Matrixx marketing has a "mainbank" facility in Ogden, Utah. It is a single room sliced into 300 calling stations. Unlike agents in "dedicated" call centers for clients like Sony and DirecTV, mainbank reps handle calls for all kinds of products, particularly those advertised on TV. The work is grueling, but you get to chat with the most interesting people. Listen to Matrixx agents swap stories about notable calls:
"I had a friend who had Madonna call," says Karen Rich. "She just pushed back from the desk and screamed. It was for a donation. I don't know if Madonna gave her last name or not."
"I had friends of Charles Manson call," says Linda Thornock. "They wanted to order him an air purifier. They said he didn't like the smell of the air in prison."
Matrixx fields roughly 4 million calls a week. A few generate funny stories. Some present thorny problems. Each registers a useful nugget of data. Taken together, they become something different -- something powerful. These calls change how companies do business. They help anticipate problems, improve products, revise strategies.
That's because Matrixx doesn't just talk. It learns. It collates data from millions of conversations and thinks about what the data mean. A food company that's test-marketing products to help people manage chronic diseases gets a daily sales report from Matrixx. The baby-formula company has online access to 40 categories of call information, updated every 10 minutes. The information is considered so valuable that the company's laboratory links directly to the Matrixx database.
One of Matrixx's oldest clients is Guthy-Renker, the country's largest infomercial producer. The company sells dozens of products -- from Tony Robbins's motivational tapes to exercise equipment -- in dozens of markets. Calls pour into Matrixx at the rate of 40,000 a week, mostly in "spikes" of 1,000 to 2,000 during the first minutes after an infomercial finishes.
The agents answer questions and take orders. Matrixx also generates a daily report to analyze what's selling where. "It comes early in the morning," says Charlotte Spielberger, Guthy-Renker's director of marketing services. "I look at it right away." She uses the data to tweak sentences in particular infomercials, compare different wordings, and test special offers.
But even in-depth data communicate only so much. Most Matrixx reports end with a fistful of pages filled with customer anecdotes, comments, emails, and notes of complaint or compliment.
Matrixx frequently arranges for client executives to sit down with small groups of agents. These aren't formal presentations -- just a chance for the people who wear the headsets to tell the people who wear the suits what the people who buy the products are saying.
"This is one of our most valuable services," says Terry Belush, director of operations for the company's health care and consumer products call centers. "We can run reports and provide quantitative analysis, but the way something is said -- what a customer emphasizes -- can be more important." The sessions also take the politics out of communicating bad news. Matrixx reps don't sugarcoat the truth. After all, they don't work for the people in the suits.
"We are a more objective vehicle for acquiring information," says CEO David Dougherty. "We don't have to worry about the political consequences of finding out that a product's bad."
Learning from customers means more than just reacting to their comments. By dealing with similar issues thousands of times, Matrixx develops the capacity to anticipate problems. Its computers are filled with data about the people who call in. The DirecTV database records in which rooms customers have their television sets. The baby-formula database lists the ages of customers' children, whether mothers are using formula as the principal form of feeding, and how parents nursed each of their previous children.
"This is better than research data," says Christine Wright, Matrixx's vice president for research services. "This is behavioral data."
The power of behavioral data is that it can affect behavior. Wright's group is working with a food company to predict which customers are likely to place big orders. For a regional telephone company, it's developing algorithms to predict which businesses are likely to buy what services. "The amount of wireless service a business does is highly related to the number of sales employees it has," Wright says. "We can develop algorithms that predict customers' abilities and propensities to purchase."
But Matrixx understands that even the most powerful database is built around data -- bits of information that trace their origins to individual phone calls. The most powerful computers in the world are no substitute for the power of an intelligent conversation with Lovemoney, Cathy from Simi Valley, or Connie Selleca. So Matrixx keeps answering the phone, and its call centers keep filling with the soft, ceaseless murmuring of the most important sound in business:
"Hello, this is Jennifer. How may I help you?"
"Your first name, please? Gerald? You'd better spell it."
"I absolutely agree ma'am, that shouldn't have happened."
"You'll have that in 48 hours. Have a good day now."