Digital reservations and orders: the alternatives that survive a P&L review

The same customer is worth eleven dollars less when the order comes through a marketplace

A 38 USD ticket through Rappi leaves 27 USD in the till before you touch a single ingredient, while that same ticket placed through «Order online» on Google Maps leaves 36.7 USD, and the gap is eleven dollars per order that nobody writes down anywhere. Marketplace commission runs between 15% and 30% of the ticket depending on the plan you signed, and local payment gateways now charge between 2.9% and 3.5%, so the spread is not a consultant's opinion but plain cash arithmetic you can rebuild this afternoon with last quarter's closings. Latin America's online food delivery market moved 23,783.7 million USD in 2024 and grows at 8.1% a year through 2030, according to Grand View Research, so the channel is not going away. It is going to get more expensive. The marketplace stops working for you the moment your repeat-order rate climbs past 40%, because from there on you pay a discovery fee for customers who discovered you months ago.

When the marketplace stops being enough?

That is the number that gives it away, and it sits inside the platform's own dashboard:

if 60% of a month's orders come from phones that already ordered before, then you are paying between 15% and 30% for retention work that an owned channel does for 2.9%. There is a second signal, less obvious and far more expensive: when the platform asks you to join a 20% discount campaign so you don't lose ranking in the listing, your effective commission climbs above 40% of the ticket and what you have is no longer a channel. It is a majority partner who never put in capital. The first alternative costs nothing and remains switched off in a number of restaurants that is genuinely hard to explain. Google Business Profile routes the «Reserve» and «Order online» buttons straight to whichever provider you designate, and that click pays no discovery commission for one simple reason: discovery already happened inside Maps.

Google Business Profile with native booking and ordering: zero cost, two hours to set up

Setup takes about two hours and address verification another 48. Who is it for? For absolutely everyone, from the twelve-table place to the nine-location chain, with no exception I can think of. Its limit is real and worth stating before anyone gets excited: Google does not manage the order, does not charge, does not dispatch and does not keep the customer, it only routes. You need something behind it to receive, and that something is the decision that actually matters. Swapping per-cover commission for a flat fee pays off from 60 covers per service onward, and the math is as dull as it is conclusive. OpenTable on its no-commission plan, SevenRooms or Meitre in Latin America charge between 49 and 199 USD monthly depending on modules and volume, which means a restaurant doing 60 covers a day at a 30 USD ticket carries a booking cost near 0.4% of sales, against the 1.50 to 2.50 USD per diner the commission model bills.

Fixed-fee booking engine: between 49 and 199 USD a month, no charge per cover

Configuration takes 6 to 10 hours of work plus a week of shakedown on the floor, because your staff has to learn to block tables on the tablet without the queue at the door piling up on them. Below 40 covers, that flat fee does not pay for itself. Putting ordering on your own domain and charging through a local gateway drops transaction cost into the 2.9% to 3.5% band, and that is where the real money shows up. A restaurant billing 40,000 USD monthly in delivery that manages to shift just a third of that volume off the marketplace recovers between 1,600 and 3,500 USD a month, depending on the commission it pays today. The switching effort is the heaviest on this list: 20 to 30 hours to set up catalogue, photography, delivery zones and payment rules, plus a courier budget that no longer comes bundled.

Your own ordering engine on a local gateway: 2.9% to 3.5% against 15%-30%

Who it works for: restaurants above 1,500 monthly orders with a customer base that already repeats. Below that floor, you will be administering software instead of cooking. A marketplace's visible cost is the commission, but the expensive cost is that you never learn the name, the phone number or the frequency of the diner who ordered on Friday. Diego F. Parra keeps pressing this point at Masterestaurant because it lands straight on the P&L: with no data of your own there is no win-back campaign, no reactivation of lapsed guests at day 45 and no menu segmented by consumption habit, which are the three levers that raise frequency without touching plate price. The restaurant management software market will move from 6,540 million USD in 2025 to 14,730 million in 2031, at a 14.52% CAGR, according to Mordor Intelligence, and that growth is explained precisely by owners who stopped renting their customer base.

What the public conversation skips: who owns the customer data?

Commission is paid once per order. Somebody else's data is paid forever.

If you wipe the marketplace off the map this Monday, first-month revenue will drop somewhere between 25% and 35% and will not fully recover before month five, even doing everything right. The reason is that discovery traffic does not transfer on its own: the diner who found you searching «sushi near me» inside the app has no reason whatsoever to type your restaurant's name into a browser. That is why the right alternative in 2026 is not to switch off, it is to DOWNGRADE. Keep the marketplace alive as a shop window, with the menu trimmed to your best-margin dishes and no promotions, and push the recurring customer toward your own channel with an incentive that costs you less than the commission: a dessert costing 1.20 USD buys an order whose avoided commission is worth 6 USD.

What would happen if you switched the marketplace off tomorrow?

That is the whole trick.

Stay where you are if you bill under 8,000 USD monthly in delivery, if your repeat rate sits below 20% or if you still have nobody with a first and last name responsible for reading the dashboard every week. In those three scenarios the marketplace is cheap for you even taking 25%, because hiring the commercial work it does on your behalf would cost more. I got this wrong for years, recommending owned channels to restaurants with no volume, and the outcome was predictable: they paid an 89 USD monthly platform to process fourteen orders. The real threshold sits around 1,200 monthly orders, the point where avoided commission comfortably beats the sum of flat fee, gateway and management hours. Before that point, your problem is not the digital channel. It is visit frequency, and that gets fixed in the kitchen. ALTERNATIVE 1 — Google Business Profile with native booking and ordering.

The alternatives, one by one, with price and fine print

Cost: zero. Curve: 2 hours to set up plus 48 hours of verification. Who it suits: everyone, and the number of venues still skipping it surprises me. Google routes the «Reserve» or «Order online» button to whichever provider you pick, and that click pays no discovery commission because discovery already happened on Maps. Its limit: it routes the order, it does not manage it, so you need something behind it. ALTERNATIVE 2 — Flat-fee reservation engine (OpenTable on commission-free plans, SevenRooms, Meitre across Latin America). Cost: 49 to 199 USD monthly by cover count and modules. Curve: 6 to 10 hours of setup plus a week of floor practice. Who it suits: venues above 60 covers per service or with a firm booking policy. The real gain is not the diary, it is the history: by month three you know which table cancels, at what hour and with how much notice.

The alternatives, one by one, with price and fine print — in practice

ALTERNATIVE 3 — Owned online ordering on your domain (Ordering, Toast Online, a POS module). Cost: 0% to 3.5% gateway plus 59 to 179 USD monthly. Curve: 10 to 15 days including catalogue, photography and kitchen printing tests. Who it suits: anyone past 400 digital orders a month. Below that line the flat fee eats the saving and you end up working for free for your restaurant software vendor. ALTERNATIVE 4 — WhatsApp Business API with catalogue and payment. Cost: 0.005 to 0.08 USD per conversation plus the provider. Curve: 4 hours, the gentlest of the set. Who it suits: neighbourhoods where guests already message you, and venues under 25 tables. Its weak spot is volume: past 60 daily orders without automation a human falls behind and reply time blows past the 4 minutes guests tolerate. ALTERNATIVE 5 — Marketplace on self-delivery or reduced commission. Cost: 12% to 18% instead of 27% to 30%.

The alternatives, one by one, with price and fine print — key points

Curve: 3 hours, plus hiring riders. Who it suits: venues with two in-house riders and a radius under 3 kilometres. I got this wrong for years by recommending self-delivery as the default: under 20 daily orders the cost per own delivery climbs to 3.80 USD and the commission works out cheaper. ALTERNATIVE 6 — Channel aggregator with unified KPI dashboards (Otter, Deliverect, Cuenta Digital). Cost: 89 to 299 USD monthly. Curve: 8 hours plus POS integration. Who it suits: operations running three or more apps with a kitchen that saturates. Here is algorithmic hospitality done properly: the system tells you when to switch a dish off because the line is 14 minutes behind, and you kill it across all four apps at once.