By Charlotte Slaiman and Elise Phillips
October 9, 2023
Each week, Public Knowledge competition policy experts watch the Department of Justice’s landmark U.S. v. Google case live from the E. Barrett Prettyman United States Courthouse. Due to very strict confidentiality rules from the court, you can’t watch or listen to the trial unless you are there in the building and following the court’s rules prohibiting using electronic devices. Join Vice President Charlotte Slaiman and Policy Counsel Elise Phillips for their recaps highlighting the main discussions as the trial continues.
This week, Microsoft CEO Satya Nadella grabbed headlines as the latest witness in U.S. v. Google. If Google planned to argue that ChatGPT’s introduction into the search market creates substantial competition for Google, Nadella was not the witness to validate that story. Yes, folks may remember his quote from that interview, “I hope with our innovation they [Google] will definitely want to come out and show that they can dance. I want people to know that we made them dance.” But under oath, Nadella explained that this was the “early exuberance” of “a 3% competitor excited he might become a 3.5% competitor.” Fundamentally, generative AI will mean big changes to the search market, Nadella said, but not in ways that change the incumbency advantage that Google has locked in with its exclusive default contracts. Any potential business customer that might be interested to set Bing as their default search engine because of its cool new feature would still have its Google exclusive contract to contend with.
In fact, Nadella revealed another really concerning trend. Already he’s seeing ways in which large language models (LLMs), like ChatGPT, can be subject to a powerful anticompetitive moat that prevents new entrants from being able to compete. These models depend on large amounts of content as an input. Nadella told the court that publishers, the copyright holders of large amounts of content, are already granting exclusive licenses to Google as the only ones who are allowed to use their content to build generative AI models.
The court also heard this week from Sridhar Ramaswamy, a former Google engineer and executive who went on to found a competing search engine called Neeva. Neeva had attempted to compete with Google, but eventually failed and had to shut down their search product. Judge Mehta asked a good question, “Why did you think you could compete with Google?” Ramaswamy explained, he thought he had a strategy to get around the default contracts, instead of making his revenue from advertising, Neeva was a paid subscription service. Unfortunately, even this unique strategy was not enough to overcome the default contracts: they were not able to become profitable quickly enough, and when interest rates changed, they couldn’t continue to operate at a loss.
The government questioned Joshua Lowcock, the Global Chief Media Officer of Universal McCann (UM), a media agency company. Lowcock has been previously critical of Google’s advertising practices, notably calling for Google to refund advertisers for allegedly charging companies for video ads that were skipped, a violation of Google’s internal ad policies. No surprises here, the focus of Lowcock’s testimony was Google’s dominance in search advertising.
Lowcock outlined the primary objectives for his clients in online search: advertisers may utilize different types of campaigns to attract consumers, whether it be search, display, shopping, video, or discovery advertising. These campaigns are designed to target consumers at different points up and down the purchase funnel — a marketing theory that describes various consumer states of mind leading up to the purchase of a product or service. Lowcock accurately described how search ads occupy a very specific place in the purchase funnel — when consumers are ready to buy. Display or video advertising are not able to reach consumers at that key moment, when they are ready to buy. That’s when consumers turn to search, and can only be reached by search advertising. With Google occupying almost 90% of the search advertising market, UM’s recommendations to clients are subject to Google’s black box auction pricing and sudden changes to their auction process.
The states’ questioning focused on the type of tools that UM advises their clients to use when bidding on search terms or generating market analytics. Here, Lowcock’s statements were particularly compelling, as Google’s dominance in the search advertising industry impacts the Search Engine Marketing (SEM) tools that companies use to bid on key search terms and generate analytics of consumer and marketing trends. Clients are drawn to SEM tools like Marin, Kenshoo (recently rebranded as Skai), and Google’s own Search Ads 360, which is well-attuned to Google’s bidding framework. While Google is allowed to optimize its own products to integrate with other services within its own ecosystem, Lowcock pointed out that companies are not inclined to use more than one SEM. This commentary added an interesting dimension to the scope of Google’s dominance in the search advertising market, as it showed another dimension of search ad market that benefits the dominant players to the detriment of consumers.
Google’s strategy in light of Lowcock’s testimony was two-fold: First, they endeavored to deconstruct and delegitimize the framework of the purchase funnel. Second, they aimed to prove that Google’s search advertising competes with “specialized search engines” such as Amazon or Walmart, and social media ads on Instagram or TikTok. However, the concept of the purchase funnel is widely shared across the marketing world, so the government should be able to point to additional sources to confirm it. And Lowcock was clear that advertisers view general search engines like Google and Bing as playing a very different role in their marketing strategies from other sites like Amazon and Walmart.
