During the earlier part of 2017, Google launched beta test that automatically launched machine-generated ads in a small set of AdWords accounts. The lack of control in that test wasn’t particularly well-received by many. Google has chosen to try again, but this time, there are some adjustments that are aimed at addressing advertiser concerns.
There were a small number of US and UK advertisers who received emails about the ad suggestions better last week. Similar to the first test, the system takes into account several factors when generating new ad variations, such as content in existing ads and ad extensions and landing pages to identify ad content that could drive the best ROI.
So what’s different with this test?
- More notification: Participating advertisers will receive notifications in the interface and via email whenever an ad variation is created in an account.
- Review period: There will be a 14-day period in which advertisers can review ad suggestions before they go live. During that review period, the advertiser can choose to edit, pause, dismiss or approve it. It’s also possible to opt out of the test altogether at any time.
A Google spokesperson had this to say, “we always listen to our customers. We heard their concerns around the initial experiment and so we have developed this updated program to reflect those concerns.”
This test is going to affect a handful of ad groups with a small number of ad suggestions per account. Google recommends that participants let the ad suggestions accrue enough impressions before making any changes.
The point behind the test is that Google has been consistently stating over the past several months that ad groups usually garner more impressions and clicks when they have three to five ads, instead of only one or two. This automates the buildup of ad variations in an ad group.
The idea is to also get a better understanding about what makes users click. This isn’t a shock, since surprise, as machine learning now underpins all new development in AdWords, and is at the core of this test. When the models gain more feedback about the sort of ad content that drives more clicks, Google will be able to adjust and improve the suggestions.
There is no specific tie to the change to ad rotation settings announced last week in which the options will be limited to be limited in order to “Optimize” or “Don’t optimize”, but they’re basically cut from the same cloth. Ad groups set to optimize are going to be eligible for this test. Advertisers are going to have to monitor conversion performance, but the test is focused on identifying ad groups where there is the greatest opportunity for improving ROI and use machine learning to automate the ad creative development and ad testing processes.
The beta is set to begin September 30. If you would like to partake in it, and you’re in the US or UK and haven’t been notified, reach out to your Google rep about it.