Digital bookings and orders: the numbers that actually rule in 2026

Why does the digital channel sell more and keep less margin?

Because an aggregator commission is not a marketing expense, it is a direct cut to your selling price, and almost nobody books it that way.

A two-location restaurant in Bogotá was billing 41% of its sales through aggregators and celebrating that at every board meeting, until we split the P&L by channel —nobody had done it in fourteen months— and digital showed 6.2 points of contribution margin against 31.4 from the dining room. The arithmetic is brutal: a 21.80 USD dish arriving through a channel charging 22% did not sell at 21.80, it sold at 17.00, and real food cost jumped from 29% to 37% without the recipe changing a single gram. Commissions run between 15% and 30% of the gross ticket depending on market and plan, while a direct reservation costs 0 to 3 USD per guest on your own engine. Technology is not the problem.

Why does the digital channel sell more and keep less margin — in practice?

Channel accounting is. Digital orders at full-service restaurants grew 237% since 2020, according to Restroworks (Restaurant Sales Statistics 2025). That figure explains why so many owners fell in love with the channel without auditing it:

the curve goes up and nobody argues with a curve that goes up. Yet 237% growth in VOLUME, arriving through channels charging as much as 30%, coexists perfectly well with an operating margin sliding quarter after quarter, and that is exactly what shows up once you open the books. Add the cost context: the National Restaurant Association measured a 35% rise in food inputs and 35% in labor versus 2019. On that cost base, giving away 22 points of price per channel is not a growth decision, it is a badly made survival decision. Before scaling the digital channel, measure what each order leaves after commission, packaging and dining-room spillage. A no-show does not cost you an empty table, it costs the empty table PLUS the guest who called, found no seat and had dinner around the corner.

No-shows: the cost that never appears in any report

With 17.4% no-shows and 2.3 turns per night, a 60-seat dining room loses roughly 24 covers per service; at a 27 USD ticket that is 648 USD a night no sales report ever records, because accounting systems only know how to subtract what was billed, never what was never billed. Multiply by 26 services a month and you get 16,848 USD a year per location in invisible losses, a figure that usually exceeds the full cost of running your own reservation engine. The concrete decision: if your no-show rate passes 10%, require a guarantee card during peak windows and send confirmation messages 24 hours ahead. That alone, executed properly, recovers half of it. A Reel that converts is not the one piling up views, it is the one pushing traffic to your own reservation engine in fewer than three taps. Three taps, not five: that funnel friction is the only thing you control end to end, and it is worth more than any paid budget.

Video content: the right metric is the confirmed reservation

Diego F. Parra, restaurant consultant and founder of Masterestaurant, insists that the content metric for restaurants is not reach but attributable confirmed reservations, and the evidence backs him: Marketing LTB reports a 30% lift in bookings the week after a creator posts (Influencer Marketing Statistics 2025), and Harvard Business School measured with Michael Luca that each additional star in review ratings moves between 5% and 9% of revenue. Social conversion gets measured wrong in most operations because the link lands on the aggregator —where you pay 22%— instead of the owned engine, where you pay nothing. These benchmarks do not apply the same way across three scales, and confusing them costs money. SMALL venue, under 60 seats and one strong turn: your lever is the no-show, not the commission; with 24 covers lost per service at 27 USD, attacking that 17.4% with automated confirmation pays better than negotiating points with an aggregator that moves little volume for you.

How to read these numbers in YOUR operation?

MEDIUM operation of two to five locations: channel P&L rules here; if digital leaves 6.2 points against 31.4 from the dining room, cap the delivery kitchen during peak hours and protect the profitable channel.

GROUP of six or more: the lever is infrastructure, and the hard data sits with Chipotle, which opened between 315 and 345 locations in 2025 with more than 80% in the Chipotlane drive-thru format (Chain Store Age, Q4 2024). At that scale, the digital channel gets redesigned on the floor plan, not in the app. Self-service kiosks lift the average ticket between 8% and 15% against the counter —Yum reports around 10% (QSR Magazine 2024)—, and McDonald's has communicated increases up to 30% in its rollouts. There is a tension worth resolving there: raising the ticket 10% does not offset giving up 22 points of commission, because the first works on the sale and the second on the effective price.

Kiosks raise the ticket, but they do not fix the margin by themselves

The correct math is simple. On a 30 USD ticket with a 10% lift through your own kiosk, you gain 3 USD and pay no commission; that same ticket through an aggregator at 22% takes 6.60 USD away. Four times worse, with more kitchen work. Contactless payment grew 260% between 2020 and 2023 per the Restaurant POS Systems Market 2024 report, which means infrastructure is no longer the barrier: guest habits are ready, the decision is yours. Whoever owns the contact sets the price, and that is the whole difference between a profitable digital channel and one that eats your margin. Stripo measured in 2025 that personalized email drives 26% higher open rates in restaurants, a channel costing cents per send against the 15% to 30% an aggregator takes, and one that belongs to you: if the aggregator raises its commission three points tomorrow, you do not negotiate, you comply.

Owning the contact beats borrowing the sale

Ask yourself what would happen if your main aggregator shut down on Monday. With your own guest database, you lose a channel. Without it, you lose 41% of your sales and have nobody to notify. That is why the right order inverts common practice: first you build the contact base and the reservation engine, then you use the aggregator as paid acquisition of new customers, on a capped and measured budget, never as the backbone of the operation. The industry figures cited here come from public, verifiable sources: National Restaurant Association (2024) for cost evolution, Restroworks (2025) for digital order growth, QSR Magazine and McDonald's for the kiosk effect, Harvard Business School for the revenue impact of reviews, and Chain Store Age for Chipotle's expansion plan. The limits deserve stating without ornament. Most of that evidence is collected in the United States, where commissions and labor structure differ from Latin America and Spain, so the magnitudes travel better than the absolute values.

Where these benchmarks come from and how far they reach?

The operating data —the 17.4% no-show rate, the 6.2 points of digital margin, the 41/59 channel split— belong to the Bogotá case described above and serve as a method reference, not as a market average.

Measure your own: two months of tickets separated by channel already give you a baseline. Aggregator commission is not a marketing expense, it is a direct price cut. When a 21,80 USD plate comes through a channel charging 22%, you did not sell at 21,80 but at 17,00, and your real food cost jumped from 29% to 37% without a single gram changing in the recipe. That is the number for the board meeting, not gross revenue. No-show cost is almost always miscalculated: it is not the empty table, it is the empty table PLUS the guest who called, found nothing available and went elsewhere.

Where the model actually breaks?

At 17,4% no-show and 2,3 turns per night, a 60-seat room loses roughly 24 covers per service, which on a 27 USD ticket means 648 USD a night that never surface in any report.

Social conversion is measured badly in nine cases out of ten because the bio link lands on a static menu. Swapping that destination for a concrete time slot multiplied conversion sevenfold across the operations we have supported, and the change cost exactly nothing: a different URL. AI agents in digital bookings and orders pay off where repetitive VOLUME lives, not where judgment lives. Confirming, rescheduling, warning about delays and basic upselling automate cleanly; deciding whether the window table goes to the couple celebrating an anniversary does not. Blurring those two territories is what produces the cold digital experience nobody can later explain. Restaurant digital transformation fails when software is bought before the indicator is defined. First the number you intend to move, then the tool; the other way round you end up with seven subscriptions and no dashboard anyone opens on a Tuesday morning.