UGC and food micro-influencers: the numbers before and after costing them properly

What does a diner brought in by a creator actually cost?

Between 1.80 and 4 USD per attributed visit when the program is costed properly, and between 25 and 35 USD when comped food gets booked at menu price.

The creator doesn't move that spread: the accounting entry does. Back to the 65-seat restaurant in Bogotá: 41,000 USD in March sales, 900 USD sitting in a «digital advertising» line, of which 640 USD were comped meals valued as if they had been sold. Re-costed at a real food cost of 30.8%, those 640 USD became 197 USD of raw ingredient, the whole program dropped to 457 USD, and customer acquisition fell from 31 USD to 15.70 USD per new diner. The campaign they were about to kill in Monday's meeting was the most profitable one of their year, and they had spent four months reading it backwards because of a classification error no creator could fix.

Discovery already moved to short video, and that changes the budget

63.1% of users discover products and trends directly on TikTok, according to The Influence Agency (2025), and Forbes points to short video as the fastest-growing restaurant discovery channel. Two Restroworks (2025) figures set the ceiling: 220,800 average views per food and beverage video on TikTok, 135,200 on Instagram Reels. Now translate that into cash. If a 220,800-view video converts at 0.05% —a conservative rate for local hospitality— that's 110 visits; at a 28 USD average check, you get 3,080 USD in sales from one piece that cost a tasting menu. That calculation, and not the follower count, belongs in your marketing budget. Even so, beware the mirage of national reach: a viral video in Medellín fills no tables in Cartagena. UGC wins on unit cost and the micro-influencer wins on speed of installation, and blending both objectives is exactly what makes owners abandon the two.

UGC versus paid collaboration: where each one wins

A diner-content program holds that 1.80 to 4 USD range per attributed visit because the marginal input is a dessert with 2.40 USD of raw ingredient plus a well-made request at the table; the payoff arrives slowly, on a curve, and depends on how many people already walk in. Working with a creator of 8,000 to 30,000 followers does the opposite: it installs recognition within two weeks, runs between 120 and 400 USD per piece depending on the market, and switches off just as fast once you stop paying. A restaurant that just opened needs the second one; a six-year neighborhood spot serving 400 diners a week is burning money if it doesn't mine the first. Attribution separates managing a channel from having opinions about it. Without a per-creator identifier, visits fall into the «organic» bucket, the channel looks free in January and useless in March, and continuing depends on the board's mood.

With no booking code, all your traffic is called «organic»

One keyword at booking time —a different one per collaborator— is enough to produce three series: cost per visit, average check, and 90-day return rate. That is where the Restroworks (2025) QSR figure, with roughly 71% of sales coming from repeat customers, stops being trivia and starts giving orders: if the diner brought in by creator A returns at 22% and creator B's returns at 6%, you don't have two similar campaigns with different reach, you have an investment and an expense, and the whole budget should move before month's end. A diner who arrives through your own channel is worth 45% more across their relationship with the house than one captured through the web alone, according to Lightspeed (Online Ordering Statistics 2025). That premium rewrites the math of a collaboration entirely. Assume a 28 USD check and three visits a year: the standard diner leaves 84 USD annually, the owned-channel diner 121.80 USD.

Lifetime value is the number that decides, not reach

Paying 400 USD for a campaign that brings 40 new diners is 10 USD a head against 84 USD of annual value, and that 8.4-times multiple is what you defend in front of a skeptical partner. LoyaltyPass (Restaurant Loyalty Statistics 2026) makes the same point from another angle: 39% of US restaurant visits already come from loyalty program members, double the 2019 share. Acquisition without a mechanism for return is a leaking bucket. Three scenarios, three different decisions off the same benchmarks. Small restaurant, up to 40 seats and under 25,000 USD a month: forget paid collaborations, build pure UGC with an incentive worth 3 USD of real cost per published piece, and cap spending at 0.8% of sales, roughly 200 USD. Mid-sized, 60 to 100 seats and 40,000 to 90,000 USD monthly: mix two micro-influencers per quarter at 250 USD each plus continuous UGC, ceiling at 1.5% of sales and a booking code mandatory from day one.

How to read these numbers in YOUR operation?

Group of three or more units: centralize measurement, negotiate package rates —the typical drop runs between 20% and 30% against one-off hiring— and demand a cost per attributed visit under 5 USD or cut the contract at the next renewal.

The ranges are mine; the base figures come from the sources cited. Reach, discovery, and loyalty figures come from public studies by Restroworks, Lightspeed, LoyaltyPass, The Influence Agency, and Statista, published between 2024 and 2026, with samples dominated by operations in the United States and Western Europe. Three honest limits before you use them. First: almost all of them measure chains and QSR, so an independent white-tablecloth restaurant should discount conversion rates by 20% to 40%. Second: the 220,800-view average per TikTok video is a mean inflated by viral outliers, and a local account's real median sits orders of magnitude lower.

Where these benchmarks come from and what they do NOT prove?

Third: none of these studies measures cost per attributed visit in independent restaurants, because almost nobody instruments it; the 1.80 to 4 USD range comes from re-costing comps at raw ingredient inside real restaurant financials, not from a survey.

Treat them as directional reference, never as a forecast. Comped food is costed at raw ingredient, always, and that single line reorders the entire conversation. A 54 USD tasting menu handed to a creator, at a 30% food cost, costs 16.20 USD of input; whoever books it at 54 USD is declaring a negative margin invented by their own system and making shutdown decisions on a fake number. I got this wrong for years, defending the accounting logic of the table's opportunity cost, until I understood that it only applies when the place is full and you are turning reservations away; on a Tuesday at 45% occupancy, that table had no buyer.

The costing rule that puts the whole program in order

Inside the Masterestaurant financial pillar this is non-negotiable: comps at real food cost, media and fees at the price paid, and everything under the same marketing line. Open your March P&L, find the digital advertising line, and re-cost the comps before Friday. The first difference is arithmetic and hardly anyone applies it: comped food gets costed at raw material, never at menu price. A 54 USD tasting menu handed to a creator at 30% food cost costs 16.20 USD of input. Book it at 54 USD and you are declaring a negative margin that does not exist, then closing decisions on a number your own system invented. Attribution is the second gap. With no booking code per creator, all traffic lands in «organic» and the channel looks either free or useless depending on the month's mood. A keyword mentioned at the moment of booking turns an opaque campaign into a data series with cost per visit, average check and return rate.

Where the real difference sits?

Third comes the split between UGC and paid collaboration, two different businesses wearing the same face. Content from a guest who already paid carries near-zero marginal cost and high credibility;

the paid collaboration buys speed and message control, which is why it runs five to eight times more per visit. Blend them in one accounting line and you cannot see which of the two is carrying your sales growth. Timing is the fourth. UGC matures slowly and compounds online reputation that keeps working months later; paid collaboration produces a booking spike lasting 72 to 96 hours, then fades. A house that needs to fill a Tuesday in September should not reach for the same tool as one building a brand over two years. And the fifth destroys the most margin: table opportunity cost. Seating a creator and a guest on a Friday at 20:30 in a room with a waitlist costs the full contribution margin of that table — between 26 and 44 USD depending on check — on top of the input.

Where the real difference sits — in practice

Off-peak that cost nearly vanishes, and scheduling well versus badly swings roughly 40% of the program's total cost.