What Advertisers Need to Know About Surges in Online Hate Speech

Hate speech content, which includes biased or derogatory language, nearly tripled after the death of George Floyd, according to new data from DoubleVerify.

Appearing alongside hate speech can have a real and lasting impact on customer attitudes and purchase behaviors. A recent DV/Harris Poll shows that 82% of consumers say it’s a brand’s responsibility to make sure their ads appear in appropriate environments, and two-thirds of consumers would likely abandon a brand they see next to questionable content.

By being aware of inflammatory content trends, advertisers can define a brand suitability strategy and take action with brand safety tools.

The Correlation Between Real-World Events and Inflammatory Content

Hate speech and fake or inflammatory news both see spikes coinciding with real-world events such as race-related controversies, violence, hate crimes and, recently, highly publicized political races. Following the 2016 U.S. presidential election, DV created the “Inflammatory News and Politics” (INP) category, which focuses specifically on identifying and avoiding fake news. And recently, after the Iowa Democratic caucus, data from the INP category revealed fake news surging in New Hampshire.

During real-world events that perpetuate fake news, DV’s technology quickly categorizes and protects advertisers from running on these sites. Since May 26th, DV has successfully prevented millions of ads from appearing adjacent to hate speech, limiting the monetization opportunities for sites that traffic in hateful rhetoric.

How DV Identifies and Categorizes Inflammatory Content

When making classifications, DV uses sophisticated approaches that rely on a combination of AI and comprehensive human review. To categorize INP content, the semantic science team analyzes websites by looking at structure, language, rhetoric and associations with other sites known for perpetuating fake or hyper-partisan news. Then, AI helps scale these insights to make classifications in real-time. To determine if content qualifies as hate speech on articles, blog posts, user-generated content, in-article comments, videos and more, DV’s AI makes real-time decisions. As new content and topics emerge, DV’s semantic science team performs additional classification updates.

What the Data Says About Content Trends

The data from DV’s categories gives insight into content trends that can help inform a bigger picture. During the pandemic, but prior to George Floyd’s death, hate speech decreased. After George Floyd’s death, when the protests began, spikes in hate speech became a national trend; at minimum, each state saw their rate of hate speech double.

At a Glance: Geographical Trends in Hate Speech and INP

• Minnesota, where the protests began, had the second highest spike of online hate speech.
• Like hate speech, every state has seen an increased rate of INP.
• Nationally, INP content has risen by 38%.
• INP content increased the least in New Hampshire, by only +5% its own average. And it increased the most in North Dakota, by +82% its own average.

How Can Advertisers Avoid Inflammatory Content?

Advertisers can protect their brand while continuing to support news by working with a verification partner that uses sophisticated, nuanced tools to identify inflammatory content. DV’s brand suitability offering safeguards advertisers while ensuring ad dollars go toward legitimate news publishers — not deceptive sites that promote hate speech, disinformation or fake content.

To learn more about the rise of hate speech, read “Hate speech has soared online since George Floyd’s death.”