Advertising disclosures: Measuring labeling alternatives in internet search engines

In an online experiment, we measure users' interactions with search engines, both in standard configurations and in modified versions with clearer labels identifying search engine advertisements. In particular, for a random subset of users, we change " Sponsored links " or " Ads " labels to instead read " Paid Advertisements. " Relative to users receiving the " Sponsored link " or " Ad " labels, users receiving the " Paid Advertisement " label click 25% and 27% fewer advertisements, respectively. Users seeing " Paid Advertisement " labels also correctly report that they click fewer advertisements, controlling for the number of advertisements they actually click. Results are most pronounced for commercial searches, and for vulnerable users with low education and little online experience. 1 Introduction Search engines combine two kinds of links: So-called " algorithmic " links (also known as " organic " links) present the results a search engine deems most relevant for a given search phrase, selected based on page contents, keywords, links, and other factors. So-called " sponsored " links give the results a search engine is paid to show, selected based on an advertiser's payment along with an assessment of the match between the advertisement and the user's search. (Edelman, 2007) Search advertisements have become a huge business-the mainstay of Google's $25+ billion of annual revenue. Many search advertisements are helpful-promoting products directly responsive to users' requests. Other search advertisements are less helpful; some promote counterfeits (Thompson, 2010), others trick users into installing spyware/adware (Edelman, 2006a), and still others resort to alternative methods of deception (Edelman, 2006b). Pursuant to longstanding legal principles and recent FTC instructions, all leading search engines now include a label near advertisements. However, no search engine uses the " Paid Advertisement " label that legal precedent has required in other media. Seeing a divergence between industry practice and applicable legal precedent, we seek to measure the effects of labeling advertisements in the way that prior cases specify. 3 Our motivations are several. For one, with more than two billion people now using the Internet, search engine advertisements are a strikingly important and sizeable market. In this context, even a small change in user understanding could result in large changes in aggregate browsing behavior. That said, other compulsory disclosures provide cause for concern; consider the ever-present " objects in mirror are closer than they appear "-often a subject of mockery. Would improved labels …