Google Chrome’s Topics API Explained + FAQs

Back in July 2021, Google Chrome floated the idea that its interest-based advertising proposal — Federated Learning of Cohorts (FLoC) — could be replaced by a topics-based advertising proposal. On January 25, 2022, these suspicions were realized when Chrome announced that it would be shutting down FLoC and replacing it with a new proposal — Topics API.

Table Of Contents

Key Points About Google Chrome’s Topics API

Frequently Asked Questions (FAQs) About Google Chrome’s Topics API

What is Google Chrome’s Topics API?

How will the Topics API work and how will topics be created?

How many topics will be available with the Topics API?

Will Topics contain sensitive categories, e.g., race, religion, sexuality, and medical history?

How does Topics differ from Federated Learning of Cohorts (FLoC)?

Why did Google Chrome change from FLoC to Topics?

How can publishers implement the Topics API into their websites?

Can publishers opt out or stop the Topics API from collecting data about users on their websites?

How can advertisers show ads to users via the Topics API?

What role will AdTech companies play in the targeting and delivery of Topics-based campaigns?

Is Topics a privacy-friendly ad targeting method?

Will the Topics API be implemented into other web browsers apart from Google Chrome?

When will Google Chrome’s Topics API be released?

Key Points about Google Chrome’s Topics API

  • Topics API is the latest addition to Google Chrome’s Privacy Sandbox and will replace its Federated Learning of Cohorts (FLoC) API, which was designed to run interest-based advertising but in a more privacy friendly way than ad targeting via third-party cookies.
  • Chrome will no longer work on the FLoC proposal and will instead focus on creating and testing Topics.
  • Chrome made this decision because of user privacy concerns surrounding FLoC, including the possibility of including sensitive topics in the cohorts and using cohort IDs to help create device fingerprints. 
  • The Topics API will enable advertisers to show ads to users based on the websites they visit. 
  • A classifier model will map website hostnames to topics. Only subdomains and root domains will be included and not the full URL. For example, the website football.news.com will have the topic of football associated with it, but news.com/sport will have topics related to news.com associated with it.
  • If a publisher doesn’t have the Topics API on their website, then topics generated for that website won’t be included in the weekly calculation.
  • The topics taxonomy will be given readable names, e.g. “Sport”, “Travel” and “Arts and Entertainment”. 
  • There are currently around 350 categories that will be included in the Topics API, but it’s likely that this number will change. However, the number of categories can’t get too big as it could lead to cross-site identification and tracking.
  • In the future, websites may be able to provide their own topics based on meta tags, headers, or JavaScript on their websites. 
  • Sensitive topics, such as race, religion, medical history, and sexual orientation, will be excluded.
  • In the future, users will be able to see which topics have been assigned to them, remove topics, and even disable the Topics API in their browser.
  • The topics will be taken from a user’s device (e.g. laptop and smartphone) and won’t involve any external servers, not even Google’s servers.
  • The Topics API will assign a user 5 topics each week, plus a randomly chosen topic to ensure that each topic has a minimum number of members (k-anonymity) as well as to provide reasonable privacy protections.
  • Each time the Topics API is called (e.g. by an AdTech platform), anywhere between 0 and 3 topics will be returned. Only those topics that the caller has observed the user visit in the past will be returned. 
  • If 3 topics are returned, then these 3 topics will represent the top 3 topics of that user for the past 3 weeks — i.e. one topic from each of the previous 3 weeks.
  • Topics associated with a user will only be kept for 3 weeks and will then be deleted. New topics will then be created.
  • When callers (e.g. publishers and AdTech companies) call the Topics API, it will return topic IDs for each of the 3 topics, which can be looked up in the topics taxonomy, a taxonomy version, and a classifier version.
  • If the user has cleared their web-browser history, has browsed in incognito mode, or has opted out of Topics, then the topics will be empty.

Frequently Asked Questions (FAQs) About Google Chrome’s Topics API

What is Google Chrome’s Topics API?

Topics (and its API) is a new initiative from Google Chrome that’s part of its Privacy Sandbox — a series of standards and APIs designed to replace the processes currently carried out by third-party cookies. Topics will replace Google Chrome’s Federated Learning of Cohorts (FLoC) and aim to show ads based on the topics a user is interested in, rather than the cohort they belong to.

How will the Topics API work and how will topics be created?

While the Topics API is still a work in progress and will likely change overtime, the general concept behind it is to assign a maximum of 5 topics to a user each week based on the websites they visit. There will also be a 6th random topic assigned to the user to ensure that each topic has a minimum number of members (k-anonymity) and to provide reasonable privacy protections. 

The 5 topics will be assigned to a user each week, but the actual start date of the week will be different for every user.

When a user visits a website, the browser will pass on up to 3 topics to “callers” — i.e. companies who have called the API, such as publishers and AdTech companies. These 3 topics will represent the topics that the user has been assigned over the past 3 weeks — i.e. 1 topic from each of the preceding 3 weeks, but in random order.

Only callers who have observed a user’s topics will be able to see them. So if a publisher has observed a user and has seen their topic “fitness”, then the next time that user visits their website the publisher will receive the “fitness” topic. The publisher will continue to receive this topic for the remainder of the 3 weeks. 

If a publisher or AdTech company hasn’t observed a user visit a website about the topic they’re targeting, then they won’t receive that topic.

How many topics will be available with the Topics API?

The Topics catalog will initially include a few hundred to a few thousand topics. Currently, the Topics taxonomy defines ~350 categories and includes main topics, such as Arts & Entertainment and their subcategories, e.g., Arts & Entertainment -> TV Shows & Programs -> TV Dramas -> TV Soap Operas.

The catalog of topics will be publicly available and will evolve over time. It’s not yet known whether websites will be able to create their own topics or be able to change the ones that have been assigned to them by the browser.

Will Topics contain sensitive categories, e.g., race, religion, sexuality, and medical history?

Compared to FLoC, the Topic’s catalog uses a readable list of topics, e.g. Fashion, Sport, and Entertainment. One of the main criticisms of FLoC was that it would be hard to exclude sensitive topics, e.g., ones that could represent a person’s sexual orientation, medical conditions, and race, because you wouldn’t know for certain whether a cohort ID represented any sensitive topics or not.

One of the main privacy considerations of Topics is to exclude sensitive topics so they can’t be used for ad targeting. In the future, Chrome users will be able to view and control the topics they’re associated with, delete certain topics, and even turn off the Topics API altogether.

We will learn more about how this will be done as time goes on as Chrome incorporates feedback and ideas from stakeholders across the industry.

How does Topics differ from Federated Learning of Cohorts (FLoC)?

With FLoC, web browsers would have been placed into cohorts based on their web-browsing behavior, e.g. Cohort 4872 could relate to people interested in tennis. With Topics, web browsers will be assigned a list of topics based on their web-browsing behavior. E.g. The topic “tennis” could relate to people interested in the sport and tennis equipment.  

FLoC would have incorporated on-device processing and machine learning to place web browsers into a certain cohort based on the websites they’ve visited. In the case of Topics, the browser will use a classifier model to link website hostnames to topics.

Why did Google Chrome change from FLoC to Topics?

The main reason why Google Chrome has moved away from FLoC and towards Topics is because of the criticism it received around user privacy. Even though FLoC was designed to be a more privacy-friendly option than ad targeting via third-party cookies, it still didn’t provide enough user privacy protections. Here are some of the main criticisms of FLoC in regard to privacy:

  • FLoC was still a form of ad personalization, which organizations like the EFF said should be eliminated altogether.
  • It could have been possible for companies to identify which cohort you belong to by matching PII (e.g. email addresses) with your cohort ID.
  • Because the cohorts would have been generated by the browser, it would have been hard for the cohorts to exclude sensitive topics, e.g. race, religion and sexual orientation. 
  • FLoC would have made it possible for companies to use the cohort ID as, along with other pieces of information, to create device fingerprints. These device fingerprints could then be used to identify individuals, which goes against what FLoC was designed to do.

How can publishers implement the Topics API into their websites?

The overriding goal of Topics API is to improve user privacy while supporting an online advertising ecosystem for both advertisers and publishers. In terms of how it works, publishers (and their AdTech partners) will need to call the Topics API to receive the topics associated with the user visiting the page. The Topics API can be combined with other contextual signals, e.g. page URL, to assist with selecting the most relevant ad for a given user.

In the future, publishers will likely be able to use meta tags, HTTP headers, or JavaScript code to help the classifier identify topics included on publishers’ web pages.

Here is a basic example of how the JavaScript code would work. Please remember that the Topics API is subject to change.

// document.browsingTopics() returns an array of up to three topic objects in random order.
const topics = await document.browsingTopics();

// The returned array looks like: [{'value': Number, 'taxonomyVersion': String, 'modelVersion': String}]

// Get data for an ad creative.
const response = await fetch('https://ads.example/get-creative', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
  },
  body: JSON.stringify(topics)
});
// Get the JSON from the response.
const creative = await response.json();

// Display ad.

Can publishers opt out or stop the Topics API from collecting data about users on their websites?

Publishers can opt out of the Topics API by adding the following code to their website:

Permissions-Policy: browsing-topics=()

Also, if the opt-out code for FLoC is present on a website, then it will also opt the website out of the Topics API.

Permissions-Policy: interest-cohort=()

How can advertisers show ads to users via the Topics API?

It’s likely that advertisers will simply need to state which topics they’d like to target and then their AdTech platforms will match those topics with the ones obtained from the Topics API. 

For example, a consumer goods company wanting to show an ad for its new line of razors could target the topic Razors & Shavers. If the advertiser’s AdTech platform receives this topic from the Topics API, then it could show the advertiser’s ad to the user.

What role will AdTech companies play in the targeting and delivery of Topics-based campaigns?

It’s not clear at the moment how the Topics API will fit into the fragmented AdTech landscape, but AdTech companies will no doubt play a key role in helping publishers and advertisers run topics-based campaigns. 

One of the privacy considerations of Topics is to limit the number of third parties that can receive information when calling the Topics API.  

One of the privacy considerations of Topics is to limit the number of third parties that can receive information from the Topics API and to not expose more information than what is currently exposed with third-party cookies. For that reason, it may turn out that there will be a limited number of AdTech platforms involved in the process. 

However, AdTech companies will look to incorporate the Topics API into their client offering to help advertising and publishers run these types of campaigns.

Is Topics a privacy-friendly ad targeting method?

When it comes to ad targeting methods, the more privacy-friendly they are, the less information is known about the user. While most people in the industry believe that increasing user privacy is a positive thing, advertisers will still want to ensure their ad dollars are spent on showing their ads to their target audience. 

But if you can’t tell whether an individual is a member of your target audience, then how will you be able to show your ads to them?

This has been one of the main challenges the AdTech industry has faced ever since the privacy movement started gaining traction — i.e. creating privacy-friendly ad targeting methods that still provide value for advertisers and publishers. 

Topics API does offer more privacy protections than FLoC, but the question is whether it will be valuable for advertisers. 

By introducing the Topics API, Google wanted to address the negative feedback it received about the FLoC proposal and the fact that it didn’t provide enough privacy protections for users.

In terms of an ad targeting method, Topics does seem to offer better privacy protections than FLoC and those currently run via third-party cookies.

Topics API also offers some other privacy-preserving settings:

  • If a user is in incognito mode, then the Topics API will return empty topics. 
  • Websites can opt out of Topics. 
  • The amount of information available in the API will be limited to reduce the chances of a user being identified, especially across different websites.
  • In the future, users will be able to view the topics associated with them and change and delete them. They’ll also be able to opt out of Topics.

Will the Topics API be implemented into other web browsers apart from Google Chrome?

Ben Galbraith, Chrome’s product director, said in a recent press briefing that because the Topics API is in the early phases, Firefox, Safari, Edge, and other browsers haven’t stated whether they will adopt it or not. From the time FLoC was released to the time it was shut down, no other major web browser said that they would adopt it. So if history is anything to go by, then we can expect that other browsers likely won’t adopt Topics API.

“We’re sharing the explainer, which is the beginning of that process to discuss with other browsers their view on the Topics API, so time will tell,” Galbraith said.

When will Google Chrome’s Topics API be released?

The Topics API has only just been announced and the other Privacy Sandbox proposals are in various stages of development. Currently, the exact release date of Topics API is not set as there are many discussions and testing phases that will need to take place before it is released. The current timeline shows that the testing will last till Q3 2022. For sure, we can expect that the solution won’t be launched before 2023.

Below is a timeline that illustrates the progress of each Privacy Sandbox standard. Google updates the timeline once every month.

Source: Privacy Sandbox timeline

The post Google Chrome’s Topics API Explained + FAQs appeared first on Clearcode | Custom AdTech and MarTech Development.

This article was originally published on https://clearcode.cc

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.