Data is Worthless if You Don't Communicate It

by Tom Davenport 

There is a pressing need for more businesspeople who can think quantitatively and make decisions based on data and analysis, and businesspeople who can do so will become increasingly valuable. According to a McKinsey Global Institute report on big data, we'll need over 1.5 million more data-savvy managers to take advantage of all the data we generate.

But to borrow a phrase from Professor Xiao-Li Meng — formerly the Chair of the Statistics Department at Harvard and now Dean of the Graduate School of Arts and Sciences — you don't need to become a winemaker to become a wine connoisseur. Managers do not need to become quant jocks. But to fill the alarming need highlighted in the McKinsey report, most do need to become better consumers of data, with a better appreciation of quantitative analysis and — just as important — an ability to communicate what the numbers mean.

Too many managers are, with the help of their analyst colleagues, simply compiling vast databases of information that never see the light of day, or that only get disseminated in auto-generated business intelligence reports. As a manager, it's not your job to crunch the numbers; but — as Jinho Kim and I discuss in more detail in Keeping Up with the Quants — it is your job to communicate them. Never make the mistake of assuming that the results will "speak for themselves."

Consider the cautionary tale of Gregor Mendel. Although he discovered the concept of genetic inheritance, his ideas were not adopted during his lifetime because he only published his findings in an obscure Moravian scientific journal, a few reprints of which he mailed to leading scientists.

It's said that Darwin, to whom Mendel sent a reprint of his findings, never even cut the pages to read the geneticist's work. Although he carried out his groundbreaking experiments between 1856 and 1863 — eight years of painstaking research — their significance was not recognized until the turn of the 20th century, long after his death. The lesson: if you're going to spend the better part of a decade on a research project, also put some time and effort into disseminating your results.

One person who has done this very well is Dr. John Gottman, the well-known marriage scientist at the University of Washington. Gottman, working with a statistical colleague, developed a "marriage equation" predicting how likely a marriage is to last over the long term. The equation is based on a couple's ratio of positive to negative interactions during a fifteen minute conversation on a "difficult" topic such as money or in-laws. Pairs who showed affection, humor, or happiness while talking about contentious topics were given a maximum number of points, while those who displayed belligerence or contempt received the minimum. Observing several hundred couples, Gottman and his team were able to score couples' interactions and identify the patterns that predict divorce or a happy marriage. 

This was great work in itself, but Gottman didn't stop there. He and his wife Julie founded a non-profit research institute and a for-profit organization to apply the results through books, DVDs, workshops, and therapist training. They've influenced exponentially more marriages through these outlets than they could possibly ever have done in their own clinic — or if they'd just issued a press release with their findings.

Similarly, at Intuit, George Roumeliotis heads a data science group that analyzes and creates product features based on the vast amount of online data that Intuit collects. For his projects, he recommends a simple framework for communicating about each analysis:

1. My understanding of the business problem
2. How I will measure the business impact
3. What data is available
4. The initial solution hypothesis
5. The solution
6. The business impact of the solution

Note what's not here: details on statistical methods used, regression coefficients, or logarithmic transformations. Most audiences neither understand nor appreciate those details; they care about results and implications. It may be useful to make such information available in an appendix to a report or presentation, but don't let it get in the way of telling a good story with your data — starting with what your audience really needs to know.

Tom Davenport

Thomas H. Davenport is visiting professor at Harvard Business School, the President’s Distinguished Professor of IT and Management at Babson College, and a research fellow at the MIT Center for Digital Business. He is coauthor of the new book Keeping Up with the Quants and the best-selling Competing on Analytics.