1. Use online research for widely used products. Use and reactions are likely to be more homogeneous. So, results are less likely to be confounded by differences between the online and offline segments of the population that NFO described in the Conference's opening talk. Don't use online research if you need to estimate incidence. Anything that requires precise demographic projections is better done offline because NFO showed attempts to correct for differences by balancing and weighting only helped somewhat.

2. Controversial questions showed fewer extreme answers when asked online. They were about gay and Lesbian rights, government help for Blacks and, abortion. The answers were described by Mike Hess, a recent addition to the brain trust at Knowledge Networks (last month's InterSurvey) who bought out his Cincinnati research firm. Knowledge Networks is the darling of the Silicon Valley Venture Capitalists started recently by two poly sci profs at Stanford, explaining the use of test questions most wouldn't touch with a ten foot pole. They are building an online panel to match total US demographics by giving panelists a WebTV to answer their questions. The comparison was with an RDD phone survey.

3. Online advertising boosts impulse sales for P&G. But it didn't have any impact on planned purchases. These were the main findings from a P&G study by IRI using 3000 online households drawn from a preexisting panel who reported purchases by scanning UPC codes with handheld devices and had their online activity automatically monitored -- in essence a single source system online. The food and beverage brand used to represent unplanned purchases showed an immediate and sustained increase from exposure to the online advertising. This brand also had its sales boosted by traditional TV advertising. Two brands from personal hygiene and household cleaning categories were classified as planned purchases, and their sales did not increase.

4. Physicians react less favorably to ad claims when interviewed online. Laurie Gelb of Cozint Interactive said it is in line with other evidence that results from online surveys tend to be more skeptical and negative. Physicians were also more likely to mention the leading brand (Lipitor) when asked to list drugs for treating high cholesterol. She said this is counter to the expectation that more socially acceptable answers will be given when a personal interviewer is involved. But since it is the brand most doctors prescribe, it also appears the web survey gave the most accurate answer. She also found attribute ratings correlated more closely with overall satisfaction, showing the web survey provided more internal consistency. This evidence of more thoughtful answers online was relevant because she said physicians are groups that already have virtually universal access to the Internet. But she mentioned response rates to both the phone and web versions of her survey were in the low single digits, causing many to feel cautious about the results.

5. AT&T finds objections to online focus groups are based on myths. Playboy feels they are real. In a direct comparison AT&T found conventional focus groups and online focus groups preferred the same ads. The online versions showed good group interaction and plenty of emotion. The problems Playboy cited as being all too real were that there were technical glitches when panelists had to click over to other sites to get a look at them and give their reactions. They didn't say as much when they had to type it all out, and Playboy found the online focus groups didn't prove to be any less expensive.

6. Opinions split on the value of "Do-it-yourself" surveys. Mark Hardy of Greenfield characterized it as advanced water cooler research. Tom DuBois and Daniel Greenberg of Active Research, a supplier of such software, felt for the experienced researcher do-it-yourself surveys are a useful tool for squeezing time out of the research process. They pointed out two ways of doing it: using an Application Service Provider's (ASP) software and fielding the survey on their web site, or buying your own software and having respondents visit a web site you set up. In either case they advised examining the questionnaire editor closely. That's the part of the program you write the survey with. The speakers didn't review or recommend specific software, and they didn't even broach the critical subject of how you get e-mail addresses of potential respondents. But they did caution inexperienced researchers could easily get in over their head. Hardy reviewed many of the same pros and cons but came down on the side of skepticism and disdain, pointing out the many pitfalls that can keep "do-it-yourself" research from being "true research". Still, he granted there is a role for "water cooler research", particularly when it is done among customers where e-mail addresses are easier to get.  

7. Those irrepressible Finns seem ready to pioneer wireless research. They are already enticing people with banner ads to play promotional games that lead to a survey. An international soft drink company generated a 4.3% click through rate. Then 51% played the game and fully 97% of those went on to register for a contest by responding to a questionnaire. 30,000 have now played using an MTV3 web site in Finland. Joseph Sakach of MSA and Matti Hamalainen of CODEONLINE described how the same concept can be implemented with current wireless phones that have Internet connectivity or data capability. Wireless phone penetration has already reached almost 70% in Finland. That's triple what it is in the US. With further advances in wireless technology a vast array of possibilities opens up. Cellular devices carry their user's identity, so that could eliminate the need to have respondents tell you about themselves. They are also "location aware" so it would be possible to track down where they are physically. And particularly in Europe, people carry them constantly so consumers are permanently and continuously reachable. All of these provide opportunities researchers have never had before. 

8. Magazine covers picked correctly and inexpensively online. Walking Magazine asked every 5th person visiting their website to take the survey. Respondents were asked to pick between several pairs of covers. They tested the alternative covers by actually putting them on magazines on news racks. The online survey correctly picked the one that generated the most sales. Interestingly, this was the second type of research they tried. First, they tried multivariate analysis of every difference they could see in years of past covers to see if they could find what led to success on the newsstand. Their results were similar to the results many of us have achieved trying to develop a magic formula for great advertising -- dismal failure! Give us a specific example of an ad, commercial -- or cover -- and researchers can do a lot better telling you if it is going to work. 

9. Prize drawing incentive didn't increase response rate, it increased reported readership! The syndicated IntelliQuest Computer Industry Media Study uses a 16-page mail questionnaire. With the ever-enduring Val Appel as consultant they experimented with an offer to enter those who replied in a drawing for several thousand dollars of prizes. This was in addition to a five-dollar bill enclosed with the questionnaire. 62% filled out and returned the questionnaire. In a matched group where no drawing was offered, 63% replied. In the survey with the drawing, respondents reported reading 20% more magazines and 92 out of 100 magazines studied showed gains of at least one percent in readership. They do not plan to continue the drawing.

10. Magazines are history, unless they go digital. This rather jarring announcement came from Robert Desatnick of Qmags.com. He reviewed the dismal financials of the industry, and how their cost problems would disappear if they became online publications. He also pictured the many benefits of researching magazines when you know every page every reader ever clicked on. 

10.5 Many old issues still face media planners working with print. Subjects that have been coming up in print research conferences for decades were also dealt with. The cumulative reach of a series of magazine ads was still reported to vary as much as 60% depending on how you use through-the-book, recent reading, diaries or face to face interviewing to make the estimates. Do print optimizers really optimize? The work of Zielske in the 50's and even Ebbinghaus in the 1900's were still being vigorously discussed. Psychographic segmentation was still being advocated over demographic segmentation.   

(And we don't want to miss items everybody was talking about. When was the last time you had filet mignon at a conference lunch? Congrats ARF! And, how about those expensive satchels Knowledge Network provided for every attendee -- to try and make up for their dumb name change during the month -- just when the world was starting to learn who they are. Truly a guilt trip that benefited everyone!)  

Don Bruzzone October 16, 2000