• by Laurie Sullivan , Staff Writer @lauriesullivan, January 17, 2025

Fake locations and illegal listings have become a problem for Google Business Listings (GBL), Near Media co-founder Mike Blumenthal said during an interview with Brad Wetherall, former director-support operations at Google.

Wetherall now runs a consulting business called Bitwise to help businesses and agencies with some of the challenges such as suspensions, verifications, and deleted reviews. The two were joined by Near Media co-founder Greg Sterling to talk about AI, algorithms, and the future of automating systems including customer service.

Overwhelming demand to resolve problems across GBL forced Google to rate each based on trust scores and other factors. Wetherall said the speed with which these problems were and are resolved relies on a voting system. The higher the trust score, the faster the problem gets resolved.

“If you go to a business that you believe to be fraudulent, and you say, ‘hey I think this is fraudulent,’ that’s one vote,” he said. “If you have product expert status or you’re a local guide level seven, that might count for five votes.”

If someone issues a fraud report to Google while standing in the location that the business should be, but it isn’t there, the report carries more weight than if it was sent from a faraway location because the algorithm identifies the IP address.

AI and algorithms have become a major focus for businesses dealing with everything from fraud to customer service.

Google’s support team has undergone an evolution throughout the years. Now it has stepped up its use of AI to automate the process.

Wetherall said that in 2020, the team tried to handle more than 3 million inquiries. Early on, Google tried to automate resolutions through email. Replying to the email would lead the person to a human response. Another option was to push people to documentation on the issue in the Help Center.

Google began experimenting with a more structured flow for customer service around 2020, Wetherall said. At the time the company had run in to a “massive backlog.”

During Covid, getting people headsets and other equipment to work from home was a logistical challenge.

“We would attempt to automate 60% [of queries], and we were successful at automating half of that,” Wetherall said. “We fully automated about 30% of our entire support base. The other half received replies and went to a human agent. We really didn’t count that as being successfully automated because it still went to a human.”

The motivation, he said, was trying to improve AI to automate a larger percentage of inquiries. AI agents should improve the process, but that’s not something Wetherall said.

Suspensions, verifications and reviews are typically the biggest challenges with search marketers. A couple of years ago there was a massive spike in suspensions, Wetherall said, driven by an algorithmic change to the suspension model. It created another backlog, and happened as a result of a bug in the code.

Increasing the safeguards in the algorithm will sometimes catch more fraudulent businesses, but it can also increase the amount of good business getting caught in the mix.