Automation is nothing new in AdWords, but this month, Google launched a pilot that adds new text ads to advertisers’ accounts. Dubbed Ads Added by AdWords, the program started on January 26.
Not surprisingly, this news has set off alarm bells among paid search managers that worry about Google usurping control over the ad creation and testing process. Here is what we know so far about this test.
The initial set of advertisers were notified of the pilot on January 12. For those that chose to participate, ads were added to ad groups two weeks later, on January 26, at which time a second wave of advertisers were notified about the pilot. Currently, 2,000 accounts have been selected for the test. Each has a two-week opt-out window via a form. If you do not receive an email, you haven’t been selected for the pilot.
What accounts were considered for this program? Google looked at campaigns with ad rotation settings of either “Optimize for clicks” or “Optimize for conversions” that have ad groups with few ads in them.
If you’ve opted out of automated extensions or are in a vertical with privacy sensitivities such as pharma, your account was not selected for this program.
How are the ads generated? We’re told that for the test, the ads were generated by people (as opposed to auto-generated) based on the existing ads in the account and the landing page content. The ads went through review by the product team, among others, for quality assurance. The sales teams were also involved in creative review and account selection for the pilot.
From the Help Center page on this new program, we also know that any ads generated for the pilot will be labeled “Added by AdWords.” In the example below (yes, all of the ads are terrible, but try to look past that for now), Google has added two test ads in an ad group that had just one ad. Notice that the headlines, description and paths are all being tested.
Google says on that Help Center page, “We believe that adding more ads to the affected ad groups can improve these ad groups’ performance by 5 to 15%.” The new ads are set to run indefinitely, and Google recommends pilot participants not pause the ads. Theoretically, if they perform worse (based on conversion or click-through rates), the ads will be shown less. But certainly, review the ads if you’re participating in the test, as Google also advises.
This program obviously raises more questions about advertiser control and the role of machine learning in ad creation. If Google deems the pilot successful and rolls out Ads Added by AdWords more broadly, it’s hard to see how the current ad creation and vetting process can scale without automation. One can assume that the machines will be learning from this pilot.
If you think you might miss the notification email from Google about this test, Frederick Vallaeys has written a script for MCC accounts will send an email if it finds automatic ads: https://gist.github.com/siliconvallaeys/a42e247d7923a3181b0a9eb7332fdbad