Revising an Outdated Business Model? Try Predictive Analytics

The engines of a Swiss airliner on a recent trans-Pacific flight were being monitored by sensors sending back data to the offices of SAS, a global analytics provider.


At one point over the ocean, the information going to SAS turned life threatening.

"One of the engines was in trouble," said Mike Newkirk, director of manufacturing and supply chain at SAS. "But because of the predictive analytics we were using, the plane was able to land safely in time and the engine was removed."

Most uses of predictive analytics — the gathering and analyzing data to predict future events — are not as dramatic as an engine malfunction thousands of feet in the sky. And the idea is not new — predictive analytics is as old as the Information Age itself. (More:15 Surprising Global Technology Cities)

But what is dramatic and very new is how industries across the board are embracing predictive analysis as a way to revise outdated business models.

"It's the speed of real time data being generated that makes it so appealing to businesses," said Anindya Ghose, professor and co-director of NYU's Stern Center for Business Analytics.

Just look at Macy's , and Sears , Ghose said. "Macy's used it to determine how many customers they will get at Thanksgiving, and can plan on what to sell in real time. And Sears personalized promotions for customers in a just a week when it used to take them eight. There's so much that can be done with it."

Predictive analytics works this way: Statistical information — like consumer spending, merchandise sales, machine performance, eating and drinking habits — is gathered by human input or by technical monitoring and then sent to a data base where it's sorted and analyzed under specific guidelines to reach a conclusion.

One of the well-known uses of predictive analytics is credit scoring — data models processing a consumer's credit history with the idea of predicting if they'll be able to make future payments.

But the business analytics market has grown beyond simple credit reports to a $31.7 billion market, according to IDC Manufacturing. Now, predictive analytics are used by legal firms to sort through case documents, by pharmaceutical companies to enhance sales, by health care firms to measure costs of health claims, by law enforcement to battle crime, and by casinos to keep customers betting.

For the online firm Yodle, which provides technology and marketing platforms to small businesses, predictive analysis is a way to find the right customer.

"We've been using it the last two years to evaluate which segments we should go after," said Louis Gagnon, chief product and marketing office at Yodle. "And we can figure out how much it costs to get a customer and how much it costs to keep them. We spend millions on it, but it's very cost effective."

For a textile manufacturing firm like BGF Industries, which makes technical fiber materials, using predictive analytics has become a valuable product monitoring tool since it started using the process in 2007.

"On my desk every morning is a report on all our products," said Bobby Hull, manager of corporate quality assurance at BGF. "We can look at everything across the many industries we serve and keep potential problems from happening." (More:How 3D Printers Are Reshaping Medicine)

While predictive analytics is gaining wider acceptance, some analysts say it's not a cure all.

"Rarely in business does one come across a decision that doesn't require some gut instinct and practical wisdom," said Ben Piper, president of Ben Piper Consulting, which regularly advises firms on tech issues. Piper pointed to a case where he said high tech analysis wasn't needed.

"Orbitz used Big Data to discover that Mac users spend $20 more a night on hotels than Windows users," said Piper. "But it's a well-known fact that Mac users tend to spend more than Windows users in general, and Orbitz didn't need to invest in Big Data to tell them this. Market research would have revealed the same thing at a much lower cost."

Some business execs are just plain skeptical of the whole idea, said Jerry O'Dwyer, of Deloitte Consulting.

"Many CEOs don't understand what analysis like this can do," said O'Dwyer, who is leader for Deloitte U.S. Sourcing and Procurement Service Offering. "They get frustrated when they learn that analytics can be a very detailed effort."

It's detailed work but worth it

And some companies are lost even if they have the data, said Aninday Ghose. "They don't have the skills in-house to mine it, and that's a problem with the process," said Ghose.

Ian McKinnell | Getty Images

Sometimes, getting through the process is not easy, said Hull, who has two people besides himself to work on data entry using SAS software.

"Setting up the ground rules on what data is to be analyzed can be 'bloody,' said Hull, whose firm employs some 700 people across four states. "We often have different opinions. And you can't have pre-conceived ideas on what the data will say. But once we see the information results, we all agree on what we should do."

Agreements on what to actually do with the data may be easy enough to reach, but getting there often means updating the methods used, said Michael Goul, a business professor at Arizona State University. (More:Get a Tiny Job, Earn a Tiny Payment. Repeat.)

"Some models for data analysis decay and don't seem to work anymore for businesses and they wonder why. Realizing that decay is as important and changing it is as important as having the skill sets to build the models in the first place," said Goul.

Any business thinking of getting into predictive analysts, needs to do their homework first, said Steve Jones, Global Lead for Master Data Management at Capgemini, an IT business consultant firm.

"Look at where chaotic value networks exist in your business and see what an increase in accuracy will deliver," said Jones. "Before you worry about the technology, find the business case."

But for those already immersed in using predictive analysis, the cost and effort involved are worth every penny.

"If I assigned a dollar for every data point, I'd say we got three dollars back in return," said Hull whose firm has spent around $5 million on analysis.

In the end, experts say predictive analysis will only keep growing as a business model—with some areas of analysis that have yet to be fully utilized.

"As computers get faster and data servers get bigger, more social media and crowd source data will be coming in," said O'Dwyer. "Some businesses are already posting important issues on public forums to get answers to their problems."