Nearly six months have passed since Google reduced third-party cookie support for 1% of its Chrome traffic. Immediately after the support was removed, we at Relevant Digital began actively monitoring the performance of cookieless traffic and testing various alternative advertising targeting methods to see their impact on inventory value.
Three months ago, we shared our initial observations on testing the Topics API, and at that time, it was still challenging to identify clear trends. After a few more months and with more data collected, it is time to share updated findings. Generally, during this period, the use of the Topics API has somewhat increased, and more publishers have tested its performance.
This text focuses on the Privacy Sandbox's Topics API and shares insights regarding its performance on over 50 domains.
Privacy Sandbox
The Privacy Sandbox project aims to minimise the information shared between websites and advertisers and store only a large part of the visitor's information on the visitor’s device. Google’s project envisions targeted advertising and measuring conversions through Application Programming Interfaces (APIs) in a browser environment. A group of different interfaces has been designed, each meeting different needs.
Topics API
The Topics API categorises websites into broad, easily identifiable topics, such as "Sports", and tracks the most popular topics in your browser. This allows advertisers to display more precisely targeted ads while protecting your privacy by not revealing which sites you have visited.
With Topics, the Chrome browser learns what users are interested in when they visit different sites online. To understand visitors' interests, Google categorises sites based on a limited number of categories. The categorisation is currently designed to work only on a domain basis, i.e. the various topics on the news site will not be categorised separately. Up to this day, there are around 496 topics.
The data collected using the machine-learning algorithm is stored in the browser for three weeks and is not stored on an external server. When a visitor visits a site that supports the Topics API, Topics draws one topic each week, distributing it to target advertising. A shared topic category is either one of a visitor's top 5 topics from the last three weeks or 5% likely to be a random topic. Advertisers use these categories to target more relevant advertising to their visitors.
Check out more detailed technical information about The Topics API here.
General Conclusions on the performance of the Topics API
As mentioned, we have been testing the performance of the Topics API for just under six months. The test has included over 50 domains from our publisher clients. We have made the journey more accessible for our programmatic advertising experts by utilising the Relevant Yield HB Analytics module, enabling effortless reporting on the Topics API.
As we have collected data, analysed numbers, and gained insights, we have gained a new understanding of the efficiency and functionality of the Topics API. Based on the data, we can draw the following general conclusions:
Performance Variation: We have found that the impact of the Topics API on ad unit eCPM varies significantly between different domains. However, overall, its influence has been positive in 75% of the domains in our sample compared to the performance in traffic where the Topics API was unavailable.
Impact on eCPM: According to our data analysis, the Topics API has, on average, significantly increased the eCPM of ad units. Generally, we have seen eCPM rise to 50% or more in 40% of domains where the Topics API is in use. The best-performing domain in our sample increased by up to 172%.
Impact on Fill Rate: The Topics API's positive impact on eCPM is largely due to the fill rate, which has been higher in the inventory where it is in use. This is likely because better ad targeting attracts more advertisers.
Need for Further Research: The amount of data is still limited, and it is too early to draw comprehensive conclusions. It is important to continue testing and analyzing more data better to understand the long-term effects of the Topics API.
So, Testing is Important (You will hear this a lot this year)
This phase of exploratory testing is crucial. It’s less about immediate gains and more about collective learning, about giving feedback to Google and refining our strategies. Testing with tools like Relevant Yield, which gives you real-time analytics and a no-code UI for configuration, could save resources, making it less challenging to get started and maintain.
Aaaand… It’s Just 1%. For Now.
Deciding to phase out third-party cookies on just 1% of traffic is a double-edged sword. On one hand, it’s a wake-up call for programmatic sellers who’ve been on the fence about adapting their strategies. The time for hesitation is over; the need for action is now. However, this doesn’t mean diving headlong into the deep without a life jacket.
Google's staged approach offers a unique opportunity for measured experimentation. It’s a chance to dip your toes in the water with Topics API and other Privacy Sandbox initiatives. This phase allows you to test, learn, and iterate without the pressure of a total commitment. It’s about making informed decisions based on data and insights rather than scrambling in the dark.
Join the Conversation
The narrative of digital advertising is still being written, and your voice matters. Privacy Sandbox is far from perfect at this stage, so Google needs constant feedback to make it a better product that is fair for all players. That’s why everyone must test their theories! It’s for the common good of digital advertising and, most importantly, for yourself to be ready for whatever is coming.
For those eager to stay informed about these developments, our monthly newsletter is your gateway to the latest insights and updates. Join us, and let’s turn today's challenges into tomorrow's opportunities.