In an August 2014 article, Groupon Director of Product Management Gene McKenna assessed the impact of Google’s changing presentation format on paid advertising clicks. McKenna reports that, under Google’s prior format (including a distinctive background color for top-of-page ads), Groupon received approximately 15% more clicks on algorithmic results than on paid results. But after Google removed the background color on top-of-page advertisements, paid clicks jumped sharply — at one point 75% more paid clicks than algorithmic clicks, roughly double the paid clicks Groupon received previously.
McKenna points out that paid advertisements tend to be less relevant than algorithmic results. (My research is in accord. For example, in the third chapter of my Ph.D. dissertation, I found that paid advertisements were nearly twice as likely to fail SiteAdvisor’s safety metrics (testing for adware, viruses, spam, and links to other unsafe sites, among other problems) as were algorithmic results for the same search terms.) By diverting users from high-quality algorithmic results to lower-quality advertisements, Google sends users to less useful destinations.
An equally troubling outcome is that Google’s advertisements send users to the same sites they were already trying to reach. For example, if a user searches for Groupon, McKenna’s research indicates the user is now more likely to click a paid ad to Groupon, whereas in the past the user would often click a no-cost algorithmic link. That’s a transfer of wealth from advertisers to Google. Given Google’s $60+ billion of cash, it’s hard to see how this makes the world a better place.
McKenna finds that the effect of Google’s format change ultimately wears off: Three months after implementation, most users had nearly reverted to their prior behavior of skipping over Google’s prominent top-of-page ads. But the effect remains notable and, to my eye, worrisome. For one, consider the users too busy or naive to notice — users who, to this day, will succumb to less relevant advertising and will continue to drive up advertisers’ costs. Furthermore, the three-month period is harmful in its own right. Google changes page format often. If every change yields a several-month period where users are confused and advertisers’ expenses increase, that’s a large fraction of the time when outcomes are needlessly poor. Rather than celebrate the ultimate diminution of the problem, we should bemoan three months of bad outcomes as three months too long.