Business Information

Technology insights for the data-driven enterprise

agsandrew - Fotolia

Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Gazing into the future of predictive analytics ROI

Companies are investing in the software to gaze into the future and learn more about customers and business opportunities. Is it prophesy -- or fallacy?

Predictive analytics is all the rage right now, and it's easy to see why. After all, who wouldn't want to know the future? But perhaps it's time to gaze into our crystal ball to ask what kind of value businesses really reap from predicting the future. What is the predictive analytics ROI?

Predictive modeling has become popular in marketing. It's now possible for marketers to predict that people who have searched online for, say, socks, may be interested in seeing ads for socks in the future. It has also enabled marketers to finally indulge their darker sides. After all, who wouldn't want to follow the clickstream of an average Web user? I'll bet that turns up some interesting insights! And anyway, users are mostly fine with the practice. People generally take stalking behavior as a sign that they're loved, right?

Healthcare is becoming another major user of predictive analytics. People are using it to predict all kinds of things, like which discharged patients are likely to be readmitted, and which diabetic patients have not received recommended preventative care. But given the shape healthcare is in as an industry, it may be just as good to assume all discharged patients are likely to be readmitted and every diabetic patient has not received recommended preventative care. N=All!

It goes to the deeper point of the value of prediction. In the 18th century, French astronomer Alexis Bouvard successfully used a mathematical analysis of unexpected changes in the orbit of Uranus to predict the existence of Neptune. While the discovery may be interesting, it didn't exactly lead to any new revenue for anyone; humanity is still waiting on the ROI from that one.

In 1996, then chairman of the Federal Reserve Bank and general sage of the stock market Alan Greenspan predicted there was irrational exuberance in the market, which would inevitably lead to an economic slowdown. Some commentators at the time said this statement may have actually contributed to even greater irrational exuberance -- people who weren't in the market thought, if everyone else is in, they should be too. Predictably, the Greenspan-induced exuberance contributed to the stock market slowdown in 2000. In the end, it was just Greenspan's prediction that was irrational.

Nate Silver made headlines in 2012 when he accurately predicted state-by-state returns in the presidential election as a columnist for The New York Times. Other than helping to advance his own career -- he has since stepped up to editor-in-chief at ESPN-supported -- his predictions didn't influence much. I think Karl Rove is still waiting on results from Ohio.

Proponents of predictive modeling like to talk about how data speaks for itself and how it helps make decisions objective. But as the Greenspan example shows, there are always biases in data, and sometimes the very act of trying to predict an outcome changes that outcome.

So does this mean predictive modeling has no value for businesses? Will organizations who have bet the house on predicting the future ever see an ROI from predictive analytics?

Only time will tell. But my guess is businesses who ask themselves this question are on the right track. Those that use the technology simply because it's the newest, latest-and-greatest thing, may come up with some interesting predictions, but these predictions will have little value. But those that have more specific uses in mind will find real value from the technology.

At least, that's my prediction.

Ed Burns is site editor of SearchBusinessAnalytics. Email him at and follow him on Twitter: @EdBurnsTT.

Article 10 of 10

Next Steps

Predictive modeling is more than a math problem

Predictive modeling and big data can be a tough fit

When is enough data enough in big data predictive analytics?

Dig Deeper on Predictive analytics

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.

Get More Business Information

Access to all of our back issues View All