Google Topics Will Require Supplemental Signals to Succeed

Since Google first announced Topics API, their new solution to help the advertising industry cope with the loss of third-party cookies, it has been met with mixed reviews. On GitHub, Google even admits that “the mapping of sites to topics will not always be accurate” and the “training data is imperfect" since it's created by humans—Google is rocking the digital advertising industry as we’ve known it, but the intention of Topics is only for “the labeling to be good enough to provide value to publishers and advertisers, with iterative improvements over time.” Privacy-safe should not equate to “not always accurate.” Privacy is nonnegotiable as we move into the next phase of online advertising, so our infrastructure has to put privacy first without jeopardizing the precision and accuracy that are required for performance. Even though we’re moving in the right direction, Topics API has an overwhelming number of limitations. Public testing of the solution is now underway, but “good enough” and “not always accurate” could prevent it from getting the industry buy-in that is required to put a final nail in the coffin of the third-party cookies. Advertisers’ Sweet Tooth Fades, With Many Unfazed by Google’s Latest Cookie Removal Delay Limitations that need to be addressed Topics will identify just that: “topics” of interest based on users’ Chrome activity. Defining consumers’ interests by only five broad categories will offer superficial data and will not give an accurate picture of what users are interested in across both their work and personal lives. Topics are decided based on sites that users frequently visit, with each publisher site only representing up to three categories. With a large news organization like USA Today that covers a wide swath of topics from politics to business to sports to tech to lifestyle and more, it will be extremely difficult for Google to accurately represent the inventory and reader interests of such publishers. After testing Google Topics for 90 days, the Gannett vp of product shared the topics breakdown and the percentage of results each made up—52% was broadly labeled as “News.” https://twitter.com/burk504/status/1550183541509685251 Topics does not account for consumers’ interest outside of Chrome, so there is no consideration for what the user engages with on connected TV or Safari, for example. Without an understanding of the topics people read on their iPhones, marketers have a very incomplete view of consumers’ interest, which makes it tough to reach them with relevant advertising—which makes it even tougher for their advertising campaigns to drive sales. This will be especially challenging for performance campaigns to drive conversions. The fact that each week one of the five topics is calculated as a “noise” barrier makes it even more difficult to gain an accurate understanding of users and their interests. Marketers cannot achieve meaningful performance metrics if they are targeting based on a random noise barrier. Topics also only identifies domain-level interests, which creates another limitation for fully understanding what is relevant to the consumer and advertising products or services they are likely to buy. For example, say a user is interested in golf and Topics only analyzes the URL: Topics would pick up if the user visited Golf.com but would fail to signal that a visitor to ESPN.com is interested in golf. Topics would miss out on golf-related stories covered on ESPN.com, leaving out potential articles of interest for the user. At the domain-level, Google is essentially packaging Chrome browsing history and rebranding it Topics. Nuances with taxonomies The initial design for Topics taxonomy only included a minimal 350 topics, which categorizes consumers much too broadly to truly understand the nuances of what is relevant to them. An advertiser is much more likely to drive conversions with a performance campaign if they understand a consumer is researching new golf drivers rather than just reading about golf in general. Google quickly built out this taxonomy using keywords and humans. While Google does plan to work with third parties to expand its taxonomy, it is still building a standard taxonomy, and while standardization across various vendors may appear helpful, it actually unintentionally limits the power of true semantics. Any standard content taxonomy leaves no room for the dynamic nature of content, which hinders both the buy-side and the sell-side. A new solution that promotes a reliance on another static taxonomy paralyzes the industry’s understanding of the true opportunity in content across the bidstream. Even though Topics API lacks the granularity needed to form a deep understanding of who someone is, people may not want to be lumped into some of the categories in the Topics taxonomy, like legal services or dating and personals. There are far fewer of these sensitive categories than the IAB Taxonomy has, but sensitive materials could be statistically correlated with the topics within the taxonomy. Supplemental signals The current proposed state of Topics API limits advertisers’ ability to precisely reach their target audience and lacks real-time optimization.While Google Topics is a step forward for consumers’ data privacy, this wide swath of interest-based targeting is a step backward for advertisers who need precision to drive performance. Privacy and granular targeting do not have to be mutually exclusive. Advertisers should consider that Topics API is not a standalone tool or replacement for cookies. Agencies should begin testing out different alternative solutions to pair with Topics across a range of vendors to see what kind of performance they can get. Now is the time to be testing first-party data or contextual signal strategy to supplement Topics API. For example, for USA Today and their “News”-focused domains, contextual signals could help advertisers hone in on the actual subject matter down to the paragraph-level. As marketers focus on building their first-party data infrastructure or contextual intelligence based on audience campaigns, Topics could nicely complement the consumers’ digital profiles. Overall, I want to see Google address the limitations with Topics head on so they can get the industry on board. I do applaud the Privacy Sandbox efforts to offer different strategies that prevent cross-site tracking and preserve audience privacy. Although logistically Topics solution has some kinks to work out to truly be a performance tool, it is a privacy-safe step in the right direction. How Trusted Media Brands Is Getting Advertisers to Ditch Third-party Cookies