The real reason content goes viral (it's not the hook)

In 2012, Jonah Berger and Katherine Milkman tracked almost 7,000 New York Times articles published between August 30 and November 30, 2008, and recorded which ones made the paper's "most emailed" list. The articles that made people feel awe, anger or anxiety were more likely to make it. The ones that made people sad were less likely.

What 7,000 articles showed about sharing

The researchers had a few thousand of the articles coded for the emotions they caused and for how useful the information was. An article one standard deviation higher in a quality than average (a standard deviation is the typical distance from the average) had these changes in its chance of being one of the most emailed:

• Anger: up 34 percent
• Awe: up 30 percent
• Practical value: up 30 percent
• Interest: up 25 percent
• Anxiety: up 21 percent
• Sadness: down 16 percent

Then they ran lab experiments to find out why. Making a story more infuriating raised people's willingness to share it, and making a story sadder lowered it. In both experiments, the change ran through arousal, which is how physically activated someone feels: a racing heart and alertness are high arousal, and heaviness and calm are low.

Sad is heavy, and heavy doesn't get shared

Sadness feels like a strong emotion, but it's a low-arousal state: people who feel it tend to slow down and pull back. Awe and anger are high-arousal states, and people who feel them are more likely to act, which includes forwarding something. Quiet content can still be high arousal. A number that changes how you see something you thought you understood is quiet, and it's activating.

Video ads point the same direction

Gerard Tellis and three colleagues studied 345 and then 512 brand video ads on YouTube from 2013 to 2017, counting how often each was shared on Facebook, Twitter, Google+ and LinkedIn. Ads built on warmth, inspiration, excitement and amusement were shared more. Ads focused on information were shared less, except for new or high-priced products and on LinkedIn. Ads that showed the brand early were shared less, while how long the brand stayed on screen made no measurable difference. The most-shared length was about 1.2 minutes in the first study and 1.7 minutes in the second.

The Tellis ads were paid ads from large brands, not posts from a small business, so treat the length numbers as a description of those ads rather than a rule for a Reel.

Where it stops

An email goes to one person, and posting publicly works differently. In Berger and Milkman's own data, useful articles were slightly less likely to appear on the paper's most-blogged list. And when Alixandra Barasch and Berger compared posting to all of someone's friends with messaging one friend, posting to all their friends made people more focused on how they'd come across. That's the subject of why people send your post to someone.

How to use it

• Decide what someone should feel before you write. Awe, amusement, anger at something unfair, or the jolt of a number that changes their mind. "Informative" describes the content, not what the reader feels.
• Give the useful part a feeling. In the New York Times data, practical value and awe each raised sharing by about the same amount, so a useful post that also causes awe has both working for it.
• Put your brand at the end. In the Tellis ads, early brand placement went with fewer shares.
• Keep the feeling and the finish together. Platforms track whether people watch to the end, so a video that delivers its feeling and ends is doing two jobs. See what Instagram and TikTok say they measure.

Sources: Berger & Milkman (2012); Tellis et al. (2019); Barasch & Berger (2014). Every study was read in full before it was cited here. Updated October 2026.