Six interactive market-research executives gathered in a small hotel conference room near Chicago's O'Hare Airport on July 8. After a couple of hours spent talking about the progress and challenges of the fledgling industry, Rudy Nadilo, CEO of Greenfield Online, said he wanted to ask Gordon Black, the CEO of Harris Interactive, a question.
Nadilo then accused Harris of spamming people to recruit survey respondents. The exchange not only brought the meeting to an abrupt halt, but also triggered a defamation suit that Harris Interactive filed against Greenfield Online in federal court last week.
Though reluctant to comment on the specifics of the lawsuit, both CEOs agreed to talk to The Standard. "I personally have received [unsolicited e-mails from Harris] and know others who have," Nadilo says.
Black counters, "From our standpoint, it is not pleasant to be accused of something that is not true." Black won't comment on Nadilo's e-mail specifically, but says Harris knows "when and where and how every single person in our database came into our database."
Only a couple years ago Nadilo and Black were, more or less, on the same team: They were both working to convince traditional market researchers that the Net was a valid medium for conducting the consumer polls and surveys that are at the heart of much market research.
The latest numbers help to explain what all the fuss is about. Jack Honomichl, publisher of Inside Research, tracks the interactive revenues of 21 major market researchers. In 1997, the market totaled just $9.9 million in Net revenues. Now 1999 revenues are on track to reach $72.4 million. Honomichl says most of the online research has been funded by moving money from existing research techniques.
"It is a very good period right now in the research industry," says Bob Lederer, publisher of the Research Business Report. Many companies cut their research departments in the early 1990s, yet demand stayed the same or increased. Outsourcing projects to major firms has helped spur domestic growth in the $6 billion industry. "The Internet is one of the areas where there is dynamic change," says Lederer.
Information is the currency of the Internet Economy and hardly anyone has more of it than market researchers. Historically, all responses to a survey are supposed to be kept anonymous and confidential; recontacting survey participants in the hopes of selling them anything is absolutely forbidden. But as increasing numbers of traditional researchers embrace the Net, the Harris-Greenfield spat may simply be the first of many closely watched quarrels between experts about the ethics of emerging business models.
Most observers agree that the turning point came sometime early last year. As online penetration approached 25 percent of Americans, researchers began to release statistically reliable studies that suggested it might be possible to conduct surveys online and produce results representative of the U.S. population, with at least the same accuracy as those using offline research techniques.
One such study, Harris Interactive's Harris Poll Online, accurately predicted the results of 21 out of 22 election races in 1998, matching or beating the record of telephone pollsters. George Terhanian, Harris' director of Internet research, is working on perfecting new analysis techniques, such as "propensity scoring," that balance the results of online polls to that of the U.S. population. The practice has not yet become popular in market-research applications, but is common in clinical research. "Some people say there's no right way, there's no science, there's no theory. But we say there is science," says Terhanian. The Harris Poll Online will conduct monthly election polls from now until the November 2000 election.
While many major market researchers have embraced the Web, Harris has few Net allies in the public-opinion polling field. Andy Kohut, director of the Pew Research Center, says entire demographic groups are missed by Net polls. Offering an analogy, he says, "If I went down to K Street in Washington and interviewed people and adjusted the data, it wouldn't make them look like
