“The utility of search advertising has been a controversial question,” says Navdeep Sahni.| iStock/Nuthawut Somsuk
Revenues from search ads are expected to exceed $300 billion in 2024 — making search the world’s largest advertising channel online or off. The ads are essential to search companies, but their value to users, who collectively make more than 1.2 trillion queries per year on Google alone, has always been something of a mystery.
Some experts argue that search ads are intrusive and even scammy — a distraction users must tolerate in exchange for free access to search engines. Others see the ads as a convenience, enhancing the search experience by offering users additional information and easy access to products and services related to their interests. “The utility of search advertising has been a controversial question and people have written positive and negative points of view on it in the media for a long time,” says Navdeep Sahni, an associate professor of marketing at Stanford Graduate School of Business. “But it is a question that needs to be answered with data.”
Sahni now has that data. Sahni and Charles Zhang, PhD ’22, then a GSB graduate student focused on quantitative marketing, got it from real users and real ads in a large-scale field experiment on a widely used U.S. search engine. While there has been copious research on the efficacy of search advertising for ad buyers, this experiment was unique for its scale and empirical focus on the value of ads to search users.
Collected over a period of five months in 2017, the data reports on queries submitted to the search engine by nearly 3 million unique users. For two months in the middle of the experiment, half of the users saw search results that included the usual number of ads that appear among the top results and in the middle of the page, known as mainline ads. “Whenever there’s a search query,” Sahni explains, “search engines use a proprietary algorithm that scores every ad that could appear with the results. Only those ads whose quality exceeds a certain preset threshold get placed in the mainline positions.” These ads are the most visible on the page and have the most effective positioning.
During the same two-month period, the search engine tweaked its ad-scoring algorithm so that the other half of the user group saw fewer mainline ads with their results. “The experiment increased the threshold cutoff of that algorithm just enough so that 17% of the ads that would have received mainline positions got pushed to less visible positions on the side of the page,” Sahni explains.
Ads Add Value
Analyzing search engine usage before, during, and after the experiment, Sahni and Zhang discovered that the users who saw fewer mainline ads did not seem more satisfied with their search experience. Instead, when the number of mainline ads was reduced, they used the search engine less. This decrease in search engine usage did not kick in immediately, but it was sticky over time.
“We used a revealed-preference approach in which we assumed that people are going to use a product more when they like it and less when they don’t,” Sahni says. “So based on that, we concluded that users liked the search engine less when the ads were removed from their search results than when the ads were present.”
Bolstering this finding, the reduction in usage was especially pronounced among users who used the search engine least frequently and those who used several search engines. “They should be the ones who are most likely to increase their usage in response to positive stimulus,” Sahni says. “Yet they were the ones who reduced their use of the search engine the most when ads are sidelined.”
What explains these findings? One explanation was revealed when the researchers found that mainline ads were featuring new and unique websites that were not being included in the search results — “a clear measure of extra information getting added to the space by ads,” according to Sahni. It may be that such search ads solve a fundamental problem with search engine results: the time it takes for high-quality results to appear in and climb up search rankings.
“It’s hard for search engines to figure out whether a new website is actually good or not given the scale at which they operate,” Sahni says. “So, they rely on an indirect measure: backlinks, which are the number of other websites that are linking to the new site and referring traffic to it. The more backlinks, the better the new website becomes in the eyes of search engine.”
This process takes time, however, and that gives rise to information asymmetry between the search engines and the entrepreneurs behind the new websites. Search ads give the entrepreneurs a means of closing the gap between search results and search users. “Good quality websites will find it profitable to advertise, because when search users go to the sites, they are more likely to complete a transaction and therefore support the search ad spending,” Sahni says.
Overall, Sahni and Zhang’s findings suggest that search ads are not a necessary evil but a useful feature of search results. “We are not saying that all search ads are good, because that is certainly not the case,” Sahni concludes. “With all the usual caveats about relying on one sample, our point is that search ads, on the whole, can be a positive for users, as well as the owners of websites that might be passed over by search engines. That goes against the widely held point of view that ads are a high price that users are forced to pay for access to search.”
For media inquiries, visit the Newsroom.