Attribution 12 min read

What is Google’s Topics API – and is it Really a Replacement for Cookies?

By QueryClick 11 May, 2022

When Google announced back in early 2020 that they intended to phase out support for third-party cookies it was clear that the advertising industry was in line for some fairly fundamental changes.

The decision by Google was driven by a range of factors from the increasing need to protect consumer privacy – to legislation including GDPR, the ePrivacy Directive and CCPA – and a fundamental need to rebuild trust with consumers.

For marketers, the decision has left even more confusion around what the removal of third-party cookies really means for them and how they can plan for post-cookie marketing success.

So, without doubt, you need to have a plan in place for the change. But what are Google proposing as a replacement for cookies? And what, in particular, is Google Topics?

Let’s dive into those “topics” below…

Google’s solution to replace cookies is evolving

The first thing to point out is that Google’s approach to proposing a solution to effectively replace third-party cookies has been iterative.

The Privacy Sandbox initiative

Back in 2019, Google announced the launch of its Privacy Sandbox initiative aimed at developing a set of open standards to enhance internet privacy.

With the clear intention of working with industry to build new internet privacy standards.

Federated Learning of Cohorts (FLoC)

One of the early proposals to come out of the initiative was FLoC.

In essence, FLoC was developed to provide a solution that delivered an increased level privacy relative to existing cookie-based solutions.

It does this by bundling consumers into specific cohorts based on their browsing history which is used to identify specific buying interests. With a view to targeting them on this basis.

However, the industry reacted negatively to the move as there was concern that the depth of data available still allowed advertisers to identify individuals. For example, combining the underlying FLoC data with personally identifiable information enabled ‘fingerprinting’.

This method potentially allowed individual user profiles to be built up – along with possible issues with existing data privacy laws including GPDR. So, ironically a solution that was aimed at increasing privacy was deemed to be erring toward exactly the opposite.

The upshot? Google is effectively abandoning FLoC.

So, Google Topics is an attempt to redress the balance towards increased privacy here. In fact, critics argue that it shifts the balance back too much in this opposite direction.

What is Google Topics?

Source: Google

Google Topics is an API that marketers can use to deliver targeted advertising in a way that protects and respects the privacy of individuals.

How does it work?

At a basic level, as a user visits websites the Chrome web browser gathers information about the topics of the websites.

For example:

  • “autos”
  • “dogs”
  • “men’s clothing”

Each week Google Topics associates five of these topics to the user – including a random ‘wildcard’ topic that is designed to increase user privacy. And advertisers are able to access up to three topics for the site visitor to use for advertising purposes.

Each topic is kept for a period of three weeks and can be used to deliver relevant advertising as the user revisits the site.

What are the pros and cons of Google Topics?

One caveat in all of this is that we are making assessments that are very fluid – things are evolving on a monthly, if not weekly, basis.

So, our analysis here is based on what we can glean right now and is likely to change as more information emerges.

So, what are some of the potential pros and cons of Google Topics as we understand things right now?

Pros of Google Topics

Simpler structure

At the most fundamental level, Google Topics will be simpler for marketers to understand, evaluate and implement than FLoC.

And it has the potential to improve things for advertisers and consumers alike.

Better data protection that FLoC

We touched on it above but it is worth re-iterating here. Google Topics provides better data protection than was being proposed under the previous FLoC proposal.

The balance in all of this is to meet heightened expectations from consumers around privacy and the needs of marketers who are hungry for the type of data they need to target messages effectively. And Google Topics potentially does a better job of this than FLoC did.

Cons of Google Topics

Concerns that categories in Topics lack sufficient depth

This is a fairly central concern for advertisers (and potentially ad-tech providers alike).

If we rewind to the 30,000 plus classification groups that were originally being mooted with FLoC – and also consider the depth and flexibility to targeting that is available via third-party cookie driven solutions today – then only 350 categories in Google Topics feels narrow in anyone’s language.

So, let’s take the example of a specialist coffee provider who currently has quite a degree of accuracy in terms of their ability to target prospective buyers. Under Google Topics that is reduced quite markedly.

The closest topic that you would be able to associate your bid against in Chrome is not single origin, coffee or speciality coffee.

It’s not even coffee.

It’s food and drink.

Not a close match we think you will agree.

An example of targeting using Google Topics

It’s worth noting again here that it is early days here in terms of the classifications but as things stand it certainly looks restrictive.

The 3-week time period restriction on data could be an issue

The lack of real-time data could be problematic for advertisers – many of whom have become accustomed to using up-to-the-minute data to adjust and refocus campaigns ‘on-the-fly’.

Because advertisers only receive the data on users once every three weeks, it is entirely possible that ads will be reflective of where the user ‘was’ in terms of buying intentions some weeks back and not where they ‘are’ right now.

For example, a user who three weeks ago was in the market to buy a new motorcycle may well have already made that purchase and may have changed their search focus to buying a helmet and leathers – but the ads being served would still be for the motorcycle due to the data timing disconnect.

Randomisation in segments has a cost

As mentioned previously of the five topics that are allocated to a user on a weekly basis, one will be randomly generated with a view to protecting user privacy.

But what does this mean in practice?

Essential it means there is a 5% chance that your data will include a classification that bears no true relation to the browsing behaviour being shown by the user. Which has clear implications for advertising waste and a degradation in the personal experience of the end user – not to mention the knock-on effect on analysis of performance.

No guarantees around adoption of Google Topics at scale

This is an interesting one.

Judging on the initial response of other browsers like Firefox, Vivaldi and Brave to the FLoC proposals – their opposition was vocal and no other browser adopted the technology – there is no cast iron guarantee that they won’t follow suit on Google Topics.

Throw in the fact that the Apple iOS 14.5 release shows Apple appear to continue to be putting user privacy at the forefront of its plans. And it certainly raises concerns that Google won’t be able to rely on widespread adoption of Topics.

Depending on which figure you believe – Google has somewhere been 64-67% of the global browser market. But driving penetration of Google Topics past this segment of the market may be challenging based on industry indications to date.

What is the alternative to Google Topics?

It is clear from our analysis above and discussion around the industry that there are very real concerns around the proposed solutions that Google has for replacing third-party cookies.

At QueryClick, we have long had concerns over the effectiveness of cookie-based solutions.

In fact, our own work with clients has shown that as much as 80% of the data generated by solutions that rely on third-party cookies like Google Analytics, GA360 and GA4 is actually incorrect.

This was the driving force for the development of our own Corvidae solution which features the pioneering use of AI and Machine Learning to effectively replace third-party cookies.

If this gives you food for thought – and you are concerned about the third-party cookie replacement solutions now being proposed by Google – then you may want to visit our website and learn more.

Is cookie-free attribution a myth?

Download our eBook to find out.

Identify wasted spend with our ROI calculator

Own your marketing data & simplify your tech stack.

Have you read?

See all articles