How to fix Google search

Google pays the cost to be the boss.

The company properly known as Alphabet has two monopolies, a US Judge ruled last month: one in general internet search, and text search advertising.

There are two main reasons it was able to obtain and maintain this enviable position, both of which are that it has the best product.

At the heart of Google Search — by far the dominant force within general search engines, with roughly 10 times the traffic of its nearest rival, Microsoft’s Bing — is a process called retrieval and ranking.

Google runs programs known as crawlers that index much of the internet into an immense database, filters relevant results based on search terms, determines which are worth giving a relevance score to, ranks them, and presents them to the user in rank order. Better ranking = better results = better search business.

In the early years of the company, Google was much better at doing this than anyone else. So it became a default, a verb, and an internet giant.

Dominating search has a number of subsequent benefits for Google:
— it can gather more information on how people interact with search results, helping it generate better rankings.
— it can charge more for advertising in and around those search results.
— it can better afford to continue indexing as the internet grows.

For a long time, this was arguably a good thing: Google got better as more people used it, launched other useful and notionally free products to support its key search and advertising function, and maintained the capacity to keep up as the size of the public web exploded through the ’00s and ’10s.

Google’s position has made it the alpha[bet] and omega[bet] of web search. Not only is it the first port of call for nearly all generalised searching and default choice for advertisers, but it’s also become the shining light towards which much of the internet bends. Things are made to be Googled.

In search, first you get the content, then you get the power, then you get the users, then you really get the power, then you get the women.

Throughout the recent Department of Justice case, which led to Judge Amit Mehta’s ruling, Google insisted that the quality of its search mattered above all: people simply use it because nothing else is better.

This is right and wrong. Yes, Google’s product is widely considered to be the best. But underpinning its search dominance is its position as the default search engine across a variety of contexts: most notably in Apple’s Safari browser, the default internet portal on the iPhone; and on devices running Android, the mobile operating system Alphabet bought in 2005, via tight agreements Google has with device makers. About half of all general search queries in the United States are through an “access point” captured by one of these contracts, and so is half the search revenue. Chart via Bernstein Research:

Apple and other partners do not do this out of the kindness of their hearts. Testimony during the trial revealed Google paid $26.3bn in “traffic acquisition costs” — a share of revenue paid to partners — during 2021. Of this, about $20bn went to Apple. In addition, Google held back from advertising Chrome, its own browser, to Apple users. “The true value of the defaults is undoubtedly far greater,” Mehta wrote in his judgments.

As well as representing a cornerstone of their services income, this agreement benefits Apple in other ways. With no general search engine of its own, Apple has claimed that offering Google means offering its users the best.

“I think one of the benefits, for example, that Google gets from Apple is that we are telling the world that Google is the best search engine, because that’s what they would expect Apple to pick,” Eddy Cue, Apple’s senior vice-president of services, told the trial last year.

Microsoft would beg to disagree. In fact, it would pay handsomely to get the chance. Satya Nadella, its chief executive, told the trial that making Bing the default search on Apple devices would be a “game changer”, and claimed he would be willing to give up all the associated search revenue — practically hand Bing to Apple as a white-label product — in order to grab the data that would break out it of a losing position in a winner-takes-more market.

“[T]his vicious cycle that [Microsoft is] trapped in can … become even more vicious because the defaults get reinforced,” said Nadella.

Of course, Microsoft (itself no stranger to monopoly allegations) doesn’t deserve much sympathy — it has been thoroughly beaten. Over almost three decades, its search products (as well as the browsers and mobile devices that help convey them) have been far less successful than its rival’s. The fate of Bing is about Internet Explorer’s complacent decline, the fundamental badness of Windows phones and the torments of predecessor MSN Search as much as Bing itself.

But it’s difficult to argue that Google isn’t now in a monopolistic position: testimony from the trail revealed, crucially, that Google found it was able to change its advertising prices without concern for the wider industry, a clear sign that it lacks genuine competition.

Attempts have been made to fix this dynamic. One way — in use in Europe — is a so-called choice screen, in which users are presented with a range of search engine options.

It’s cute, but it doesn’t work. The evidence shows they may even entrench market leadership further: 97 per cent of European users pick Google when offered a range of choices. There’s a decent argument — despite a perceived decline in Google’s quality — that they are making the right choice. And an important finding of the case is that Google has not rested on its laurels: it has continued to innovate from a position of dominance.

Conditions would suggest that general search is a broken industry. Microsoft’s apparent desperation to capture Apple’s search defaults is telling: the inexorable logic of ranking and crawling means money and data beget one other. Bing and Google could exist in perfect balance, but the natural state is one of single-company dominance, and then potential enshittification. It’s also possible that the battle is truly over, and no shift in the ranking balances could now unsettle Google’s publicly perceived superiority.

So can this situation be fixed? Definitely maybe, says Bernstein’s Mark Shmulik and team:

While we consider ourselves somewhat savvy on Google, Search, and the Digital ad market, we’ll admit that we were stuck scratching our heads trying to come up with remedies that stood a chance of addressing the alleged bad behavior… Armed with incremental data points, specifics in the ruling, and the judge’s color commentary, we update our remedy trees.

Here is a, uh, “remedy tree”:

Bernstein’s focus at this stage is on what the remedies might be rather than their impact, but it’s a starting point.

The most eye-catching approach — one the DOJ is already considering, according to Bloomberg — is a forced asset divestment, ie breaking Alphabet up.

Bernstein reckons this is an unlikely outcome — but how could this work? If the goal is to break off a piece of Google, three main targets jump out, they say:

1. SA360/Google Ads (AdWords portion): Instead of opening the search ad delivery software, the Court could compel Google to spin off SA360 and/or the AdWords portion of Google Ads. It’s unclear if such a separation is even technologically possible, but the idea would be to create a demand-side platform separate from Google that can buy keywords across various search publishers — a Trade Desk for Search. Presumably Google would still be able to sell ads directly, though an independent third-party may, in theory, be able to keep ad prices from inflating while creating ad demand for smaller search engines.

2. Android: Trying to work through the various second-order effects of a stand-alone Android hurt our brains. We don’t see the emergence of a third mobile operating system, thus the ecosystem of players remains the same. Now Android needs to monetize, which means they’ll likely charge the device manufacturers – meaning either the OEMs or consumers are now worse off. Presumably, Google Play would have to go with Android to monetize apps on the OS which seems well out of scope.

3. Chrome: We’re not entirely sure that a web browser is a stand-alone business. Google accounts for 80%+ of Mozilla’s revenues with donations the second largest source of funds. Even if you spun off Chrome as a separate entity, it would still need to find a way to generate revenues, Google search remains the gold standard here.

Much more likely is some kind of crackdown on Google’s use of default contracts, either by forcing an adjustment to the language used within them, or ditching them altogether. In such a situation, it stands to reason that Bing would probably simply pay to take Google’s spot:

While the contracts may be ripped up, it’s hard to see how the courts can prevent a browser or device manufacturer from setting a search engine as a default with their own revenue-share terms similar to the Play or App store. Sure, these partners could select a different default partner, which is really just Microsoft/Bing, or go the choice-screen route, but the courts have made it clear that Google is the best consumer search product that monetizes search text ads best. We’ll likely end up in the same place.

While Microsoft has been more or less clear about its ambitions, the situation would appear to be somewhat more fraught for Apple. Its choice could be between either pushing people to use a search engine that it doesn’t currently endorse, or giving up a big chunk of services income. For Google it would hurt a lot, Bernstein notes:

According to Google’s internal estimates, if Google was replaced as default GSE on Apple devices, they could lose between 60-80% of iOS queries, which would translate into net revenue losses between $28.3B to $32.7B. In the worst case scenario for Google, Bing is picked as default across all distribution channels covered by the RSA/ ISA today, 10-15% of Google’s global search revenue could be at risk (40% in the US, 55% covered by these contracts, and Google is only able to crawl back ~40-50% of the queries).

If the idea is to curb Google, a choice screen seems like the worst idea: the evidence suggests choice would entrench the company’s dominance while actually saving it money on revenue sharing.

Another possibility is to force Google to open up its databases via an API. In effect, this would force Google to allow other companies to license its database, shattering the advantage it has been given by economies of scale. This isn’t such an alien arrangement: Bing already has a similar relationship with DuckDuckGo and Yahoo, both of which piggyback on Microsoft’s indexing efforts. Bernstein:

If the goal is to really open the floodgates to competition, such a remedy makes the most sense – imagine what a SearchGPT and Perplexity could build if working off the same quality library that Google has today. Such a remedy also wouldn’t necessarily impede on Google’s IP around understanding, ranking, and delivering query results.

There are two big ironies here.

The first is that the playbook for reining in Google would likely to be the same regulators used against Microsoft at the turn of the century. Then, Microsoft was seen to be using its operating system dominance to shanghai users into its other products. Despite initial talk of break-ups, the company eventually reached an consent decree deal with the DOJ that allowed more third-party products into Windows. The fate of Google may be similar, with enforced transparency a likely focus.

The second is that this major invention — which follows Google being ruled an illegal monopolist on Google Play, the default Android app store, and ahead of another DOJ challenge to its ad business — comes just as the search industry faces its most meaningful shift, through generative AI.

GenAI chatbots, which rose to prominence after the DOJ began its case, were not a major focus of Judge Mehta’s attention. But if AI does become a widespread alternative to search, the DOJ risks fighting yesterday’s battle. Shmulik et al:

With the benefit of hindsight, rules around Google’s ability to compete in a changing search universe could have the greatest impact of any remedy. Rules here feel increasingly likely following the DOJ’s comments around forward-looking remedies. Google has won search, but the antitrust law exists to make sure that there is a fighting chance for new disruptive technologies to emerge, which is AI today.