Unredacted Antitrust Complaint Unsealed: Google Internal Documents Show AMP Pages Brought 40% Less Revenue to Publishers

In December 2020, we reported on a new antitrust lawsuit against Google that claimed AMP was created for the purpose of pushing publishers away from “header bidding.” This is an advertising mechanism that allows sites to route their ad inventory through several ad exchanges and sell the space to the highest bidder. At that time it was clear that these were troubling allegations regarding AMP’s performance and how Google may be using it to impede header bidding, but many key parts of the complaint were redacted.

The full text of the newly unredacted complaint, which was unsealed by a federal judge last week, references research from internal Google documents. It states that internal Google communications identified header bidding as an “existential threat.” The complaint alleges that Google throttled non-AMP ads in order to give AMP a “nice comparative boost:”

After crippling AMP’s compatibility with header bidding, Google went to market falsely telling publishers that adopting AMP would enhance page load times. But Google employees knew that AMP only improves the “median of performance” and AMP pages can actually load slower than other publisher speed optimization techniques. In other words, the ostensible benefits of faster load times for a Google-cached AMP version of a webpage were not true for publishers that designed their web pages for speed. Some publishers did not adopt AMP because they knew their pages actually loaded faster than AMP pages.

The speed benefits Google marketed were also at least partly a result of Google’s throttling. Google throttles the load time of non-AMP ads by giving them artificial one-second delays in order to give Google AMP a “nice comparative boost.” Throttling non-AMP ads slows down header bidding, which Google then uses to denigrate header bidding for being too slow. “Header Bidding can often increase latency of web pages and create security flaws when executed incorrectly,” Google falsely claimed. Internally, Google employees grappled with “how to [publicly] justify [Google] making something slower.”

The unredacted filing also states that internal documents show that AMP pages brought 40% less revenue to publishers:

Google gave publishers a Faustian bargain: (1) publishers who used header bidding would see the traffic to their site drop precipitously from Google suppressing their ranking in search and re-directing traffic to AMP-compatible publishers; or (2) publishers could adopt AMP pages to maintain traffic flow but forgo exchange competition in header bidding, which would make them more money on an impression-by-impression basis. Either option was far inferior to the options available to publishers before Google introduced AMP. Just how inferior? According to Google’s internal documents, 40 percent less revenue on AMP pages.

The complaint succinctly summarizes the reason many publishers felt under the gun to allocate developer resources for AMPing up their websites, and why Google was in the position to force the issue despite widespread criticism of the AMP project. It also describes how Google’s anticompetitive tactics and control of the market essentially has small publishers over a barrel:

Direct evidence confirms Google’s monopoly power in the display ad network market. GDN charges high double-digit commissions of at least 32 percent on advertising transactions, which, according to public sources, is double the “standard rate” elsewhere in the industry. Internally, Google acknowledges that its fees are very high and that it can demand them because of its market power. For example, in an internal 2016 conversation, Google executives commented that Google’s ad networks make “A LOT of money” with its commission, and they acknowledged that they do this because, quite simply, “we can.” “Smaller pubs don’t have alternative revenue sources,” explained one Google employee when addressing the lack of viable competing ad networks available to its customers.

The suit, led by Texas Attorney General Ken Paxton and nine other state attorneys general, also exposes a number of code-named programs. Project NERA is the most insidious among these and one that publishers should know about:

Project NERA was Google’s original plan to create a closed ecosystem out of the open internet. Google documents reveal that Google’s motive was to “successfully mimic a walled garden across the open web [so] we can protect our margins.” For Google, Project NERA’s walled garden meant two things: controlling the design of publishers’ ad space, then forcing those publishers to sell their ad space exclusively through Google’s products. According to internal Google documents, this strategy would permit Google to extract even higher intermediation fees. A Google employee aptly described Google’s ambition for Project NERA by acknowledging that Google wants to “capture the benefits of tightly ‘operating’ a property … without ‘owning’ the property and facing the challenges of building new consumer products.” Google’s nickname for this walled garden plan was “not-owned-but-operated,” or “NOBO” for short.

The complaint also alleges that Facebook and Google colluded to manipulate header bidding auctions, among many other anticompetitive practices.

Google has not yet responded to the AMP-specific allegations but has published a response to the Department of Justice, calling the complaint “a deeply flawed lawsuit that would do nothing to help consumers.” The post attempts to refute the DOJ’s “dubious complaint” with demonstrations of how easy it is to change the default search engine on different devices.

Earlier this year, Google’s Director of Economic Policy, Adam Cohen, addressed the claims that AMP was designed to hurt header bidding, saying that it was created in partnership with publishers and other tech companies to help webpages load faster and improve the experience on mobile devices.

“AMP supports a range of monetization options, including header bidding. Publishers are free to use both AMP and header bidding technologies together if they choose,” Cohen said. “The use of header bidding doesn’t factor into publisher search rankings. 

The AMP project has not officially responded to the allegations in the unredacted complaint. Google transferred the project’s governance to the OpenJS Foundation in 2019, a move which skeptics hailed as “mostly meaningless window-dressing.” In August 2021, ex-AMP Advisory Committee member Jeremy Keith gave a behind-the-scenes look at the project in his resignation announcement. “It has become clear to me that AMP remains a Google product, with only a subset of pieces that could even be considered open source,” Keith said.

In the beginning, AMP was not a user-friendly product. It severely encumbered publishers and and was roundly denounced by advocates of the open web. In Google’s zealous drive to get publishers to adopt AMP, the company began investing heavily in developing WordPress plugins that would make make its products easier to use. More than 500,000 WordPress sites are now using the official AMP plugin.

The DOJ’s complaint alleges that Google is exploiting the position of smaller publishers that do not have any other options for revenue. It identifies AMP as a vehicle for anticompetitive practices and exposes viperous initiatives like Project NERA that do not have publishers best interests in mind. Any project that would seek to build “a walled garden across the open web” doesn’t seem particularly complementary to democratizing publishing. Getting to the bottom of these concerns should be a priority for the WordPress community and should inspire more scrutiny over Google-led core projects.