Google Removes Fewer Bad Ads in 2018 by Targeting Root Causes
Google removes 2.3 billion bad ads in 2018, marking a 28 percent decrease from the previous year. The company attributes this drop to improved machine learning that targets bad actor accounts before scams spread.
Google reports that it removes 2.3 billion bad ads in 2018, which reflects a 28 percent decrease from the 3.2 billion ads taken down in 2017. Instead of simply reacting to individual policy violations, the company shifts its focus to stop bad ad experiences before they ever reach users. This approach results in Google taking down fewer than 75 bad ads per second, compared to over 100 per second the previous year.
The tech giant relies on improved machine learning technology to tackle the root of the problem by terminating bad advertiser accounts rather than just deleting individual ads. In 2018, Google bans nearly 1 million bad advertiser accounts and almost 734,000 publishers and app developers, which doubles the account takedowns seen in 2017. The company also deploys 330 new detection classifiers to catch policy violations at the page level across 1.5 million apps and 28 million web pages.
Beyond its internal technology upgrades, Google collaborates with cybersecurity firm White Ops and the FBI to dismantle 3ve, a massive international ad fraud operation. This coordinated takedown stops a network that generates over 3 billion daily bid requests through counterfeit domains and malware-infected computers. To further protect users, Google introduces 31 new advertising policies throughout the year to address emerging abuses in the digital advertising landscape.