June 20, 2005
Mistaken Judgments for Content Labeled Advertising
Last week I applauded the FTC for arguing against the mandatory labeling of commerical emails. In that post, I argued that the labels would increase the rate of erroneous judgments by recipients, because the recipients would mistakenly believe that the advertising was lower-value content than it actually was.
Coincidentally, over the weekend I came across a study by Jansen and Resnick called "Examining Searching Perceptions of and Interactions with Sponsored Results." The researchers were trying to prove that consumers don't like sponsored search results. In the process of doing so, they presented searchers with identical search results, one set characterized as organic results and the other set characterized as sponsored links. As the press release says:
"While study participants rated 52 percent of the organic results as "relevant," searchers described 42 percent of sponsored links as "relevant" even though both sets of results were identical."
As this result demonstrates, 10% of the results were graded irrelevant solely because of the sponsored link label.
From this, I derive a contrarian policy judgment. Consumers don't necessarily benefit by knowing that content is advertising vs. organic. Indeed, such labeling may mislead the consumers--exactly the effect the consumer protectionists are trying to avoid. Consumers think they want to know if something is advertising, but if it leads to sorting mistakes, maybe they are better off not knowing.
Your contrarian policy judgment makes a lot of sense just looking at the data point that you describe. But another finding of our study and others is that consumers have strong negative reactions when they find out that links are paid for unless they are clearly labeled as such. Unlabeled (or vaguely labeled) sponsored links are attributed significantly lower credibility.
So I guess there are two choices. From a purely technical point of view, removing any indication that a link is sponsored may give consumers more choices. Of course, most search engines already return millions of hits so I am not sure how many more we need.
On the other hand, we can listen to what consumers want, which is to know which ones are which. While this may not be a great help in filtering relevant from irrelevant results, it doesn't hurt either. If the content of a sponsored link is truly relevant, it will be listed high on the organic list too.
However, another question is whether we should leave this to the market to decide, or ask the FTC to step in. For engineers (like me), this decision is based on how much sense we attribute to the average consumer. For philosphers, it may depend on where you fall on the libertarian/paternalist scale.
Posted by: Marc Resnick at June 20, 2005 10:57 AM
Actually, I am in agreement with your assessment.
I am doing some research with a major Internet marketing firm aimed at comparing the relevance of organic to sponsored links. Basically, it appears that the relevance may be the same (i.e., they both address the searcher's information need in a satisfactory manner). If this is true, sponsored links have done in 5 years what took algorithmic research to do in 50 years.
And, if true, does the searcher really care how the links are classified? As you point out, it may have a detrimental effect.
I have some survey data that shows searchers really don't care to know (although, some surveys do show otherwise).
Posted by: Jim Jansen at June 20, 2005 01:50 PM