In the world of mining Big Data, Stephen Brossette is among those who struck gold.
Brossette immersed himself in data in the mid-1990s when he was in medical school and — at the same time — earning a doctorate in computer science at the University of Alabama at Birmingham. Drawing on insight from both of those endeavors, he worked on statistical models that merged medical data patterns with patient demographics and other information to predict hospital-borne diseases. That eventually led to the creation of MedMined, a company that the then 30-year-old Brossette co-founded in 2000 with long-time friend Patrick Hymel, also a physician-entrepreneur. The two later sold it for more than $100 million.
Brossette, 42, is now the chief science officer and a co-founder of Birmingham-based MedSnap LLC. The 2½-year-old company developed technology that accurately identifies pills and helps avoid human error in medication adherence. The system works, in short, by placing pills on a high-tech surface developed by MedSnap and then photographing them with an iPhone. An app in the phone provides positive identification of the pills. The company has plans to roll out new programs for properly monitoring and managing medication intake.
From Brossette’s perspective, “There are tons and tons of data out there, health care and otherwise, ” he says. “It’s a playground for people like me, who like to build statistical models against data.”
The amount of data being generated is practically incomprehensible, thus the Big Data tag. According to Google Executive Chairman Eric Schmidt, from the dawn of time through 2003, human civilization generated roughly 5 exabytes of aggregate information. In 2009, that much information was generated every two days.
Vast amounts of data can now not only be accessed but also linked, aggregated, organized and sold, providing new marketing opportunities not seen since the early days of the Internet. The amount of data being generated is doubling every two years, providing even more opportunities to extract, or mine, information.
Michael Hardin, dean of the Culverhouse College of Commerce and Business Administration at the University of Alabama, explains that the variety of data “is really important, because that’s where a lot of data mining challenges come in.”
Data, for example, can be extracted from relatively new digital sources: a social networking site like Facebook; warranty or product information from a company’s web site, and ideas or opinions found in blogs. That information can then be merged with more traditional sources of data, such as how many units of a product or service sold, at what price and in what zip code.
According to Hardin: “I want to know about customers, and in times past we might have tried to think through logically what customers wanted, but nowadays we have all these large sources of data, and we say, ‘OK, we think our customers want this, but what do the data say our customers want?’ And if they really want that, then how do we predict their future behavior so we can better meet their needs?”
“It’s technically difficult, so if you emphasize some of the technical skills, that’s important, ” Hardin says. “But you also have to look at it from the business side. How do you understand marketing in the environment, and what business actions do you take?”
Such thinking has led to software applications that revolve around consumer needs and wants. Internet users shopping online, for example, are often shown “You-might-also-be-interested-in” prompts to buy something similar to what they just selected, especially for music, books and apparel. More data leads to more new product and service offerings.
Working with Big Data is now common among Alabama companies. BBVA Compass is using data mining and analytics to assess performance and develop strategies throughout the organization. Among other things, that includes prospecting for new accounts and determining the best channel for communicating with customers.
“We use data mining modeling that directs us not only to identifying the best prospects but also the best products and/or services to sell them, ” says Emmett Cox, manager of customer experience campaign development at BBVA Compass in Birmingham. “There are roughly 2.5 million to 3 million small businesses out there that you can target as a client, but it’s like apples in a barrel trying to find the right one. The techniques we use provide a higher percentage of acceptance, they limit the cost of whom we’re actually going to pursue and increase the likelihood that we’re going to attract prospects.
“One of our huge tasks right now is to determine not only the solution but the communications as well, ” Cox says. “What is the channel preference for the customer? Which channel do our customers most want to be contacted on, and how do we maintain that channel preference through advanced behavioral analytics? That gets into extremely complex types of data modeling, very precise and specific modeling and merging it with different types of data.”
Al Schellhorn, senior vice president of underwriting at Alfa, says the Montgomery-based insurer committed to data mining in 2007. Although the scope of Big Data and difficulty of data mining can be staggering, it’s really all about making better decisions with the customer’s needs foremost.
Alfa collects consumer data — lifestyle and demographics information, for example — so it will have a better idea of what customers and prospects need or want. The more you know about what makes a customer tick, the better the place your company is in to serve their needs, Schellhorn says. “You can tailor your offerings to be more effective.”
Schellhorn sees a future of those companies that have data mining analytics and those that don’t. “The ones that have them are going to use them against the companies who don’t, ” he says. Allocating resources to data management “has been a reasonable investment by Alfa. It’s not easy, it’s not always cheap, and a lot of sweat equity goes into it, but we think it’s a key to our future.
“I think there’s a lot of (unstructured digital data) that is coming, ” says Schellhorn. “In our own industry, cars that drive themselves and start braking themselves are going to generate a lot of data, and we need to understand that data and have the kind of (corporate) environment that supports that kind of analysis.”
Historically, Blue Cross and Blue Shield of Alabama depended on community referrals for finding fraudulent or abusive healthcare practices. But in the past few years, most successful cases against such practices resulted from proactive data mining and analysis.
“Ultimately we are looking for patterns, trends and opportunities in our data that will allow us to establish programs, networks and products that support high quality, cost-effective healthcare for our customers, ” says Koko Mackin, the company’s vice president of corporate communications and community relations. “In addition, we utilize sophisticated analysis and data mining to detect and prevent healthcare fraud, to identify members in need of outreach, and to analyze patterns of care within disease categories that result in high medical adherence.”
Sunshine Mills, headquartered in Red Bay, has turned to new software to help improve its operations day-in and day-out. The company, which has four locations and more than 700 employees, produces dog and cat food and pet treats, most of which is sold under private labels throughout the United States and in 30 different countries. What it’s doing with data right now isn’t sexy; rather, it’s simply getting down to old-fashioned blocking and tackling.
“When we got the new software, we weren’t looking for new ways to do things, ” says Gene Matthews, Sunshine Mills’ CFO. “We wanted something to help us operate day-to-day by giving us good information so we can better plan our production for the next week. We wanted a system that gets information to us readily if something happens or changes, so we can quickly get to it by having data readily available.
“Right now, we’re looking at things like making a dashboard that includes certain information we can see on the screen instead of having to ask someone to print out a report. We’re still tweaking our system, and we’ve got a way to go. But we have seen that the more data we put into the system, the more information we get back.”
Big Data companies are well-represented in Harbert Management Corp.’s venture capital portfolio, which has more than $200 million in committed capital. For example, Clarabridge, based in Reston, Va., provides a software platform that allows a worldwide hotel chain to gauge the perception of its brand globally, regionally, by state or at the city level.
Another big data equity is JackBe, a Chevy Chase, Md. company that provides a flexible interface that allows users to define what data they want to see and what relationships they want to draw from that data. The user, for example, could structure a screen that shows real-time pricing and availability of steel throughout the world, along with shipping options and the ability to place an order. A former member of Harbert Management’s venture capital portfolio, Charlotte, N.C.-based YAP, was a pioneer in developing software that converts voice into text.
“Data is everywhere, ” says Will Brooke, executive vice president of Harbert Management and the managing partner of Harbert Venture Partners. “It is all around us, and the proliferation of devices to generate data is growing exponentially. It’s amazing what people are doing. You’ve got a number of sensors that are going to be in everything from operation machines to light bulbs to televisions, you name it.
“We’re going to be awash in data. The question is, when you start collecting available data like that, what can someone do with it that helps them better understand consumer habits, preferences, desires and helps them produce better products and also helps you target your market.”
The volume of Big Data provides opportunities, but there are no guarantees for those who mine it. “If you understand statistics and fundamentals of data analysis and model building, then this is a very exciting time, because we have lots and lots of information to build models from and lots and lots of data to test models with, ” Brossette says. “But these tools have to be used very carefully to get anything out of them. And it’s hard. It isn’t easy.
“If you don’t really understand those concepts but you have lots of data to work with, you’re going to get lost very quickly. If you understand those techniques but don’t have access to data, or you don’t understand your problem domain deeply enough, you’re probably going to end up at a dead end pretty quickly.”
All agree that having people who understand data mining and analytics is a must if you’re going to play the Big Data game. At BBVA Compass, “The biggest challenge for us in the next two to five years is going to be talent, ” Cox says. “The technologies are changing so fast, and the data realms and complexity of the data is changing so fast, real good talent is probably going to be more and more difficult to find, and that’s where relationships with universities will become more critical.”
UA’s Hardin is a specialist in data mining who has made the discipline a priority for the business school. He says his students are heavily recruited by the best Big Data practitioners in the country, starting with consumer giant Procter & Gamble. Hardin has taken students to explore an operation at P&G’s Cincinnati headquarters he calls “analytics on steroids.” P&G calls it “the Business Sphere, ” a data command room with floor-to-ceiling projector and computer displays, where teleconferences are held and managers can immediately drill down to real time information about the company’s business anywhere in the world.
Questions flow in meetings held in the Business Sphere. What are a detergent’s sales in a particular country? Who are the competitors there and how are they doing? After a mouse click or two, the information pops up, and it’s there for everyone to see.
“They say there’s nowhere to hide in those meetings, ” Hardin says. “The data is what it is, and sometimes the CEO is sitting there when all these tough questions are being asked. But then it becomes a matter of problem solving, and they can start looking at data as part of solutions to a given situation.”
Hardin points out that an Alabama graduate who went to work for Procter & Gamble played a key role in the development of the company’s Business Spheres, which can be found at different P&G offices. Alabama’s business school is one of Procter & Gamble’s preferred colleges for recruiting business analytics talent, and several Alabama graduates now work at North Carolina-based SAS, a leader in business analytics software and services. Alabama has had a relationship with SAS since 2002, and the university’s Big Data graduates can also be found at Regions, BBVA Compass, Alfa and other Alabama companies.
To prepare them for the real world, Hardin would like to see more Alabama business students working with companies on collaborative projects prior to graduating. He likes the chances for those studying data mining and business analytics.
“Big Data is here, ” Hardin says, “and it’s only going to grow.”
Charlie Ingram is a freelance writer for Business Alabama. He lives in Birmingham.
Text By Charlie Ingram