What a delivery app costs
This sum has something the others do not: you are already paying for one, even without having it. Marketplace commission is the price of the delivery you use today, and it is what any alternative has to be measured against.
The cost that already exists
A marketplace charges a percentage of every order. The range runs from roughly 12% when you deliver yourself up to around 30% when the platform delivers, plus a monthly fee on some plans.
The number matters less than its shape: it is a percentage, so it grows with your revenue and has no ceiling. A restaurant doing R$ 40,000 a month in delivery pays, at 20%, R$ 8,000 a month. Close to a hundred thousand a year.
The three routes, with the sum done
An agency. A delivery app is among the priciest projects on the menu, because it has a catalogue, a cart, payments, delivery, a courier area and an admin. Tens to hundreds of thousands, plus maintenance.
A freelancer. Much cheaper, same scope, the same continuity risk as always. Add servers and a database, which in a delivery app are not token costs.
A platform. A fixed subscription. On Fabapp, Free builds and publishes on the web, and paid plans run from R$ 99 to R$ 1,099 a month. You pay the same at 50 orders or 5,000.
Against 20% commission, R$ 249 a month breaks even at R$ 1,245 of delivery revenue. Above that, every real that goes through your app instead of the marketplace stays with you.
What drops out of the quote and comes back later
- the payment processor's fee, which exists in every scenario and is not the platform's;
- delivery itself, which is the real cost of the operation and does not change with the app: your own courier, a courier app, or counter pickup;
- a developer account, if you want the stores: US$ 25 once with Google, US$ 99 a year with Apple;
- acquisition. This is the big one. The marketplace charges a lot because it brings new customers. Your app does not, so you have to count who will tell people it exists: the packaging, the QR code at the counter, and the social accounts you already have.
The arrangement that usually works
Almost nobody leaves the marketplace in one go, and nobody has to. What works is running both and moving people who have already ordered to your own app: the repeat customer, who would have come back anyway, and whose commission hurts most precisely because it bought nothing new.
A voucher printed on the bag with your app's link costs pennies and converts somebody who already liked the food.
The short version
- the reference cost is not zero, it is the commission you already pay;
- commission is a percentage and grows; a platform is fixed and does not;
- your own app does not bring new customers, it retains the ones you have;
- payment fees and delivery costs continue in every scenario;
- start with repeat customers and keep the marketplace for discovery.
What you can build and how is in delivery app, and the feature comparison is in how to build an app like Uber Eats.