Last Updated on: 2nd February 2024, 12:47 pm
It’s time for our 4th Media Review. This month, we selected 5 stories, such as an update from Chrome on the progress of the Privacy Sandbox and an analysis of the future of reporting.
Table of Contents:
- Progress in the Privacy Sandbox (September 2021) – Google Chrome
- What is an Identity Strategy – AdMonsters
- User-Agent Reduction Origin Trial and Dates – Google Chrome
- The future of attribution is data-driven – Google Ads
- Yahoo! JAPAN’s Feedback of [Conversion Measurement API] Origin Trial – Github
Progress in the Privacy Sandbox (September 2021) – Google Chrome
= In the blog post accompanying the update to the Privacy Sandbox Timeline, the Google Chrome team provides information about the progress in the development of proposals from each of the groups:
= Prevent covert tracking – a more detailed timeline for User-agent string reduction was published (we covered this part in this Media Review down below)
= Strengthen cross-site privacy boundaries – the Origin Trial of First-Party Sets (proposal allowing to exchange information between websites of the same owner) has concluded and the work continues based on the feedback
= Show relevant content and ads – the discussion period for proposals from this group, FLEDGE, and FLoC has been extended to Q4 2021, which means the testing stage will be postponed until Q1 2022. As it will retain its duration, it will end in Q3 2022, not Q2, as previously assumed
= Measure digital ads – the Origin Trial for Attribution Reporting API, a proposal enabling to measure the outcomes of advertising, has been extended after the Chrome team received feedback from the industry (we covered Yahoo! JAPAN’s feedback in this Media Review down below)
= Fight spam and fraud on the web – the Origin Trial for Trust Token API, a proposal allowing sites to detect if the visitor is human, has been extended to Chrome version 101
What is an Identity Strategy – AdMonsters
= Wenda Zhou from AdMonsters emphasizes in her article that Google’s extension of the timeline for cookies should not stop the industry from planning for the future. She proposes four steps for publishers to take:
= Understand the identity options and determine where to focus – naming Google Privacy Sandbox, publisher first-party data, third-party IDs and contextual as the key groups of methods
= Ensure your SSP has a digital identity strategy – while it’s clear that demand-side platforms will be leveraging available tools, it’s just as important for SSPs to support them so that publishers do not lose out on advertiser spend
= Develop an identity testing strategy – with so many methods available, it’s important to develop a way to accurately test and measure their efficacy to make an informed decision on choosing the right partner
= Seek out new opportunities – the phase-out of cookies brings many opportunities for publishers to take advantage of, such as the possibility to better leverage their first-party data, to strengthen direct relationships with valuable partners, monetize browsers – which already banned cookies – and become closer with their audiences to understand what they actually expect
User-Agent Reduction Origin Trial and Dates – Google Chrome
= After announcing the intention to reduce the User-Agent String, an important piece of information about the user, Google published a more detailed timeline of when it is going to happen, measured in Chrome versions rather than dates
= The first stage will be Origin Trial which will allow for testing and feedback regarding the reduced UA string, which will last for about 6 months (Chrome 95 to Chrome 100)
= The second stage will be the roll-out of the reduced UA string, starting from the Desktop (Chrome 107) moving on to Mobile (Chrome 110)
= The third stage will be the completion of the reduction, which is expected to happen in Chrome 113
= Despite the whole process taking a while, as it is expected that it will end in Q2 2023, it is expected to significantly impact the efficiency of fingerprinting mechanisms aiming to subside third-party cookies, as it was described by Zach Edwards on Twitter
The future of attribution is data-driven – Google Ads
= The Google Ads team announced that data-driven attribution will become a default option in their Ad Manager from now on, replacing last-click attribution
= This approach, aiming to model conversions using machine learning algorithms, allows to fill gaps when user identifiers are not present, which will become a common scenario after cookies are phased-out
= Google claims that this new approach gives more accurate results by analyzing many different touchpoints that led up to a conversion
= This move has been expected by the advertising industry, because the diminishing possibility of receiving user-level signals requires adaptation and an increased use of probabilistic attribution methods
Yahoo! JAPAN’s Feedback of [Conversion Measurement API] Origin Trial – Github
= One of the key proposals for reporting and attribution purposes is Conversion Measurement API, recently renamed to Attribution Reporting API. The team from Yahoo! JAPAN published an extensive feedback report based on the data of the proposal from Origin Trial, up until July 15, 2021
= The company claims to have received more than 200 thousand conversion reports in total, finding out that:
= Around 10% of the conversion reports were lost in the delivery
= Based on the three deadlines of the reporting window from the impression, 30% of the reports were received after two days, a total of 58% after 7 days, and the rest after 30 days
= 5.13% conversion data got noised, in comparison to 4.375% that was supposed to be noised based on the theory
= 17.13% of the reports were discarded, as they would indicate more than 3 conversions per impression, which is forbidden based on the proposal’s specification
= There may be an issue with falsely attributed conversions between different services under the same effective Top-level domains (eTLD+1)
If you have any questions, comments or issues, or you’re interested in meeting with us, please get in touch.
