Why your restaurant's food reels get likes and no customers

Most restaurant owners believe their reels should show the food. Close-up of the pasta, steam rising, a trending song, 12 seconds. It gets likes. It gets a few "need this" comments from people two time zones away. It does not get the couple down the street to book a table on Friday, and the data suggests it rarely gets shared, which is the only signal that predicts reach.

So what makes a restaurant reel get shared rather than admired, and how do you produce enough of them to matter without a marketing team?

Likes are not the signal. Shares are.

Based on our analysis of 4.5 billion TikTok posts (2014 to 2026), the early signal that predicts whether a post reaches the top 1% by views in its country and week is shares per view. At 0.5% to 1% shares per view, a post has a 2.5% chance of becoming a top 1% hit. At 2% or more, 5.6%. At zero shares, 0.01%. Saves per view at 2% or more give 4.0%. Comments do not differentiate at all.

That is the whole problem with the pretty food reel. It is built to be liked. A share happens when someone sends it to a specific person with a reason: "we should go here", "this is the place I told you about", "this is you". A plate of pasta rarely carries a reason. A story does.

The formats that get shared, ranked

We tagged the opening hook of each post and measured how much more often each type appears among top 1% posts than average. Globally:

• GRWM, vlog and ASMR style: 3.32
• "Part N" series: 2.90
• How-to: 2.65
• Tips: 2.49
• Question: 2.08
• Storytime: 1.81
• POV: 1.39

For a restaurant, the first line is the kitchen: prep, plating, the sound of a knife on a board, the hiss of a pan. ASMR style footage of real cooking carries the highest lift of any hook type, and it is the one thing every restaurant has an unlimited supply of.

The second line is the one to build the account around. Most hook types lose 10 to 15% of their lift every year. The only one whose lift is growing is the "part N" series, from 2.81 in 2024 to 3.01 in 2026. "Every dish on our menu, part 7." "Things our regulars order that are not on the menu, part 3." "A week of specials, day 4." A series gives the viewer a reason to come back and the platform a reason to show them the next one.

Ideas by format

Series: one dish per episode, with the story of where it came from. One regular per episode (with permission), what they order and why. One supplier per episode: the farm, the baker, the fishmonger.

How-to: how the kitchen makes one thing most people get wrong at home. How to order at a place like yours if you have never been. How a table of eight gets fed in fifteen minutes.

Tips: five things to ask for that are not on the menu. Three dishes to order the first time. What to get if you are in a hurry.

Question: "What would you do with 40 kilos of tomatoes that arrived a day early?" "Why does every table ask for the recipe for this one sauce?"

Storytime: the night the power went out mid-service. The customer who proposed at table 4. Why the owner opened the place at all.

If you want to see how these translate for your menu, paste your restaurant's website into /tools/reel-hooks and it will return five hooks based on what the site actually says.

Length: the 12 second food clip is the weakest format

This is the number that most restaurant social media gets backwards. Videos between 61 and 120 seconds appear 3.08 times more often among top 1% posts than average. Videos of 8 to 15 seconds appear at 0.63. Photo posts appear at 0.24.

A 12 second plating clip is not just shorter. It is the format the platform rewards least. A 75 second reel that tells the story of a dish, from the delivery to the pass, has room for a setup, a turn and a payoff, and it is the format the platform rewards most.

In our own 100-reel experiment (one new account, 100 reels, 30 days, all produced by the pipeline that became Ultim), reels built around a named person held attention about twice as long as "mechanics" reels that explained how something works. For a restaurant: a reel about the chef, a regular or a supplier is a story. A reel about "how we make our sourdough" is mechanics. Both can work, but the story holds the viewer, and holding the viewer is where a share comes from.

Sound: your kitchen, not a trending track

A post using the creator's own voice or own recording as the sound carries a lift of 1.09. A post using someone else's music carries 0.79. For a restaurant this is a free win: the kitchen is already the sound. Sizzle, chop, the espresso machine, a voiceover explaining what is happening.

An AI voiceover counts as own sound, because it exists only on your account. Ultim records a fresh voice track for every reel and lays it over the footage with word-by-word captions, so a reel can be watched muted in a queue and still land. The trending song, by contrast, is something the platform has heard a million times, and it treats it that way.

Caption and hashtags: one sentence, a few labels

The caption under the video should be one plain sentence of 21 to 100 characters. That carries a lift of 1.7 to 1.8. An empty caption carries 0.27. One or two emoji carry 1.53; more does not help.

Hashtags are a niche label, not a reach lever. Globally, 6 to 10 hashtags carry 1.65 and zero carry 0.55, but in Poland zero hashtags has the highest lift and 11 or more hurts. For a restaurant, the useful labels are the cuisine, the town and perhaps the neighbourhood. Three to six, not thirty.

Posting time: lunch service is also prime time

Posting between 10:00 and 18:00 local time carries a lift of 1.04 to 1.12, with the peak at 15:00 to 16:00. Posting between 21:00 and 02:00 carries 0.78 to 0.84. Restaurant owners post after close, at 23:30, in the weakest window there is. The reel should go out mid-afternoon, which is exactly when nobody in the kitchen has a free hand. That is a scheduling problem, and a scheduler solves it. Day of week barely matters (0.90 to 1.05), Sunday weakest.

How many reels before it works

Based on our analysis of 4.5 billion TikTok posts (2014 to 2026), among 2.87 million creators with at least 3 posts, the median number of posts before the first top 1% hit is 5, the third quartile is 16 and the 90th percentile is 38. The chance an account has at least one hit is 1.6% at 3 to 4 posts, 4.9% at 10 to 19, 10.9% at 20 to 49 and 39% at 100 or more.

A restaurant that posts eight reels in its opening month and then goes quiet has barely started. A restaurant that reaches fifty has a one in ten chance of having had a real hit, and one that reaches a hundred is better than one in three. Posting 1 to 3 and 3 to 7 times a week give the same share with a hit; above 7 the chance per post drops. Four a week, every week, for six months.

In our own 100-reel experiment, going from one reel a day to four reels a day in four different formats raised the daily total from about 800 to about 1,700 views, with views per reel falling less than proportionally. For a restaurant: a series episode, a kitchen ASMR clip, a tips reel and a story, each on its own schedule, each getting its own test pool.

What this looks like in practice

The restaurant that gets customers from reels is not the one with the best-looking food. It is the one that tells the story of a named dish or a named person in 60 to 90 seconds, with its own kitchen sound and a voiceover, numbered as part of a series, posted at 15:00, four times a week, for long enough to pass post fifty. Every one of those is a production and scheduling problem, which is why most restaurants never get there.

Ultim reads the restaurant's website and menu, writes the series and the stories, records the voice, edits the reel with captions, and publishes on schedule to TikTok, Instagram Reels and YouTube Shorts, with results checked at 24 and 48 hours. The first step is to see whether the hooks it writes sound like your place: paste the site into /tools/reel-hooks.