The Impact of the Google Topics API

It’s been a slow, lingering death, but the deprecation of cookies is finally in sight. Google announced it would disable third-party cookies in Chrome in 2022, but delayed it until sometime in 2023 – and then again to 2024. There is still no precise date in place at this time.

The delays were caused, in part, by the decision not to move forward with Google’s early attempt at another solution — the Federated Learning of Cohorts (FLoC). FLoC received such serious pushback and negative reviews from privacy experts that Google dumped the idea and instead moved to its new solution — Google Topics API. While the solution hasn’t been finalized, Google has announced plans in the company’s Privacy Sandbox and is preparing a developer trial of its Topics API solution shortly.

How Does the Google Topics API Work?

Google Topics API will review a user’s online behavior to determine interests based on a broad recognizable high-level topic. For example, Chrome would examine a user’s browsing history to determine general interests based on website topics. 

Websites have the option to declare which topics they want associated with their site. 

Each week, the browser selects three topics for each user to show to publishers and advertisers. One topic per week will be chosen from each of the previous three weeks. Thus, frequent topics will change weekly as behavior changes.

How Does Google Assign Topics to Websites?

The application programming interface (API) labels each website with high-level topics. The API would collect the topics from among a user’s most frequent visits and shares a selection of them with websites and advertisers for contextual advertising.

Right now, the topics are limited and broad. For example, websites about basketball, football, volleyball, and hockey would likely be labeled as Sports. Unlike cookies and granular behavioral data, Google topics of interest are, as the company says, “coarse-grained advertising topics that the page visitor might currently be interested in.”

The handful of topics will last for three weeks, allowing advertisers to target based on these topics with one significant caveat: websites and advertising partners, for example, will only be able to target if their topics align with the user. If the assigned targets are parenting, music and movies, a website for auto enthusiasts would not be able to access any of those topics.

However, if a user was assigned auto and vehicles as one of their three topics per week based on their behavior, websites and advertisers would be able to deliver ads based on those targets.

How Does Google Develop a List of Topics for Users?

Currently, the list of topics is just 350, although Google says it is a trial of topics and anticipates growing the number. If you want to see the topics yourself, there is a list available on GitHub, sorted by taxonomy.

This is significantly different than the type of information advertisers have had access to in the past, including deep behavioral data, intent data, and third-party data. Google says the limited number of topics is intended to reduce matching topics to individual users. Google also says it will not be including any identifiers that are considered sensitive or would allow more narrow matching. This would include gender, race, religion, and more.

To show how limited this list is, FLoC had more than 30,000 classifications.

To make it even more difficult to identify individual users, Google will also include random topics in the handful it provides to advertisers.

How Does Google Assign Categories to Advertisers?

According to Google, three steps occur before advertisers see topics.

  1. Map website hostnames to topics of interest
  2. Identify the top topics for users 
  3. Provide topics of interest to ad tech platforms

As it currently stands, a machine learning model would infer the topics from the hostnames of the pages visited, using classifier models distributed with the browser. These are available for development, while the classifier would store the top five topics. 

To keep privacy advocates happy, Google says it will not store these topics. Instead, they will reside in the browser. When a user visits a website, and there is an ad call, the Topics API will provide the list of Google topics of interest to the AdTech platform. The API will return three of the stored topics at random and, according to the privacy sandbox, may also include a random topic chosen from the full list.

An important thing to note for advertisers is that this does not bring in any outside data. If a topic is assigned, it will be based on observable behavior from prior website visits. You can see a more detailed explanation of how this happens and some of their more technical aspects on Chrome’s developer site.

Programmatic advertisers would have access to Google Topics API during ad calls through the ad vendors, which will be crucial since consumers today shift channels and websites frequently. In today’s fragmented environment, advertisers need a holistic approach with an integrated platform that can help target consumers across the entire omnichannel digital ecosystem.

Websites will need to embed the Topic API (or a service that allows topics to be generated) for their site. If they don’t, they will not be able to participate. As a note, if a website previously opted out of FLoC, the Google Topics API will be disabled unless publishers enable it.

How Google Topics Improve User Privacy for Interest-Based Advertising

Google Topics’ goal is to provide some contextual information for interest-based advertising while significantly increasing user privacy. In fact, some users may be hidden from Google Topics entirely as user controls will allow people to opt-out completely.

Users will also be able to see the topics that are assigned based on their behavior and remove any that they do not like. To make it easier for users, the identifier uses common language rather than a long string of numbers and letters that you typically find with cookies or its earlier privacy solution, FLoC.

Behavioral Targeting Without Third-Party Cookies

Traditionally, advertisers used third-party cookies to target and retarget ads aimed at select groups or even individual users. With the vast amount of data currently available — much of which can tie back to individuals — advertisers can narrowly target using a combination of first-party, third-party, and intent data as well as vast amounts of other data amassed in the advertising ecosystem. 

Since Google API reflects only broad user behavior from site visits, it will be less granular than what advertisers are used to or like. For example, someone who is assigned the topic of fitness might include someone shopping for fitness clothes, hard-core bodybuilders, or someone trying to drop a few pounds looking for at-home fitness routines. Cookies provided a way to segment within a category like fitness to target users. If you’re selling muscle-building supplements for people that do intense workouts every day, you probably don’t want your ads running on a site for trendy workout outfits. 

Likewise, if someone is assigned a topic like auto, there’s no way to know if they are interested in luxury vehicles, rental cars, or used cars. Christine Zirnheld, Digital Marketing Manager at Cypress North gave the example of advertising coffee products. Google Topics API would classify an interest in coffee as Food & Drink. By comparison, the IAB audience taxonomy has roughly 1,500 audience segments, including coffee, coffee and tea, coffee creamer, coffee filters, and more.

To target more precisely, website owners and advertisers will need to put a premium on acquiring and managing first-party data.

Emphasis on First-Party Data

Digital marketers need to create a compelling value proposition to get potential customers to voluntarily share information.

“To succeed in a world of consent-based advertising, digital marketing leaders must accelerate when, where and how they collect, aggregate and deploy first-party data,” said Chelsea Gross, Director Analyst at Gartner. “They’ll need best-in-class tactics for incentivizing customers to share first-party data, as well as data and analytics (D&A) management capabilities to handle the data itself.”

First-party data, gathered when users interact directly with you, can provide a significant amount of data to help target. Examples include:

  • Clicks from emails
  • Purchase history
  • Product preferences
  • Time spent on certain pages
  • Demographics information
  • Social data

While users must voluntarily provide this information, the more first-party data you acquire, the better you can target.

First-party data is, by nature, passive. There’s also zero-party data that is proactive. For example, potential customers share their name and email address to receive gated content in B2B or a discount coupon in B2C. Because this zero-party data comes directly from users, it’s incredibly valuable for publishers to accommodate advertiser targeting.

Zero-party and first-party data will be essential for publishers and advertisers that may not have direct visibility into users’ past behavior otherwise. Without this data, advertisers would be totally reliant on their ad tech vendor for general interest insights into a topic.

Advertisers can take steps now to grow their first-party data and diversify their data collection, including taking proactive steps such as:

  • Email marketing to capture data, including demographics, email addresses, and interests
  • Use surveys, customer feedback, and other customer touchpoints to consolidate data into your CRM or database.
  • Increase the number of touchpoints throughout the customer journey to capture additional information, behavior, and interests.
  • Invest in a data management platform (DMP) or customer data platform (CDP) to develop more robust customer data profiles.
  • Capture campaign data from cross-channel marketing campaigns.

This data will help you create more contextual campaigns. You can get more ideas on how to build out your first-party data and strategies for the cookieless future by downloading our free guide, The Great Cookie Countdown.

Topics Does Not Assign Cohorts to Users

Google Topics API does not assign cohorts to users. Google’s first attempt at replacing cookies — FLoC — worked on the concept of cohorts. By creating cohorts of users with similar interests, such as those with the intent to buy a new car, advertisers could target a group of potential customers that had demonstrated intent.

Privacy advocates demonstrated that these cohorts could be supplemented with additional data to uncover individual users. As such, Google dropped FLoC.

Topics API will not use cohorts, which Google says will make it significantly more difficult to identify individual users. As such, advertisers will have significantly less precise targeting than cookies with fewer options for audience matching.

Partner With AUDIENCEX for Emerging Tech Updates

The rules and configurations for Google Topics API are not finalized. Google says it is still collecting feedback and testing different iterations before settling on its final configuration. So, it’s possible things will change again.

There are still privacy hurdles to clear, such as the W3C Technical Architecture Group (TAG) and UK’s Competition Market’s Authority (CMA), which have raised concerns about First Party Sets (FPS). FPS would allow multiple domains with a common owner to “ allow related domain names to declare themselves as the same first-party.” The concern remains that they would give an unfair advantage to media conglomerates.

Whatever the final rules, the impact will be significant. Safari and Firefox already block third-party cookies by default. When Chrome joins in — with its nearly 67% share of the browser market — it’s going to be a big shift. Considering nearly all current programmatic ad spend uses third-party data, advertisers need to look at their current campaigns and plans to strategize the path forward.

Advertisers need to stay on top of new development because the cookieless future will be here sooner rather than later. One way to do that is by keeping an eye on AUDIENCEX Insights for the latest trends and updates.

AUDIENCEX works to support brands and agencies, providing seamless access to the omnichannel media landscape along with a fully integrated tech stack to provide solutions for optimization, audience insights and segmenting, and futureproof targeting. Our platform is backed by our team of experts, who provide the technical and strategic support needed to leverage and optimize campaigns to maximize performance. AUDIENCEX is the leading integrated platform offering agile and scalable solutions across more than 24 DSPs, including programmatic options for:

  • Display
  • Video
  • Audio
  • Native
  • Connected TV (CTV)
  • Social
  • And more

AUDIENCEX provides end-to-end campaign management, full-service creative, and real-time performance reports to drive optimized results. As we monitor the ongoing changes to the digital advertising landscape, we continually work to integrate futureproof solutions into our unified platform for brands and agencies. Reach out today if you’d like to learn more about how AUDIENCEX can support your marketing and ensure performance outcomes today and well into the future.