“We’re seeing a big move toward automation and machine learning in 2020,” said Matthew Mierzejewski, SVP search capability lead at Merkle. “Those were buzzwords for the past few years, but now we’re starting to see new product.”
As an example, Mierzejewski pointed to auction-time bidding in Search Ads 360, a smart-bidding feature that analyzes contextual signals at the same time of the action to set the bids.
Brands simply need to trust that Google's automation can make the best decisions and accomplish the best job.
Google handles the search query, understands the intent, the user and the context before deciding how likely it is that the consumer will take a favorable action such as buy the product. They do this knowing keyword mapping, time-of-day settings, reading remarketing list tags, and more.
The more automation, the less brands understand the reasons behind bidding decisions. “This is where Google’s black box gets darker,” he said. “The advertisers' understanding of individual ad auctions gets more opaque. The more automation, the less advertisers will understand the 'why'."
This is true of all artificial intelligence, he said, because advertisers have less understanding of the automation that made the decision.
Trust also plays a factor, he said. It’s difficult to trust a platform when bids are lost or return on ad-spend falters. It’s complicated to gain the measurement and data signals to pass through to Google, Mierzejewski said.
“Bad inputs equal bad outputs,” he said. “It’s a focus primarily on search, but there are elements in Google’s Floodlight tag on DoubleClick ID that will span search and display.”
He said the auction-time bidding is specific to search, but if advertisers can get that right and get measurement correct on the floodlight element, hey can expand automated strategies into display on the Google network.