Publishers are concerned Google’s Topics will lead to a higher ‘ad-tech tax’

Google’s sand(box) is starting to get in between publishers’ toes.

Though far from realized yet, Google’s Privacy Sandbox, the company’s ambitious attempt to make the web less invasive while still efficient for advertisers, has found a number of critics in publishers, concerned that some proposals won’t actually marginally help them without an overreliance on ad tech.

Rewind: Google’s long said it’s crushing third-party cookies—just yesterday it kicked the deadline (again) to the second half of 2024—but it still wants to keep advertisers happy, so it’s testing several alternatives that promise some level of audience targeting.

First there was FLoC, or Federated Learnings of Cohorts, which segmented audiences into thousands of categories based on their browsing history. But some privacy advocates were concerned that these audiences could be small enough to identify individual users.

So Google scrapped FLoC for Topics, assigning three topics to each individual user based on their internet habits and enabling advertisers to target those individuals based on those assigned topics.

• Google has roughly 350 topics for testing, which could grow to the low thousands.
• Every three weeks, topics are automatically deleted and replaced with three new ones.

Topic time

Here’s where it gets wonky. When publishers call the API, they’ll only see topics that Google has already assigned to the publisher. Sometimes a publisher only gets one, sometimes seven, with no limits set yet. (Subdomains can also have their own topics).

However, that’s what publishers already know:

• So, for example, if you’re a news publisher and the API has tagged your readers as interested in news, then that’s the only data point you’ll be able to share with advertisers.
• Conversely, if you’re really into motorbikes and you, as a reader, are assigned the motorbike Topic (stay with us), because USA Today probably wouldn’t share that label, USA Today won’t be able to tell advertisers you might be interested in buying a motorbike even though you’re on their site. They will just see that you are interested in news.

That’s limiting, argued Jeff Burkett, VP of product at Gannett, which tested Topics and found that (unsurprisingly) 52% of its readers were labeled news over 90 days. “We knew that already,” he told Marketing Brew. “We’re just getting into a world where it’s harder and harder, potentially, for publishers to survive…this doesn’t help.”

According to Burkett, who shared a screenshot of the API with Marketing Brew, USA Today is still tagged with only one topic: news. Conversely, CNET, the consumer-electronics site, was given five: consumer electronics, news, science, computers and electronics, and consumer resources.

“Ad-tech tax”

Here’s what Burkett said some publishers also don’t like: Middlemen and third-party ad-tech vendors that are plugged into many different publishers and can “see across all domains” will likely have access to a broader swath of Topics, and would be able to identify more audience interests, Burkett argued.

That’s intentional, explained Grant Nelson, a project manager at SSP TripleLift. “Chrome’s explainer states that the Topics API will in practice be called by third parties much more than by the publisher itself, but for the benefit of [the] publisher,” he told Marketing Brew.

That will “extract even more ad-tech tax” for publishers, argued Aram Zucker-Scharff, engineering lead for privacy and security compliance at the Washington Post, which hasn’t tested Topics.

“Buyers flock toward the perception of accuracy, and the presence of Google’s strongly trusted technology will create a situation where CPMs may become highly reliant on Topics being active, giving publishers who may see little direct benefit little choice but to activate the system,” he wrote to Marketing Brew via email. “The logic is—if a publisher doesn’t like how a third party operates, then why would they run them? Of course, reality doesn’t actually work that way. Publishers often find market or buyer pressures forcing them to implement vendors they would not work with on their own.”

In April, Alexis Sukrieh, CTO of the martech company Weborama, posted on GitHub that Topics had “very low (to zero) business value for the publisher.” Others had earlier expressed similar thoughts.

Josh Karlin, a software engineer at Google, responded and thanked Sukrieh for raising the issue, but said that “it is up to publishers to determine which third-party service providers they include, and up to service providers to determine what they’re willing to divulge to publishers.”

“From the point of view of Google, any publisher market power is a problem. They need to limit the access of the publisher to any valuable information they can,” said Don Marti, VP of ecosystem innovation at CafeMedia, who also participated in Topic’s origin trial. “It incentivizes every possible ad-tech provider to get onto as many possible pages as they can.”

It’s up for debate whether publishers should be able to set their own topics. A concern is that some no-name publishers would potentially be able to assign themselves a highly valuable topic.

“Some random cat blogger is going to say [their site] is all about BMWs and Palo Alto,” joked Marti.

Meanwhile, both Marti and Zucker-Scharff said they fear Topics could allow clickbait to fester, the same way it does under cookies.

And, because it’s Google, publishers may feel as though they need to opt in for fear that it could negatively impact their SEO. “If it’s a Google thing, whether it’s Google Plus, Google Amp, any kind of new thing from Google that the publisher is not convinced is in their interest? Everybody looks at the SEO consequences,” Marti said.

To be fair, both Marti and Burkett said Google had been communicative and open to hearing criticism—anybody can track feedback on GitHub. Meanwhile, some publishers have pitched their own standard called “seller-defined audiences,” which as AdExchanger recently reported, hasn’t yet been picked up by the industry.