Getting Smart With Big Data

Sub-Zero, a mid-sized company based in the Midwest, has been able to create a brand of stainless steel, side-by-side refrigerators that have become status symbols for Hollywood starlets and Wall Street investment bankers alike. Its reputation for 20 years of reliability is one reason Sub-Zero can command premier pricing and compete against much larger rivals, such as Whirlpool and General Electric’s appliance division. One of the keys? Sub-Zero is a master of manipulating Big Data.

Like a growing number of other small and medium-sized companies, Sub-Zero has learned how to move beyond simply analyzing structured data, the type that exists in neat little rows and columns in a spreadsheet. It is now also analyzing “unstructured” data from customer service calls and warranty claims. Unstructured data does not exist in numerical form and often has to be massaged into a form that allows it to be blended to create a coherent picture of a company’s operations. Data-savvy companies are now also analyzing social media and capturing information from mobile devices, which are other signs of growing sophistication.

When smaller companies reach the big leagues of data analysis, it pays off, often giving them capabilities that larger firms lack. In the case of Sub-Zero, the company uses software from SAS, based in Cary, North Carolina, to search recordings of customer service calls for word pairs such as “compressor” and “failure.” The system applies “fuzzy logic” to analyze fields of text. If the system detects a problem, the company seeks to match the information up with the serial number of the refrigerator in question to determine exactly when it was made and what parts from which vendors were used. It also mines warranty claims to look for any patterns of quality or reliability problems. That detailed information flows back to product development and reliability teams, who are then able to improve the way the refrigerators are made and work out any kinks with suppliers. In one specific case, the company used its data system, which was upgraded last year, to identify a risk of failure in an electronic component many months earlier than its older system would have allowed. Thanks to that insight, Sub-Zero reckons it shipped 5,000 fewer refrigerators with risks of failure, avoiding as much as $250,000 in warranty expenses.

The end result is that Sub-Zero, a privately held company based in Madison, Wisconsin, with an estimated $700 million in annual sales, competes robustly against multi-billion-dollar giants, whose big legacy systems make it harder for them to exploit their data. “Our investment in this hardware and software continues to pay dividends,” says Scott Lafleur, Sub-Zero’s COO. “We have been able to use these advanced tools to significantly improve the accuracy of our data analysis, which—in turn—allows us to identify improvement opportunities in our design and manufacturing processes.”

According to IBM, the world is awash in quintillion bytes of data—so much that 90 percent of the data that exists today has been created in the past two years. The data comes from everywhere: posts to social media sites, digital pictures and videos, purchase transaction records, cell phone records, RFID sensors and GPS signals, to name a few.

Small and medium-sized enterprises (SMEs) can often leapfrog larger rivals in analyzing this torrent of data because they are nimbler and don’t have large, expensive legacy systems. Larger companies also suffer from “siloes,” or different arms of the company that capture different sorts of data about customers or suppliers but are unable to consolidate that data to provide genuine insights. “Our experience tells us that smaller companies are doing a better job of integrating and optimizing cross-channel data, due in part to their ability to be more agile and make quicker decisions than their enterprise counterparts,” concluded a briefing paper by marketing and analytic firm 89 Degrees. The company, based in Boston with less than $50 million in annual sales, uses SAS software to provide services to other companies.

One underlying factor allowing SMEs to analyze data better is the increase in raw computing power, allowing more data to be “crunched” faster. “Not too many years ago, a small business couldn’t access this kind of talent and data-crunching capability in an affordable manner. You had to be a Fortune 500 company,” says Bruce Rogoff, CEO of GroundLink, a New York-based global airport limousine company with more than $30 million in sales. It works with 89 Degrees to analyze its best customers and how to market to them. By combining GroundLink’s own information with databases on credit card use, credit ratings, home location, hobbies and even preferences in pets, 89 Degrees helps GroundLink zero in on marketing that works.

In addition to raw computing power, the information technology industry has made rapid strides in improving its analytical software. Only two years ago, for example, IBM unveiled its Watson computer that was able to defeat human contestants on the television game show, “Jeopardy.” Now the same software is available in IBM’s new Power7+ servers for prices starting at about $6,000. “We’ve taken the intelligence of a Watson and put it in systems that are priced to the small and medium-sized company range,” says Ed Abrams, vice president of IBM’s Midmarket Business, based in Norwalk, Connecticut. “The small and medium-sized business, when armed with big data analytics, can drive true change and true innovation into the marketplace.”

Not surprisingly, providing analytic hardware and software to SMEs is one of the hottest spots in the entire IT industry. It’s a very fractured industry, but IBM says its sales of analytics to SMEs are growing at a faster rate than IBM’s sales as a whole. It is competing aggressively in the space by offering $4 billion in financing to buyers of its systems. SAS is a long-time big player in the space, but EMC, SAP, Oracle and every major IT company are targeting this sector.

Even decidedly low-tech companies are realizing gains from modest investments in data analytics. IBM’s Abrams tells the story of a California distributor of cantaloupes and honeydew melons that acts as a middleman between growers and wholesale and retail customers. “They’re using predictive analytics to look at heat indexes, the size of picking crews at various growing sites and other factors to get a very good sense of the inventory they will acquire,” he explains, noting that a single manager working with an IBM Power7 systems handles the process. “That helps them to know how to price the fruit and tells them how much they can provide to retailers and wholesalers.” The size of picking crews is considered unstructured data, while structured data includes the number of bruised or damaged pieces of fruit.

It seems that when an SME company combines structured and unstructured data, it is able to achieve higher levels of insights than it otherwise can. Fiona McNeill, global product marketing manager for text analytics at SAS, says her company had a customer that evaluated insurance claims in hopes of minimizing workplace injuries in police departments, among other workplaces. The system analyzed handwritten accounts of hundreds of police officer accidents and found a pattern. Officers, particularly women, were suffering wrist injuries when they were putting an offender in the rear seat of their cars. The insurer suggested that all officers wear wrist guards. Result? A major reduction in accidents.

Of course, the most obvious gains from advanced data analytics are coming in the retail world. Digital advertising agencies can evaluate the success of a company’s television advertising or online campaign within the first few hours by analyzing how much “chatter” there is about it on social media. Advertisers can tell whether a campaign is going to be successful or not and modify it or pull it altogether. Retailers that create their own Facebook pages can monitor what teen-aged girls are saying about their latest fashions in real time and redesign the clothes to be even more attractive to them.

The Holy Grail in this emerging industry is integrating data from multiple channels, including social media, e-mail click-throughs, databases and direct mail, says Phil Hussey, president and managing partner at 89 Degrees in Boston. His company serves big companies, such as IKEA and Hyundai Motor, as well as smaller companies and it relies on SAS software.

The key is measuring a customer’s “engagement.” “We try to match up and understand what customers are doing across multiple channels,” says Hussey. “We can intuit a lot of things they are doing in [the] social sphere. We can measure how customers engage with the brand through different channels. The more channels they engage in, the better customers they are and the more likely they are to shop.”

Those types of insights allow even small and medium-sized retailers to create profiles of their customers and to keep track of their preferences and buying patterns, much as does. That type of more specific “behavioral” marketing is much more effective. The click-through rate in response to a traditional email advertising campaign might be 5 percent of customers, says Hussey, but it is 15 to 25 percent for behaviorally targeted emails.

Even a true believer like Hussey warns, however, that companies can make mistakes in plunging into big data analytics. One caveat is that some companies collect data but then don’t know how to take action on what the data is telling them. They become paralyzed. “If the data isn’t put into action, you’re not going to get a rate of return because then it’s just data for data’s sake,” Hussey says. “You’re not truly engaging and interacting with customers.”

Another trap is not building in the capacity of staff to interpret mountains of data. “I don’t believe any software allows you to just push a button and get all the right answers,” Hussey says. “The software world takes you up to a certain point; and then, it’s up to analysts, who take you the rest of the way.” He notes that there is a national shortage of statisticians and data analysts.

When done correctly, building data sophistication into an SME company can allow it to punch above its weight. Very few CEOs worry about their rates of return on this type of investment because it fundamentally changes the nature of the company and its products, as happened at Sub-Zero. CEOs can see huge payoffs in productivity, product quality and innovation. The bottom line? It makes big sense to get smart about Big Data.

A Data-Driven Airport Service Company

GroundLink, a limousine car service based in New York City that caters to “road warriors” around the world, has come up with some winning technologies on its own steam, thanks, in part, to research and development offices in India and Serbia. GroundLink allows customers to order pickups at airports from their smart phones and then track where their driver is on the screen of their handheld devices. That avoids the age-old dilemma of arriving at an airport, not finding your driver, calling the limousine company to find the driver but never really knowing when the car will show up. Altogether, GroundLink handles more than 1,000 pickups a day in 75 countries around the world.

CEO Bruce Rogoff relies on an outside vendor, 89 Degrees, to really dig in and try to understand who his best customers are. “We want to get more of the best customers and pay less attention to the others,” he says. “We are a data-driven organization.”

GroundLink provides its own data to 89 Degrees, which then combs through various public and private-sector databases, including those of credit-rating agencies to draw profiles of the most active customers. Thanks to its advanced analytics software and hardware, 89 Degrees can overlay GroundLink’s data with that of many other databases. “Are my customers investment bankers in Westchester [an affluent county north of New York City] or consultants in Boston?” Rogoff demands to know. “What is their job title and household income? Do they live 10 miles from airports or 20 miles? Are they tennis players? Are they platinum Marriott people? Do they have a particular status with United Airlines?” One discovery: a surprising number of his customers own golden retrievers.

Once it has an accurate portrait of a customer segment, GroundLink tries to craft a specific marketing campaign to reach them. “It’s not just knowing that I serve a bunch of frequent fliers; it’s a question of, how do I speak to those people?” Rogoff explains. So the company started offering 1,000 airline bonus miles to customers who were found to have a particular loyalty to an airline or a particular interest in amassing bonus miles. The key, he says, is not just collecting information—but learning how to act on it.

IBM’s Advice to Mid-Sized Company CEOs

Ed Abrams, IBM’s vice president for the midmarket and the company’s top official concentrating on small and medium-sized companies, offers these tips on how to proceed with analytical projects.

“Your first step is identifying areas of future growth. Where is the opportunity going to come from? Then, work with IBM and business partners and your Information Technology staff to unlock the insights that are in the information you are already collecting. Chances are, you have a ton of information and you don’t know you’re collecting it, say, from mobile systems being used by the sales force.

“Working with our partners is the way most CEOs in this segment want to work. They want to work with someone local, who knows them and knows their business. They don’t want to get lost in a big company. Our resellers have real expertise in that regard.

“How do you put it together in a way that makes sense for you? The smart answer is to start small and grow. Take pieces of the information you have at your disposal to start this process. What often makes these analytic projects difficult is that a CEO will try to solve for everything at once. Our advice is, start with an aspect of business. Build, learn, develop and identify the things that work for you; then build off of that. Don’t go in and say, “I’m going to build a giant data analytic group within my organization.” Start with simple dashboards that give you insight into what you’ve already got.”

Desert Mountain Uses Data Analytics to Keep Golf Courses Green

The Desert Mountain golf club has a unique set of challenges. As the largest private club in the U.S., it boasts six 18-hole courses designed by Jack Nicklaus and needs 1.2 million gallons of water each night to keep the courses in tip-top condition. But the club is located on 8,000 acres in Scottsdale, Arizona, where water and the electricity needed to distribute it are in scarce supply and therefore very expensive.

Desert Mountain, with $63 million in annual sales, turned to technology to help address the challenge. It bought IBM Intelligent Operations Center software for $150,000 that allows it to analyze data from individual sensors located on all six courses about levels of humidity in the soil. The system is still being rolled out, but each course has as many as 1,000 sensors because different species of grass and other plants require different levels of moisture to survive and thrive. On 10-minute intervals, the sensors transmit their findings wirelessly to the club’s operations center, and employees also transmit findings from hand-held devices as they survey the courses and its lakes.

The system crunches both the structured and unstructured data—such as weather forecasts—and helps managers make decisions about which individual sprinkler heads on which courses need how much water each night. The water is recycled waste-water purchased from the city of Scottsdale. It is dispersed at night because that is when electricity is cheapest. Altogether, Chief Operating Officer Robert Jones estimates that the total cost of equipping each course with this system is about $250,000, including elements like the advanced sprinkler systems, sensors and wireless transmission gear.

But over the course of a mere three years, Jones reckons that the investment will pay off. “It represents millions and millions of dollars if we can control the amount of water we use,” he says. Because the club uses more than 1 billion gallons each year, even a 10 percent improvement in water usage will make the investment pay off in that three-year time frame. “Our water doesn’t go on unless our system tells us that the ground has reached a certain temperature of dryness stage and the turf needs it,” Jones explains. “Then it tells us how much water we need to buy.”

Desert Mountain relied on a local computer systems integrator, Element Blue, to put all the elements of the state-of-the-art system together. Like many small and medium-sized companies, it did not deal with IBM directly. But IBM’s analytics was the right solution because it could accommodate the data from other companies’ sensors and devices and translate it into one common “language.” Jones is confident that the net result is the most efficient use of water. “What used to be somebody’s guess or a Scientific Wild-Assed Guess (SWAG) has now turned into a very sophisticated decision based on what’s the best way to get the water out there,” he says. “It’s new technology managing old stuff.”


  • Small and medium-sized companies can buy sophisticated analytic tools for very little money
  • SMEs often can absorb and exploit these technologies faster and better than larger organizations
  • The rate of ROI in these technologies is high when implemented correctly
" William J. Holstein : William J. Holstein is a journalist, consultant and speaker. He is the author of, "The Next American Economy: Blueprint For A Sustainable Recovery." For more of his work, visit"