TikTok has spent years being the place people find restaurants by accident. On Wednesday it started trying to be the place they look for one on purpose.
"Must Eat" is a community-driven restaurant guide inside the app, built from more than 40 million posts and comments. Users browse by city, filter by cuisine, price, rating and location, and see results split into "Everyday Eats" for casual neighbourhood places and "Elevated Dining" for occasions. Select restaurants can be booked directly through TikTok via OpenTable. It launches first in New York, Los Angeles and San Francisco, reachable from search, the Local feed, restaurant pages, and location links attached to videos.
The strategic logic is obvious. Yelp and Google Maps are built on explicit reviews and structured listings; TikTok is built on implicit recommendations — someone filming a plate, a comment saying "the one on 6th". That signal already exists in enormous volume. Must Eat is an attempt to index it, which is the hard part: turning unstructured video and comment text into a stable list of businesses with consistent prices, cuisines and hours.
Why it matters: local discovery is one of the few remaining advertising categories where Google has not been dislodged, and it is high-intent — someone searching for dinner is minutes from spending. A recommendation feed with booking attached closes that loop without leaving the app.
The open questions are the ones that decide whether this works. Whether recommendations extracted from 40 million posts are accurate enough to trust is unproven; a feed that sends someone to a closed restaurant once loses them. Whether merchants will treat a TikTok listing as canonical is another. And the guide covers three cities at launch — where the density of food content and advertisers is highest, not where the harder problem lives.
A note on the framing: TikTok calls the guide community-driven. It is, in the sense that the raw material comes from users. The ranking, the categories and the booking integration are TikTok's.
