Data Mining vs. Privacy

Trolling for new business leads has been the bane of insurance agents for decades. Electronic data mining has eased the pain, but there are many gray areas as the call for consumer privacy grows louder.

It was a bet with a Dun & Bradstreet sales agent several years ago that got Richard Beedle thinking that there had to be a better way to dig up new business.

He had wagered that the agent couldn’t pull together a pure list of retail-category SIC codes with 80 percent virginity. The agent bet he could not only do it, he could come up with 85 to 90 percent virginity. “I told him, if you can hit that, I’ll buy the list,” Beedle said, recalling the conversation. The Dun & Bradstreet agent returned some time later with a shoebox filled with three-by-five note cards, with the names of at least 3,000 businesses.

“It was filled with liquor stores, Bible book stores, lots of businesses that just should not have been there,” Beedle said. “I won the bet.” But he wasn’t winning in his quest to market insurance products to a well-known national chain of gift/card shops. He needed to find a data-mining panacea.

Many businesses, not just insurance agents, are faced with the same dilemma—getting better, more complete customer information faster in an understandable format. They realize that not only does knowing more about customers provide them a better chance of courting or persuading them to purchase or use their products, but that new technology is instrumental in their quest.

The demand has resulted in numerous companies popping up all over the country, providing businesses and individuals with, in many cases, as much information as they can afford. And as that information becomes more readily available and more detailed, questions are being raised and lines are being drawn over what information is public, private or proprietary.

Click of a mouse
Not surprisingly, many of the questions surrounding the selling and gathering of customer information stems from the use of the Internet. This relatively new marketing tool has the ability to glean information from millions of users with virtually no overhead costs. And, as yet, little regulation.

A recent Georgetown University study directed by Mary J. Culnan, Ph.D., surveyed a random sample of 361 commercial U.S. web sites included in the top 7,500 sites ranked by unduplicated traffic. The premise was to determine how many web sites were collecting information about visitors, how many were providing visitors with a choice about the gleaning of this information through the posting of a privacy policy, and to what extent the privacy disclosures reflected fair information practices.

Of those sites surveyed, 92.8 percent collected personal identifying data; 56.8 percent collected demographic data; 6.6 percent collected no personal data; and less than 1 percent collected only demographic data.

The survey also found that of the 236 web sites that collected personal information and posted a privacy disclosure, 89.8 percent included at least one survey item for notice, 61.9 percent contained at least one survey item for choice, 40.3 percent contained at least one survey item for access, 45.8 percent had at least one item for security and 48.7 percent had at least one item for contact information.

Only 13.6 percent of those same web sites contained an item in all five elements deemed as fair information practices: notice, choice, access, security and contact information.

Ripe for the pickin’
The possibility for abuse by these web sites has attracted legislative attention, with at least one electronic privacy act having been introduced in both the House and Senate. U.S. Rep. Edward J. Markey, D-Mass., introduced H.R. 3321 last year in an effort to ensure consumers are aware that personal information is being collected.

The legislation calls for conspicuous online notice if the recipient of that information intends to share, reuse or sell it. Also, the legislation provides for consumers’ control over the amount of personal information that is collected about them.

Colin Crowell, spokesperson for Rep. Markey, said the bill has yet to be filed with committee, so it has not received any formal opposition, however, “the online industry still opposes any Internet privacy legislation, preferring instead to rely upon industry self-regulation.”

President Clinton has also spoken out on the issue, saying consumers should not have to worry about their personal information being accessed without their knowledge.

At a recent commencement address at Eastern Michigan University, he told graduates “In this information age we can’t let new opportunities erode old, fundamental rights. We can’t let breakthroughs in technology break down the walls of privacy.”

The source of most of the concern is companies that track consumers online without their knowledge. Everything from consumer names and addresses to income, race, religion and favorite pastimes is compiled and sold so companies can more successfully market their products.

Other companies use the information to more closely scrutinize which customers are high-value and which could cost them money.

And the situation will only heat up as banks, insurers and financial service outlets move their businesses onto the Internet. By next year, as much as 60 percent of companies will use the Internet to micro-segment customers, according to Forrester Research

Murky water
The real problem is the potential for companies to snub consumers based on irrelevant information as varied as the types of books the person likes to read to what kind of car he or she drive.

At first glance, it looks like good business. Less desirable customers who are less likely to buy will be automatically eliminated. But there are pitfalls for businesses using such systems.

Essentially, Consumer A could live in the right neighborhood, have the right credit cards and receive the right catalogs, thereby receiving a high profile ranking. But Consumer B, who lives in a less-desirable neighborhood, drives a cheaper car and does no catalog shopping could wind up being more liquid, more fiscally responsible and, in the end, a better risk. But because the company’s consumer profiling software did not give Consumer B a high ranking, they lost a good customer.

Dr. Linda Gerber, a senior marketing lecturer at the University of Texas in Austin, says that’s a risk businesses will end up taking if they segment their markets too narrowly. The ability to more closely predict consumer behavior is, overall, good news for business. But taking consumer profiling too far can have a negative effect for companies.

“What happens to the company that makes a mistake and minimizes a profile on a customer they should have had as a valuable prospect?” she asked rhetorically. “The company is the loser. The customer will simply go elsewhere. The customer and society as the loser almost never happens unless there are ramifications on the way, say, insurance rates are set.”

And when that happens, it’s the age-old concept of redlining, given a computer-age, jargonized twist—Weblining. Gerber, who also heads up the Center for Customer Insight, a cross-disciplinary center in
UT’s business school, said while there is room
for such abuses, most data gathering has remained quite cursory.

“Now, the high-level profiling—that concerns me as far as the abuse potential,” she said.

The data miners
That’s where data brokers like Axciom, which offer such services as InfoBase Ethnicity System, enter the picture. Axciom officials did not return numerous IJ Texas calls, but InfoBase is described on the company web site as a breakdown of ethnic, religious and minority classifications. In seconds, the system can match names with a wide range of demographic data including income, education, religious affiliation and race. The company provides this and other services to a wide range of businesses, including Conseco Insurance Co.

Conseco sees no harm in using the information Axciom provides. Naveen Passey, a Conseco spokesperson, said Axciom compiles the data from public sources, including census data. The company has been using the geographic, demographic and lifestyle data for more than four years to market its products and service existing customers, he said.

“The data is used to identify the types of customers we have so that we can create better products and services that meet our unique needs,” he said. “The information enables Conseco to better understand and predict our customer’s wants and needs. The single view of our customer enables the service providers to answer questions based on a customer profile that includes the necessary product and descriptive data in one place.”

Axciom is not alone in collecting consumer information. Unisys, Naviant Technologies and Hot Data are just a few others dealing in the consumer profiling trade.

Based in Austin, Hot Data provides clients with databases that link directly to their existing customer-tracking software, such as ACT, as well as data cleaning services.

Marketing Director Ellis Oglesby explained that, unlike Axciom, Hot Data does not provide individual consumer data. Rather, the company provides carrier route demographics including psychographics such as the percentage of residents that enjoy outdoor activities like backpacking.

“We can deliver to them in a single day data such as, on average, people here are married, have children, go to Home Depot,” he said. “Those may or may not be traits of individuals you want to insure. The information allows the company to personalize their pitch, etc.”

The company says it is very sensitive to the privacy issues surrounding its line of business. Oglesby explained that Hot Data purchases its information from 15 suppliers, including Experian and Dun & Bradstreet rather than collecting the information surreptitiously as some other companies do.

“Everything we provide is based on information that has been willingly provided by consumers,” Oglesby said. “They’ve put that information out there. We are really committed to consumer privacy and we have some mechanisms in place to make sure we provide that.”

The company will yank access to their system from anyone who is found to be selling the data to someone else. Besides consumer data, the company also provides business profiles and data cleansing software that allow users to update address, zip code and area code information virtually with the click of a button. That’s where agents like Richard Beedle find the services these companies provide so useful.

Beedle had tried everything before stumbling onto Hot Data, including photocopying addresses from Sunday supplements that advertised—without a phone number—his targeted merchant. He’d then type the addresses on index cards, call the stores (after figuring out which of California’s 27 area codes it was listed in), and purify the list himself, the whole while knowing there had to be a better way.

“I’ve been an ACT proponent since their product first came out some 15 years ago,” he said. “I was thinking what if there were a way to broker a conversation between Dun & Bradstreet and ACT…What if I were able to download that information…directly into ACT?”

He followed through on the idea and, subsequently, was a Beta tester for the concept. For about five years he was able to download Dun & Bradstreet information directly into ACT databases. But after a software update about three years ago, the two were no longer compatible, and Beedle was left to once again fend for himself. That’s when he heard about HotData

“Oh my God, an answer to a maiden’s prayers,” was Beedle’s response. “I’m back in business.”