Google’s new Bumper Machine whips up 6-second bumper ads from longer videos
The tool uses machine learning to take relevant moments from longer videos and automatically convert them into the shorter ad units.
Google has launched a Bumper Machine tool to help video advertisers create six-second bumper ads from existing video ads that run 90-seconds or less. Powered by machine learning, the tool will identify well-structured moments in a longer video, and convert those moments into multiple six-second video ads.
The process takes only a matter of minutes, and video advertisers can adjust the creative before saving the new bumper ads.
Why we should care
YouTube’s six-second unskippable bumper ads have proven to be successful in terms of brand awareness and ad recall. As early as 2017, YouTube reported that 70% of the 122 bumper campaigns it analyzed drove a significant lift in brand awareness, and nine out of ten of the campaigns drove ad recall, with an average lift of more than 30%.
For advertisers with limited resources, the Bumper Machine will save time and effort by taking their longer video assets and delivering multiple bumper ads.
“Bumper Machine enabled greater flexibility by allowing Grubhub to create captivating new video formats from our existing creative. As a result, we were able to generate short-form videos to leverage as part of our ongoing campaigns,” said a Grubhub spokesperson.
Grubhub, one of the brands with early access to the Bumper Machine, used the new YouTube tool to create the six-second video ad above.
More on the news
- The Bumper Machine technology identifies elements such as a brand logo, human faces, motion or contrast to automatically produce the short ads.
- YouTube first introduced bumper ads in April, 2016, offering them via the AdWords platform.
- This announcement comes ahead of Google Marketing Live, a two-day event to introduce new Google ad products that kicks off on May 14. Marketing Land editor in chief Ginny Marvin will be reporting from the event this week.
Opinions expressed in this article are those of the guest author and not necessarily MarTech. Staff authors are listed here.