Delivery is three apps and one shared truth. Customer, merchant and courier all need a live, consistent view of the same order. Agencies quote $60,000 to $300,000. We build a working core in one week for $5,000.
Not the screens. These are the parts that decide whether a delivery marketplace works or falls over:
Three-sided consistency. One order, three live views. When the merchant marks it ready, the customer and the courier both need to know immediately.
Dispatch and reassignment. Offering deliveries to couriers, handling declines, and reassigning cleanly when someone drops mid-run.
ETAs that hold up. Prep time plus travel time, updated live. Wrong ETAs are the single biggest driver of refunds and churn.
Menus with modifiers. Options, sizes, extras and out-of-stock state, per merchant, without a combinatorial mess.
Splitting the money. One payment divides between merchant, courier, tip and platform fee, and has to reconcile.
The specific decisions that make a delivery marketplace hold up, and that the obvious version gets wrong:
One transaction decides the order. Verifying stock, creating the order and decrementing inventory run in a single serialisable transaction, so two customers cannot buy the same last item.
Prune by distance before asking for travel time. Candidates are filtered by straight-line radius first, and only the survivors go to a travel-time service. Requesting real ETAs for every merchant is what makes these systems slow and expensive.
Availability served from replicas. Menu and stock reads vastly outnumber writes, so they come from read replicas partitioned by region with a short-TTL cache in front.
Dispatch as a durable workflow. Offer, decline, timeout and reassignment run as a workflow with retries and persisted state, so a courier dropping mid-run reassigns cleanly instead of stranding the order.
One order, three live views. Customer, merchant and courier all subscribe to the same order state and receive pushed updates, so nobody is acting on a stale screen.
The gap
A studio quote for a delivery marketplace usually lands at $60,000 to $300,000 over six to twelve months. We build a launchable core in one week for a flat $5,000, deployed on your own accounts, with every line of code yours to keep.
A working product you can put in front of real users:
Customer ordering with menus, modifiers and stock state
Merchant order screen with accept and ready-for-pickup
Courier app with live location and delivery offers
Dispatch, assignment and reassignment
Live order tracking with ETAs
Split payments, tips and payouts
And what we would leave out. Being straight about this is the only way a fixed price means anything:
Batched multi-order route optimisation
Machine-learning ETA and demand prediction
Subscriptions and loyalty programmes
In-app advertising and merchant promotion tooling
You are getting the core of the same problem the big product solved, scoped to something launchable. Not a clone of the finished article.
It is one system with three interfaces, which is why it is achievable. The shared order model and the live layer are built once and used by all three.
The order returns to dispatch and is reoffered, with the customer told. That path is part of the core build, because it happens constantly in the real world.
A free 20 minute call. We will tell you if an app like DoorDash is a fit for one week, and what it takes.
Start here: what it costs to build an app in 2026, or how the week works.