“The Sell Sider” is a column written by the sell side of the digital media community.
Google recently began testing Topics API, the latest part of its Chrome Privacy Sandbox. It enables Chrome to determine a handful of topics, like “Fitness” or “Travel & Transportation,” that represent a user’s interests based on their browsing history. But unlike a third-party cookie, topics do not identify the user.
On the one hand, Topics API is a significant improvement over FLoC, the former Privacy Sandbox proposal for ad targeting. But that’s a limited gain on privacy for a poor trade-off in terms of utility and competition issues. For one thing, FLoC had 33,000 topics to choose from, while the Topics API has only 350.
The Topics API proposal also carries two unacceptable risks for the open web. First, it enriches large platforms at the expense of niche or independent sites, especially sites that invest time and skill to cover categories in detail. And second, the Topics API leaks audience information from trustworthy sites to sites that take advantage of advertisers and ad tech.
Unfair data and revenue shifts
If a user shows a strong interest in #VanLife by browsing sites with specific content about vans and related products, then a manufacturer can target them effectively on a general interest site. The blogs or niche sites created a valuable target with a very specific topic assigned to the user, which YouTube, ESPN, or any other major publisher can then capitalize on.
Meanwhile, someone who visits YouTube and is assigned “Entertainment” as a topic returns no value to niche sites. Big sites get all the value without giving any in return.
The problem is exacerbated by the fact that Google only assigns each site one or a few potential topic. Someone who visits BuzzFeed and reads a bunch of parenting stories is classified with a “News” or “Entertainment” topic instead of “Parenting,” because BuzzFeed is primarily a news and entertainment site. Smaller sites would do better with alternatives to the Topics API that promise better control over their audience data.
A step backwards on brand safety
Another issue with Topics API is that when you visit sites with commercially relevant topics – like recipes or product reviews – information about your interests can be leaked to disinformation and other unsafe sites in the form of topics.
Topics API shares this problem with other ad technologies that track users from site to site. But when the browser takes on more of the work of running ad auctions, that shouldn’t mean advertisers have to take a step backward on brand safety.
Topics API will need to enable the same level of validation that today’s ad systems do. Otherwise, legitimate publishers will drive up the value of low-quality or straight-up bad actors in digital media without any ability to investigate or curtail the practice.
Getting closer to benefiting all sides
Google has posted the Topics API proposal to GitHub for public review in the W3C’s new Private Advertising Technology Community Group. Ad industry players should participate in the discussion and file any issues on the Topics API GitHub repository to address concerns. (CafeMedia has already filed several issues, as well as a suggested change that would help Topics API treat general-interest and topic-specific sites more fairly.)
Today, Topics API is just a proposal from Google that must go through months of testing and other iterations before becoming a standard. Niche or independent publishers don’t have the time or specific expertise to participate in web standards development like the big platforms do, so their priorities can often end up lost in the process.
Let’s work together as an industry to ensure we find the best possible solution to maintaining a free, open web.
Follow CafeMedia (@CafeMedia_) and AdExchanger (@adexchanger) on Twitter.
For more articles featuring Don Marti, click here.