Study Suggests Google Harms Consumers by Skewing Search Results
Study Suggests Google Harms Consumers by Skewing Search Results
Yelp-sponsored research examines Google’s practice of promoting its own search services
By Tom Fairless
June 29, 2015 4:18 a.m. ET
BRUSSELS—New research by two U.S. academics suggests that Google Inc. is harming Internet users and violating competition laws by skewing search results to favor its own services, a potentially significant twist in Europe’s long-running antitrust investigation of the U.S. search company.
The research combines statistical testing with detailed legal and economic analysis to examine the ramifications of Google’s practice of promoting its own specialized search services, such as for local restaurants or doctors, at the expense of rivals such as Yelp and TripAdvisor.
It was sponsored by Yelp, which has filed a complaint with European Union antitrust authorities over Google’s search practices. It was presented to EU regulators on Friday.
The study’s authors— Michael Luca of Harvard Business School and Tim Wu of Columbia Law School—found that users were 45% more likely to click on results that were ranked purely by relevance, rather than as Google ranks them now, with its own services displayed prominently.
“This suggests that by leveraging dominance in search to promote its internal content, Google is reducing social welfare—leaving consumers with lower quality results and worse matches,” the authors wrote.
The results, they went on, provide “empirical evidence” that Google’s search practices have harmed consumers in some cases and as such “cannot be described as pro-competitive.”
“The demonstration of consumer harm is, we think, an important conclusion…that should influence any competition law analysis,” the study says.
Mr. Wu is one of the most prominent academics in the field of competition law and technology. A former adviser to the U.S. Federal Trade Commission, he is known for coining the phrase “net neutrality,” the principle that Internet service providers should enable access to all content equally, without favoring or blocking particular products or websites.
One official at a European antitrust authority said any study that showed Google caused “quantifiable harm” to consumers would “certainly bring things forward” for EU regulators.
Those regulators “will be delighted to have as much evidence as they can,” the official said.
EU antitrust chief Margrethe Vestager formally charged Google in April with skewing results to favor its comparison-shopping service, escalating a five-year investigation. At issue is whether Google uses its 90% share of online searches in Europe to squeeze competitors in related markets where it also competes.
The charges could lead to billions of euros in fines and requirements for Google to change its business practices. Ms. Vestager said she continues to examine other domains, such as travel and local services.
Google declined to comment on the new study. The company has repeatedly denied breaking EU antitrust rules, and has said it strongly disagreed with the need to issue formal charges.
On Monday Google said it had requested, and been granted, more time to respond to the EU’s charges, in order to review documents related to the case. The new response deadline is Aug. 17.
U.S. regulators closed their own investigation into Google’s search practices two years ago after the company agreed to voluntary changes.
Google has long argued that it serves users by prioritizing its own specialized search services for areas such as maps and travel, because it thereby answers users’ queries more precisely.
The academics agreed that might be true in some instances, such as displaying time or solving arithmetic problems.
But in other instances, they argued that Google was making its overall product worse for users to provide favorable treatment to Google content.
The authors focused on searches for local services such as restaurants or hotels, the largest single category of search requests. They randomly displayed one of two sets of search-result screenshots to more than 2,500 Internet users. One set of users saw a page reflecting results currently displayed by Google, while the other set saw a page that ranked third-party review sites such as Yelp and TripAdvisor based on their relevance—using Google’s own algorithm.
The survey found that 32% of users would click on Google’s current local results, while 47% clicked on the alternative merit-based results. That nearly 50% increase in the click rate is “immense in the modern Web industry,” the authors wrote.
“Stated simply, when it comes to local search, Google is presenting its users with a degraded version of its search engine,” the authors wrote.
The experiment was carried out on an online platform, UsabilityHub, that is used for testing website designs before a website is introduced to the market.
The authors conceded that there were differences between their click surveys and actual search results, and that they didn’t have access to internal Google data to verify their results. They nevertheless cross-checked the results using data from Yelp. The authors said the Yelp data indicated they provided a reasonable estimate of actual user behavior.
In one example, users were asked where they would click first on a screen showing results for “coffee Louisville ky.”
The authors argued that Google’s behavior could harm consumers because they might not find what they were looking for in Google’s own set of results; they took longer to find the information; or they ended up patronizing a business that wouldn’t be their first choice.
Local results intrinsically have a lower number of references from other sites or other measures of relevancy because they are of interest only to a small portion of the web.
Choosing to rate Yelp or other company's results high or low can only be done on the overall repute of that sources results as a whole, not in any particular instance.
In the Yelp case, one might give their results low relevance because of the acknowledged packing of results by shills or the alleged manipulation of results by Yelp to extort payment for better placement.
There will never be a perfect web search oracle. Consumers will still have to exercise judgement.
'Caveat emptor!' includes looking past the first three results.