

Many companies are witnessing an expanding gap between data and decisions, even with the goal of being a “data-driven organization.” The increasing complexity of data and algorithms can make it harder for decision-makers to collaborate with data analysts. For a business to thrive, it’s essential for both groups to understand and value each other’s expertise.

Many businesses find themselves overwhelmed by the sheer volume of data at their disposal. Putting decisions firmly at the center of the analytics process can be transformative. Starting with decisions and working back to the data will improve the quality of decision-making, improve the collaboration between managers and data analysts, and ultimately foster an organizational culture that is action oriented and that prizes the quality of decisions over ego or politics.

Here are the four core principles of decision-driven analytics:

Decisions. Identify controllable, relevant decision alternatives. Consider diverse perspectives and a wide array of solutions. Prioritize feasible and impactful alternatives to achieve important business outcomes.

Questions. Formulate precise questions that will help rank the identified decision alternatives. Ambiguous questions can lead to miscommunication and poor decisions.

Data. Evaluate the data-generating mechanism. While Big Data can be tempting, the emphasis should be on collecting relevant data.

Answers. When the earlier steps are done right, determining the best action becomes straightforward. Remember, acknowledging uncertainty and sidestepping overconfidence are key for informed decisions.

Decision-driven analytics is about making informed choices, not just processing data or flooding presentations with graphs. It emphasizes gleaning actionable insights from pertinent data. Embracing this approach means letting go of the notion that every data point is vital and not being distracted by the newest tools.

Data is just a means to an end. What matters is the decisions we make.