Meta's worst day was TikTok's best
Meta lost over $230 billion in market value in one day after Facebook's first ever drop in daily users. The interest graph beat the social graph.
Meta’s stock fell about 26% on Thursday, erasing more than $230 billion in market value, the largest one-day loss for any US company. The trigger was the quarterly report: Facebook’s daily active users declined for the first time in its history, and the company guided lower revenue, blaming Apple’s privacy changes and competition from TikTok for users’ time.
Zuckerberg was unusually direct on the call about TikTok. People have many choices for how to spend their time, and apps like TikTok are growing very quickly. Reels is growing too, but it makes less money per minute than the feed and Stories, so shifting attention to it hurts revenue in the short term.
In 2017 I wrote that ByteDance buying Musical.ly was really a recommendation engine buying a global audience, and in 2018 that once people got used to a feed that doesn’t need their friends to be interesting, a feed that does would start to feel empty. Thursday’s results suggest that shift is starting to hurt Facebook’s business.
Facebook was built on the social graph: who you know. Your feed was mostly things your friends posted or shared, ranked by how likely you were to engage. That was a huge advantage when your friends were posting a lot. But people post less than they used to, and most people’s friends aren’t very interesting to watch every day.
TikTok is built on the interest graph: what you actually watch. It doesn’t need you to follow anyone. Every creator’s video is a candidate for every user, and the ranking learns your taste directly from your behavior. It has a much bigger pool of content to choose from and a much denser signal about what you want. The social graph is a subset of the information the interest graph has.
Meta is now moving every product toward recommended content, with Reels in Instagram and Facebook and more “suggested” posts in feeds. They have the data, the engineers and the compute to do it well. The problem is that users have come to expect a friends-based experience from Facebook, and recommended content from strangers feels out of place there in a way it doesn’t on TikTok, which was built for it from the first day.
The lesson I’d draw is broader than Meta. For a product that shows content, what the ranking system optimizes and what data it learns from are the core strategic choices. Everything else follows from them. The social graph was a great way to find content when there was little content. When there’s an effectively infinite supply, the winner is whoever learns each person’s taste fastest.