What do Taylor Swift, the city of Dunedin, Florida, and an English cricket club have in common? They’re all leery of comments on social media. Swift recently advised new artists to refrain from reading negative comments to avoid derailing their creative process. The city of Dunedin just suspended comments on its official social media pages because of resident harassment. And an English cricket club took the same step this month, citing player mental health concerns.
While many have decried social media for misinformation, hate speech, and other problems, most concerns have focused on platform design or the governing algorithm. Comparatively little attention has been paid to comment threads below official posts. New research from Columbia Business School Assistant Professor Dante Donati and University of Illinois Urbana-Champaign Assistant Professor Lena Song puts this specific aspect of social media under the microscope.
“Social media is like a public square,” says Donati. “It’s all about two-way engagement. And a strong majority of users — 85% — say they at least sometimes look at the comments on social media posts.”
The research team found that the tenor of comment sections can directly affect brand outcomes. Supportive comments tend to have little effect. Negative comments, on the other hand, substantially increase reactions, comments, and click-throughs. Basically, outrage makes us pay attention.
Publicists are already clued in to this phenomenon — recall that “rage-bait” was Oxford English Dictionary’s word of the year in 2025. But Donati and Song found that outrage-inducing engagement comes at a hefty cost. Even as negative comment sections drive engagement, they may make consumers less likely to purchase from or donate to the organizations behind the post.
Not all publicity is good publicity
Most past research on social media hasn’t focused on comment sections because it’s challenging to isolate the impact of comments from that of the original post. Donati and Song devised a workaround to this problem by using existing platform features on Facebook to run an experiment. They randomly assigned about a million viewers of posts by a large racial justice organization to different comment sections using A/B advertiser testing tools as well as the comment hiding function. This allowed them to measure how users reacted to different kinds of comments. They found that negative or opposing comments led to more clicks, comments, and reactions.
Next, in a second experiment, they interviewed 5,000 online survey participants about their attitudes and behavior after viewing the same posts and comments. Those who had seen mostly negative comments were less supportive of the organization and less likely to make a donation afterward.
This research documents a tension that publicists have long navigated. “Brand managers would argue that there has always been this trade-off between buzz and brand safety,” Donati says. “Content that is more buzz-worthy is also more risky.”
Those relying on the “rage-bait” effect might see soaring engagement metrics for their social media posts. But they risk losing control of the narrative and turning off visitors. On the flip side, polite content is safer for the brand’s image but may not achieve the same visibility.
Pleasing all stakeholders some of the time
Brand managers whose primary goal is engagement must be willing to accept some negative comments to gain visibility. But if a subsequent purchase or donation is more important, they might seek to limit comments on posts. And in some cases, they might want to turn off comments entirely.
Donati advises brand managers to consider content moderation like a volume wheel. The noise from negative comments can be filtered more or less aggressively depending on the brand’s engagement objectives and risk tolerance.
For Donati, it’s imperative that regulators require social media sites to provide content moderation tools that allow organizations to make that decision. That’s not necessarily a given, considering that regulators must balance the needs of both brands and social media platforms — and that social media platforms generally want the engagement spike provoked by outrage.
“The algorithm is very good at amplifying extreme voices,” Donati says. “The platform benefits from the fact that you have these vocal minorities with high visibility because they are the ones that comment a lot.”
But his research makes it clear that the organizations using social media to get their message out may not see similar benefits from negative comments. Regulators must balance both perspectives when it comes to formulating policy on platform design.
“We don’t want censorship here,” Donati says. “That's why it's a delicate thing for a regulator to tackle. Comment and content moderation tools are important because they allow freedom to the brands to decide and also give people a voice.”
