Results from our display ads experiment with the Privacy Sandbox APIs

July 22, 2024

Google’s ads teams are committed to providing advertisers with effective and privacy-centric tools and have been conducting rigorous testing of the Privacy Sandbox APIs, in combination with other signals, to understand how the APIs can support ads performance without third-party cookies.

Between January and March 2024 we conducted an experiment on a portion of display ads traffic in Google Ads and Display & Video 360. We evaluated the effectiveness of measurement and audience strategies using a combination of Privacy Sandbox APIs and other privacy-preserving signals, in comparison to third-party cookies. 

The results serve as directional estimates at a fixed point in time, with limited ad traffic and inventory available for testing, and should not be considered a final reflection of Google’s display ads performance without third-party cookies. We expect performance to continue to improve as we keep innovating our entire suite of ads products including privacy-preserving technologies, Google AI and first-party data solutions. As part of this broader roadmap, we plan to expand our platforms’ Privacy Sandbox usage, which will help Google AI optimize for better results. Further improvements are also expected as the broader industry adopts the Privacy Sandbox, so we encourage more ad technology partners to test and integrate the APIs. 

What we learned 

The experiment indicates that using the Privacy Sandbox technologies (Topics API, Protected Audience API, and Attribution Reporting API) in combination with other privacy-preserving signals on our display ads platforms can contribute towards advertisers’ performance recovery without third-party cookies. The experiment showed:

  • Scale preservation: For advertiser spend—a proxy for scale—we saw an 89% recovery in Google Display Ads and an 86% recovery in Display & Video 360.1
  • Return on investment (ROI): For campaigns focused on conversions only, we saw a 97% recovery in conversions per dollar (CPD)—a proxy for ROI—in Google Display Ads, and a 95% CPD recovery in Display & Video 360.2
  • Remarketing recovery: Across campaigns using only remarketing audiences, our experiment showed a 55% advertiser spend recovery in Google Ads and 49% in Display & Video 360.3 These results are likely because remarketing today is more reliant on third-party cookies which enable a highly precise level of ads personalization, and the eligible inventory is limited because few supply side platforms (SSPs) are currently testing the Privacy Sandbox. In terms of CPD, we saw some outliers that we’ll continue to monitor. The recovery was generally better for campaigns that use a combination of remarketing audiences and other strategies like in-market-, affinity audiences, and optimized targeting.4 This suggests that advertisers may benefit from adopting a combination of audience solutions for customer re-engagement. We encourage you to read the whitepaper for more details.

Looking ahead

We will continue iterating on our API integration in consultation with the Competition and Markets Authority in the UK, and share regular feedback on the Privacy Sandbox APIs as they evolve.

With the rapid advancements in AI, first-party data solutions, and privacy-preserving innovation, marketers have the tools to be ready for a new era of growth.

Posted by Google's display ads teams

 

 

1. Google Data, Global, Display ad platforms, 1/29/24-3/10/24
2. Google Data, Global, Display ad platforms, 1/29/24-3/10/24
3. Google Data, Global, Display ad platforms, 1/29/24-3/10/24
4. Google Data, Global, Display ad platforms, 1/29/24-3/10/24 

Was this helpful?

How can we improve it?
true
Subscribe to our Best Practices newsletter

Sign up for the Google Ads Best Practices newsletter today to get actionable tips, how-to’s, and the latest product updates delivered right to your inbox.

Subscribe here

Search
Clear search
Close search
Main menu
15865837040732299937
true
Search Help Center
true
true
true
true
true
73067
false
false
false