Google Sandbox could push news publishers out of business

As Google prepares to remove third-party tracking cookies from its Chrome browser later this year, publishers, advertisers, and the UK Competition & Markets Authority are giving their proposed replacement — Google’s “Privacy Sandbox” ad system — a closer look.

Many don’t like what they’re seeing.

Google is developing Sandbox under a set of legally binding “commitments” intended to ensure that Google doesn’t abuse its dominant position in the online ad market. It can only introduce Sandbox once the CMA has signed off on the final product.

The CMA recently threatened to ban the company from implementing Sandbox entirely: “Google cannot proceed with third-party cookie deprecation until our concerns are resolved,” the CMA said in its most recent report on Sandbox. It wants “to ensure that Google does not use the tools in a way that self-preferences its own advertising services.” Currently, the CMA believes that Google is working to comply with its requirements: “Based on the available evidence, we consider that [in the most recent reporting period], Google has complied with the Commitments”.

Publisher cookies currently underpin much programmatic advertising revenue by allowing marketers to better target their messages at readers based on their interest and behaviour.


Sandbox Topics: Clunky and gives control to Google

The CMA is specifically worried about one aspect of Sandbox, named “Topics”.

Content from our partners

The Topics system allows advertisers to target readers via a new function in their Chrome browsers which will track the most recent themes those users have shown an interest in. For instance, if an advertiser wants to reach users interested in “sport”, any publisher using Topics will be able to show an ad to any user if “sport” was one of the user’s top five browsing interests of the last week. Crucially, the Sandbox system — not the publisher — chooses the topic labels. The intent is to allow advertisers to target users based on their interests but to keep their actual browsing history private from both advertisers and publishers.

The system is so clunky, however, that the CMA — along with multiple sources interviewed by Press Gazette — believes it may simply tilt the ad market in Google’s favour.

The CMA is concerned by the fact that there is one lone arbiter of what counts as a “topic” and what does not: Google.

Topics consists of a list of 469 themes that Google thinks advertisers will be interested in. Most of the topics are commercially anodyne: “pets & animals” and “shopping”, for instance.

Five Topics covering pets but nothing for science

The list is also short. For comparison, Reach’s Mantis system contains 1,600 different topics that can be targeted by advertisers. Dotdash Meredith’s D/Cipher system claims 15,000. Both systems compete against Google.

Google’s list has its quirks. It deliberately does not include “sensitive” topics. Thus there are no topics for sex or dating, but there are five different topics for pets. There is no topic for religion. There is no topic for science but there is one for “science fiction”. Running and walking are contained in the same topic. The list is generated by staff at Google. It cannot be added to or altered by advertisers or publishers.

The list’s diminutive scope has puzzled many. “If a page is about religion, it should be understood to be about that and not something else,” said Justin Wohl, chief revenue officer at Snopes.com, the myth-busting website.

“Google might advantage itself by manipulating the Topics API taxonomy which it currently controls,” the CMA said in its Q4 2023 progress report on the development of Sandbox. That language — and CMA’s outright threat to halt Sandbox — was notably harsher than its previous assessment, delivered after Q3 2023. The CMA declined to comment when reached by Press Gazette.

Google confirmed it controlled the list but said it expected the list to evolve. It also said that it was trying to balance the need of marketers for “granularity” without making the data it produces so specific as to strip users of their anonymity.

Niche publishers aren’t supported by Topics

The CMA is concerned that the 469 topics will play to the advantage of large publishers who cover multiple interests and hurt smaller, niche publishers whose content isn’t classified by Topics but who might nonetheless offer a better-defined audience. For instance, there is a topic for “Long Distance Bus & Rail” but there isn’t one for model railways — a totally different market.

“Smaller publishers and advertisers may be less able to mitigate any revenue loss from third-party cookie deprecation and the potential corresponding shift from open display towards direct deals. We are aware [of] this potential effect and will consider it as part of our overall assessment of the competition impact of the proposals,” the CMA said in its most recent assessment.

The IAB, in a technical review, said much the same thing: “Privacy Sandbox may limit the industry’s ability to deliver relevant, effective advertising, placing smaller media companies and brands at a significant competitive disadvantage. The stringent requirements could throttle their ability to compete.” Google, in a lengthy published response, said the IAB’s review “contains many misunderstandings and inaccuracies”.

Google said it wanted to avoid the problem of publishers misclassifying their own sites for financial gain: “Publishers can describe their inventory for digital ad sales [on the open market] like they do today, and the contextual information they provide will always be accessible in digital advertising auctions.”

“Really, it’s hard to shake the feeling that all of the Privacy Sandbox solutions are going to best serve Google buying platforms, right?” said Snopes’ Wohl. “It seems like it will work well for their AdX business, that they will play nicely together. But it’s TBD whether it’s going to be useful or valuable when trying to monetise with non-Google partners.”

Sandbox could lead to 30% fall in ad revenue for publishers

Publisher cookies on Chrome offer advertisers infinite ways to target users in infinite combinations. Under Topics, most publishers will be squashed into a handful of labels.

“It removes control. It moves flexibility for publishers to do what’s right for them,” says Stu Colman, senior director for European identity at The Trade Desk, an online ad platform, which competes with Google. “They’re not in control of how they’re classified, they’re not in control of the number of classifications that are available for them to be classified in. And they’ve got no ability to influence or affect what they are then represented as, that’s entirely controlled and gate-kept by Google.”

Colman says he has seen testing of Sandbox that has shown a potential 30% drop in CPMs to publishers. “If you walk into any publisher now and say, really sorry, but tomorrow you’re going to make 30% less money, that’s probably the end for a lot of publishers and that’s a really, really, really bad thing”, he said.

Execs at Criteo, another online ad platform that competes against Google, are also sceptical. In 2022, Criteo did a study of Topics’ “prediction power” and found that cookies were on average five times more relevant than Topics for targeting, according to Lionel Basdevant, Criteo’s product director.

At Reach, group digital & innovation director & EVP Mantis Terry Hornsby says he has been working with Google on Sandbox development — trying to map Mantis to the Topics categories — but “it needs more work. It needs more clarity on how it’s determining topics for that user”.

Sandbox could be data grab for Google

Insiders are also worried that Topics will trigger a net transfer of valuable data — and thus ad dollars —  from smaller publishers to Google because publisher Topic labels are often more granular than those on Youtube which are shared within the same advertising ecosystem.

For instance, a niche site for designer clothing might seed its users with the “Luxury Goods” topic label. Users who have previously looked at “Luxury Goods” websites could be targeted via that topic when they arrive on Youtube, with reasonably specific advertising from high-end advertisers.

But anyone watching a video about luxury goods on Youtube would only pick up Youtube’s non-specific topic labels: “Online communities”, “TV and Video”, and “Arts and entertainment”. If that user later arrives on a publisher website, those labels will offer no clues that the user is interested in luxury goods.

“That’s an advantage for Youtube. Because Youtube is not giving up any of their audience data to use Topics. However, the niche sites that get classified with a specific topic are passing their audience data into Youtube,” says Don Marti, VP of ecosystem innovation at Raptive, a publishing tech provider.

Google said: “We acknowledge that sites with more general interest domains are likely to contribute less granular topics than sites with more niche interest domains. However, not all niche sites contribute commercially valuable topics.”

‘Complex and convoluted’

But it is the complexity of Topics and Sandbox that unites Google’s critics. Many believe that Sandbox will be so difficult to implement by publishers —the full suite requires 16 different API integrations — and so confusing for advertisers that many marketers will simply do the easiest thing: go to Youtube, Facebook or Amazon.

Paraphrasing adtech podcaster Ari Paparo, The Trade Desk’s Colman said: “All the functionality is available, but in the most complex and convoluted way possible.”

Google told Press Gazette that it encouraged companies to test Sandbox, and that the entire scheme was voluntary: “Companies are not required to use all (or any) of the Privacy Sandbox APIs.”

Note: This story was edited on 22 March 2024 to include a fuller response from Google.

Email [email protected] to point out mistakes, provide story tips or send in a letter for publication on our “Letters Page” blog

link