Google Announces Results from Initial Cookieless Ad Testing (with Cookies)
04/19/2023
Yesterday, April 18, Dan Taylor, VP of Global Ads at Google posted results from initial in-market performance tests of ads served with Google’s new interest-based audiences (IBA). The IBA is Google’s new privacy-driven technology to replace the 3rd party cookie (3PC), and it relies on the use of anonymized browsing history placed into interest groups for ad targeting.
Taking a step back, Google has rolled out a larger privacy initiative called Privacy Sandbox covering consumer data privacy efforts across its Chrome browser, its Android mobile OS, and other related web privacy and safety technologies. Within the Privacy Sandbox, Google announced that Chrome will no longer support 3PC, and while Google pushed back the go-live date of this cookie kill-off to later this year, the company has been testing various aspects of its Privacy Sandbox, including the test results released yesterday covering a test of video and display ads using the new IBA approach.
The test was interesting for several reasons. First, they designed the test to compare ad performance across a number of KPIs between the current approach in the market which uses 3rd party cookies and the new IBA technology which serves ads based on anonymized interest groups (AKA “TOPICS”). Because the test was a randomized control test, it begins to show the market what it can expect once Google finally kills off the cookie. Secondly, Google tested both Google Ads as well as Display and Video. Lastly, and somewhat ironically, the test measured ad performance using 3rd party cookies, which have been shown to have significant measurement accuracy issues. More on using cookies to measure cookieless ad performance is broken out below.
So, what were the results? Here’s a summary (Note: for a more complete breakout of results and information about the test design, see the project page here.
| Test Performance Criteria | Performance of Test (IBA) vs. Control (3PC) | Our Thoughts and Notes |
|
Spend |
Decreased by 2-7% |
Without 3rd party cookies, the consumer interest signals are lost to some degree and therefore fewer campaigns match an impression, thereby lowering spend. |
|
Conversions per Dollar Spent |
Decreased by 1-3% |
Conversions would expect to be weaker if targeting precision drops with the use of IBA. |
|
Click-Through Rate (CTR) |
“Results within 90% of the control” |
In other words, clicks dropped as much as 10%. |
Using Cookies to Measure Cookie-less Ad Testing:
In the detailed test methodology documentation was buried an interesting tidbit: the test relied on 3rd party cookies to measure the performance of Google’s new cookieless ad targeting. We can only speculate as to why this would be as the test documentation didn’t provide an explanation. There may be a few possible reasons to use cookies to measure results:
1. The most obvious is that the user identity isn’t known at the time the ad is served. This of course is by design and is what Google is testing. But because of this missing identity, there’s nothing to match the conversion and ad served– other than a cookie.
2. This also implies that Google’s own privacy-centric measurement protocol isn’t ready yet, although it has been in two origin trials in 2022. You can read more about the Attribution Reporting piece of the Privacy Sandbox here.
3. It could also mean that Google wanted to show the strongest possible results for the new IBA approach, and one clue is that they also included Retargeting results in the conversions. And how did they conduct retargeting? You guessed it: by using 3rd party cookies. And this could be evidence that Google’s “FLEDGE” retargeting protocol isn’t ready for primetime yet either. NOTE: Google also recently renamed FLEDGE to “Protected Audience API”.
What to Make of This?
The testing, and the public release of test results, would indicate that Google continues to march down the path of the 3rd party cookie deprecation and is moving into more public testing of their Privacy Sandbox initiatives. It also should be further evidence to marketers that they need to take this pending change seriously and begin to make changes to future-proof their own marketing technologies– including marketing attribution and overall measurement.
Want to learn how OptiMine can help future-proof your marketing measurement? Drop us a note and we’d be happy to evaluate your current state and determine how we can help your brand succeed in this huge privacy-driven transition.