Google removes 2.3 billion bad ads in 2018

Google today released its latest so-called ‘bad ads report’, in which it highlighted that 2.3 billion ads were taken down across all its products, including YouTube, in 2018.

In addition, 734,000 publishers and app developers were booted from Google’s ad network, and advances in machine-learning technology meant the company could take more granular action to remove ads from nearly 28 million web pages found to violate its publisher policies.

The global report does not contain breakdowns for Asia-Pacific, but Amanda Storey, APAC director of Google’s go-to-market team, said the trends around bad digital advertising and actors were consistent around the world.

“This stuff is hard, but we’re using machine learning to identify bad actors at the source,” she said, meaning Google could terminate more bad ad creators as opposed to just their ads. Storey cited this as the reason why in 2017 Google removed 3.2 billion ads, compared to 2.3 billion in 2018.

In response to a question about Google’s current issues around unsafe YouTube content and advertising appearing against it, a Google spokesperson clarified that the report addresses bad advertising on Google’s platforms, not the content on them.

Storey broke down Google’s process for determining what constitutes a bad ad:

• Ad creative: Is it doing what it’s meant to, and technically working how it is supposed to?
• What’s being promoted in the ad­­? Is it leading to a bad offline experience, such as diverting you to opaque ticket resellers
• Landing pages­: Is the ad referring you to a landing page that is real?
• Traffic: Is the ad driven by invalid traffic or bots?

As part of Google’s efforts, the company also introduced 31 new ad policies in 2018, primarily addressing abuses in specific sectors, such as ticket resellers, cryptocurrencies, third-party tech support and bail bond providers.

Regarding misinformation, Google said it removed ads from almost 15,000 sites for “violations of policies directed at misrepresentative, hateful or low-quality content”. Furthermore, around 190,000 ads were taken down for violating Google’s “dangerous and derogatory” content policy. Advertisers do have the ability to appeal if they believe they have been unfairly removed, Google confirmed. 

Storey admitted determining whether an ad was ‘hateful’ was a constant challenge, and that Google used both technology and teams of reviewers, working together with experts and adhering to local laws, in tackling this issue.

In Asia-related news, Google said it has launched an election ad verification tool in India ahead of next month’s general election, similar to the tool launched in the US in 2018. A Google spokesperson said the company worked with local lawmakers to set up a commission in India to certify advertisers, and created a library of verified ads, which provides details in some cases on who has purchased the ad. Similar action is also being taken in Australia ahead of its general election this year.