The finding, published by hate-fighting group Anti-Defamation League on Friday, is intended to force Facebook to do more to police hate and harassment on its service. The organization has long complained that the company has merely payed lip service to cleaning up racism and anti-Semitism.

The ADL report comes a day after Facebook for the first time detailed the scope of the hate speech on its service. The company said that only 10 to 11 of every 10,000 posts that users view contain hateful content.

But the ADL said that fails to take into account the algorithms that can amplify exposure to hate speech for users who are already searching for it. And looking at Facebook’s data in another way shows that the problem is widespread rather than an aberration, the group said.

“Weekly exposure to hate speech thus could be as high as 10 billion views, monthly exposure, 40 billion views, and yearly exposure, 520 billion views,” the ADL said in a blog post. “These are startling numbers and they demand an urgent response.”

About the ADL report, Facebook told Fortune that its new metric provides a broad look at overall levels of hate speech, but that they shouldn’t be used to make assumptions about the experience of the average user.

On Thursday, Facebook described that data, referred to as ” prevalence,” as its “North Star” for how it polices content. The prevalence figure, the company said, more accurately reflects the experience of its users and the number of hateful posts the company fails to remove. 

In its report on Thursday, Facebook said it had removed 22.5 million posts for hate speech during the second quarter and 22.1 million during the third. Meanwhile, Instagram removed 3.2 million in the second quarter compared to 6.5 million in the third. 

But the ADL said it had expected to see a rise in the number of posts removed due to new Facebook policies, including a ban Holocaust denial and of images containing blackface. Instead, the number of posts removed on Facebook remained relatively unchanged, even though they more than doubled on Instagram under similar policies. 

The ADL lauded Facebook for catching 94.7% of the violating content before users flagged it, but the group said that figure doesn’t provide the full picture. Facebook should also report how many posts users flagged for hate speech because it often fails to remove them after they’re reported, the group said. 

Lastly, the ADL wanted more insight into how hate speech affected specific communities, such as Muslims, Black people, Jews, and members of the LGBTQ+ community. Without that targeted data, civil rights groups will have a harder time suggesting fixes for the problem, the ADL said.

“To show they are taking real steps to reduce hate speech, platforms must try to understand the scope of the problem by collecting the relevant data and using rigorous research methods,” the ADL said. “Failure to do so will result in vulnerable groups continuing to be at the mercy of toxic users on social media.”