Marketing assets

Google’s Topics API raises more questions than answers

Data-driven thinkingis written by members of the media community and contains fresh insights into the digital revolution in media.

Today’s column is written by Eric Schmitt, Research Director at Gartner.

Amid fundamental changes in privacy standards and regulations, Google has topics introduced. In alphabetic language, Topics is a Chrome Privacy Sandbox “interest-based advertising” proposal and initiative that aims to “display relevant advertisements”.

The rest of us can be forgiven for just calling it an ad targeting framework.

Google’s introductory thematic communications suggest a broad and crude approach to targeting.

But the limited information shared so far raises more questions than it answers.

What we know – and don’t know – about the subjects

Simply put, topics are audience segments organized into a taxonomy, which Google will define and manage. (The company has suggested, however, that it could eventually cede control to an outside party.) Google has previously published an initial “design” of 350 Topics to give users a taste.

We also know a little how topics will work. At any one time, a user’s browser will associate with up to five themes, plus one additional randomly chosen theme. To target and serve an ad, the publisher or ad technology intermediary will query the visitor’s browser, which will return up to three topics from the set of six. Browser-to-topic pairings (topic segment membership) will be refreshed weekly. Google also promises to mitigate the risk of data aggregation and re-identification and to provide end users with transparency and control.

So what do not do we know?

For one, the total number of topics, which will be crucial for the overall accuracy and effectiveness of the ads. While Google has said the taxonomy will have a few hundred to a few thousand topics, a more specific range would clarify how far topics will deviate from the current buffet of ultra-precise, highly personalized segments based on cookies and user IDs. devices.

A taxonomy of 3,500 Topics, for example, is an order of magnitude larger than the “design” sample of 350. In the context of a US population of 268 million adults, this could mean the difference between a size segment average of 75,000 people and 750,000!

The granularity of topics also remains a question mark. the IAB Taxonomy 2020 of 1,500 lines includes 500 “interest-based” lines, 800 purchase intent segments and over 200 demographic attributes. These include age, gender, income, education, occupation, marital status, household size, “life stage”, property compared to rent and “net worth”. Which of these subjects will he understand? For the moment, it is not clear.

It’s also unclear if Topics offers geo-targeting. Depending on how Topics handles geography, US media planners might be able to coordinate digital ad buys with the 210 designated marketing areas used for radio and television planning, or the ~42,000 zip codes. postal. Of particular note is the postcode extension, which can be a particularly accurate identifier – and therefore a privacy risk.

Additional questions abound:

  • How will frequency capping and additional viewership work? Assuming they will work at all.
  • How far can Topics scale beyond Chrome? Browsers other than Chrome – primarily Apple’s Safari – have a a third of the world market and half of the US market.
  • How (if at all) will Google use the Subjects with respect to its owned and operated inventory? Specifically, what about search, YouTube, and display?
  • Will Google replace its custom audience targeting features (based on first-party data) with Topics-like targeting? This would go a long way to creating a more level playing field, although it seems unlikely that Google willfully put itself at a disadvantage.
  • How the dynamics of the new Privacy Sandbox on Android get in the game? Especially when it comes to geography and location data.

An unpredictable road ahead

from google Privacy sandbox timeline does not offer any overview of the company’s calendar. For one, Topics will need to be production-ready by the time Google deprecates third-party cookies in “late 2023.” On the other hand, Google has has already delayed Privacy Sandbox’s biggest milestone by two years. This could further delay, especially if the lengthening of deadlines works to its own economic advantage.

Despite the long list of unknowns, CMOs and ad budget owners have no choice but to keep a close eye on topics. After all, Google just surpassed $209 billion in ad revenue last year and now accounts for 30% of the overall global advertising pie – $710 billion in 2021, by MAGNA.

Google dominates search and TV streaming (YouTube), is dominant in advertising measurement and powers a sprawling ecosystem of ad-relevant Android devices, including smartphones, TVs and speakers.

However Topics shakes, it seems certain that sooner or later Google will incorporate it into its assets and properties far beyond Chrome.

Follow Gartner (@Gartner_inc) and Ad Exchange (@adexchanger) on Twitter.