Google’s Latest Adjustments to Topics, Again, Favors Large Publishers

Privacy Sandbox proposal Topics is dealing another blow to smaller publishers. The Topics proposal suggests grouping users by interest-based cohorts based on browsing history, like auto or fitness, created by the browser and shared with a site—and its ad partners—when a user visits. In Google's January quarterly progress report, to assuage European antitrust regulators, it suggests assigning Topics using page-level metadata rather than a domain. Currently, Topics are assigned per domain, for example, the nytimes.com domain would be assigned the broad Topic of news. Assigning based on page-level metadata, which means larger publishers would receive Topics like cooking to the relevant page, instead of an umbrella Topic like news to a domain, in theory, is likely to benefit larger publishers, four sources told Adweek, further exacerbating the bifurcation of the web. “Essentially, this would make Topics into a pure semantic contextual solution, where page content and not the site determines the Topic,” said an ad tech executive, who requested anonymity to freely discuss industry relations. As Cookies Wane, Retargeting Protocol Fledge Emerges as Privacy Sandbox Favorite Topics has been open for testing since last year and are only available for 5% of Chrome browsers. It is just one solution marketers and publishers are testing to match an ad to someone's interest as cookies deprecate, including data clean rooms and seller-defined audiences. Still, privacy and effectiveness concerns have grown. A Google spokesperson said it is "continuing to develop Topics based on feedback from the industry, prioritizing use cases across both large and small publishers," and maintaining its current deadline of cookie deprecation of 2024. A big win for big publishers Under its current structure—where large publishers could be assigned generic Topics like news—each time a user travels from larger publishers to more niche ones, they would carry the more generic label, diminishing the smaller publisher's audience for advertisers. This structure is “a big win for publishers with lots of inventory at the expense of smaller, more focused publishers with valuable audiences,” said Loch Rose, chief analytics officer at Epsilon, since a more specific label, like auto, is more valuable to an advertiser than a more generic label, like news, Rose added. If a person visiting the auto research company Kelley Blue Book (KBB) then goes to Yahoo, the bid impression sent out from Yahoo will show an auto label, which would be more valuable for a certain type of advertiser and would enrich Yahoo’s audience. On the flip side, if the user visits Yahoo from KBB, it brings with it Yahoo’s generic label news, which could be of less interest to certain advertisers, especially when the interest of KBB’s audience is skewed towards auto. “It's a difficult trade-off,” said Rose. “And publishers are rightly complaining about that.” While under the page-level metadata suggestion for Topics, niche publishers can retain the value of their audiences, they should expect less competition for their inventory compared to what it would have been without Topics, since the advertiser can buy ads based on that niche label on all those other visited sites. “Publishers want competition for their ad inventory," said Rose. "The more advertisers are competing to buy their inventory, the higher their sales to advertisers and thus revenue." Further, a niche publisher would want advertisers to show ads to its site visitors while those visitors are actually on the site, not when they are visiting somewhere else. Some privacy concerns solved while others emerge However, focusing on the page instead of the publisher helps address web-standards body W3C’s privacy criticism that Topics tracks a user's site browsing history. This would instead focus Topics on the content that the user likes to consume, as opposed to where they go to consume it. Nonetheless, advertisers are still at risk of sending higher bids to low-quality sites because of cross-site tracking, said Don Marti, vp of ecosystems innovation at CaféMedia. Once a user has a Topics label, it is visible on any site they visit. This means a user can visit a premium publisher and then visit a lower-quality site, and in doing so, carry with them the former publisher's label that has the ability to attract a bid from a high-value ad campaign. "Topics API is technically a privacy improvement over third-party cookies because it provides only a subset of what you can do with a third-party cookie," said Marti. "But part of that subset includes adverse effects of cross-site tracking on legit sites." Aside from these niggles, there appear to be inaccuracies in the current classification of Topics. CaféMedia saw how-to tips for mechanics tagged as vehicle shopping, but pages on how to find shoes or how to tell if mayonnaise has gone bad were also tagged as vehicle shopping. After looking closely at several Topics API data, Marti found 17% of its automotive and vehicle pages were classified by Chrome as news. Given Topics’ current structural and privacy flaws, “I don't see topics having a future,” said James Rosewell, CEO and co-founder of data company 51Degrees.

Publishing News

Google’s Latest Adjustments to Topics, Again, Favors Large Publishers

The tech giant suggests assigning Topics based on page-level metadata rather than the domain

Trishla Ostwal

About

Subscriptions

Events

Publications

WORK SMARTER - LEARN, GROW AND BE INSPIRED.

Subscribe today!

Trishla Ostwal