NextRoll Creates FLEDGE Test Lab

How a NextRoll Data Scientist is Preparing Us for the Digital Advertising Future


The marketing technology industry is constantly evolving and innovating, so it’s vital to remain at the forefront of navigating and participating in these changes. At NextRoll, we’ve worked closely with the World Wide Web Consortium (W3C) right from the start and contributed to conversations around standards for the web. 

In 2020, when Google announced its intention to stop supporting third-party cookies in its Chrome browser – following Apple and Mozilla doing the same on their browsers – a large focus in the industry became about balancing preserving data privacy with delivering targeted ads to users across the web. More about this in our Rebalancing Privacy and Precision on the Open Web blogs.

To balance privacy needs with advertising needs, Google published its First Locally-Executed Decision over Groups Experiment (FLEDGE) proposal that addresses retargeting and custom audience use cases with the future deprecation of third-party cookies. The Engineering team here at NextRoll contributed to FLEDGE, along with other MarTech leaders. And since its publication, we’ve dedicated a lot of time and effort to digging into the available Google FLEDGE APIs. 

Marco Lugo, a Sr. Data Science Engineer at NextRoll, has been at the forefront of these efforts. At the beginning of 2022, he built a test harness that allows our team – and others, since it's open-sourced – to “play with the FLEDGE API” in an automated fashion. This is a huge accomplishment, as Marco has uncovered learnings from the tests and shared them with the Google Chrome team.

“Marco is one of the most knowledgeable engineers at NextRoll and in the industry when it comes to proposals in the Privacy Sandbox,” said Sr. Data Science Engineer Matt Wilson. “He drew from a deep well of technical knowledge when he created FLEDGE lab, a way for engineers to get hands-on experience with the new privacy-preserving Google Chrome APIs.” 

His creativity and hard work around creating the FLEDGE test harness won him an AFT Award – an internal NextRoll award program that awards individuals or teams that have made significant contributions towards NextRoll's goals – in the Future-Proofing category!

Marco's examination contributes to the debate around what we should expect in resource usage from FLEDGE and what algorithms we may employ to help the winning bids be selected more often. And since this test harness is open-sourced, other industry leaders can confirm results, extend its capabilities, and further the community's understanding of how we can expect FLEDGE to function.

“Creating FLEDGE lab will enable other industry stakeholders and us to quickly experiment with privacy sandbox APIs, allowing us to give Google better feedback on them,” said Matt Wilson. “In the end, this will make the APIs better and more useful to us!”

The test harness is also more valuable than just reading the FLEDGE documents themselves, as Engineers can see when and where source registration occurs. 

“Marco has really helped advance our understanding of the FLEDGE API by pushing it to its limits,” said Sr. Director of Data Science Engineering Andrew Pascoe. “His analysis of his results demonstrated some edge cases in FLEDGE that could lead to less-than-optimal auction outcomes; we took this back to the W3C for commentary and feedback. In addition, he open-sourced the test harness he wrote to get these results, which will enable the industry to do similar things more easily in the future in a standardized way.”

There’s plenty of more work to be done regarding creating the optimal environment for users and advertisers on the open web, and the NextRoll team is ready to be a leading voice along the way. 

If you’re interested in viewing or using the FLEDGE test harness, visit https://github.com/AdRoll/fledge_lab.

And if you want to join our innovative Engineering team leading these efforts and more, check out our Job Board!