Hyperlocal delivery for restaurants means getting a hot order from your kitchen to a customer within a few kilometres, usually inside 30 to 45 minutes. You have three ways to do it: your own riders on payroll, on-demand delivery partners that you pay per trip, or the aggregator's fleet when the order comes through a food delivery app. Each one costs differently, fails differently and suits a different volume of orders.
This guide compares the three on cost per delivery, reliability, control and effort, with rupee numbers for a typical city kitchen. It is the starting point for the rest of our delivery series, which goes deeper on zones, riders, packaging, tracking and the real cost of a small order.
What hyperlocal delivery actually involves
Every delivery, however it is done, has the same five steps: the order is confirmed, the kitchen prepares it, a rider is assigned and reaches the kitchen, the rider travels to the address, and the food is handed over. The model you choose decides who controls each step and who pays when one goes wrong.
Two numbers decide most of the economics. The first is your delivery radius, which for hot food rarely stretches past 5 to 7 kilometres before the food arrives cold and the rider's time is wasted. The second is your order density, meaning how many orders you get per hour within that radius. A kitchen doing 10 orders an hour in a 3 km radius can keep two riders busy; a kitchen doing 10 orders a day cannot keep one.
If the radius question is open for you, read how far a cloud kitchen should deliver before deciding on riders at all.
Option 1: in-house riders
You hire riders, pay a monthly salary plus fuel, and dispatch them yourself from the counter or the dashboard. The rider wears your brand, knows your regular customers, and can carry two or three orders on one trip when they go the same way.
The cost is largely fixed. A full-time rider in a tier-2 city commonly costs ₹12,000 to ₹18,000 a month in salary with fuel on top, more in metros, and the rider costs the same whether you send them out 300 times or 500 times. That is the strength and the weakness: at high volume the cost per delivery drops to ₹30 to ₹40, and at low volume it climbs past ₹100.
In-house also means you handle the hiring, the absences, the bike breakdowns and the accident insurance. The practical side is covered in how to hire and pay delivery staff.
Option 2: on-demand third-party riders
On-demand delivery partners let you book a two-wheeler for a single trip from an app or, at higher volume, through an API connected to your ordering system. You pay per trip, priced on distance, with a base fare for the first few kilometres and a per-kilometre rate beyond. Surge pricing applies at peak hours and in rain.
The cost is entirely variable. A 3 km trip in most cities is in the range of ₹40 to ₹80 depending on the city, time and provider; check the app for your location. There is no salary when it is quiet, and no limit on riders when it is busy. The trade-off is that the rider has never seen your kitchen before, may take 10 to 20 minutes to arrive for pickup, and carries only your one order.
The main types of provider are compared in detail in third-party delivery partners for restaurants: on-demand, hyperlocal and courier options.
Option 3: aggregator fleet
When an order comes through a food delivery app, the app's rider delivers it. You do not book, dispatch or pay a rider directly; the delivery is bundled into the commission, which is commonly 18 to 30 percent of the order value, and the customer may also pay a delivery fee to the app.
This works well for discovery orders from new customers. It works badly as your only delivery model, because every repeat order from a regular customer pays the same commission, and because you cannot use that fleet for orders that come through your own WhatsApp number or store page. The cost comparison between the two is in own delivery vs delivery app delivery.
The three options side by side
The table uses a kitchen doing 600 deliveries a month within 4 km at an average order of ₹350. Rider costs are typical ranges rather than quotes; your city will differ.
| Factor | In-house riders | On-demand riders | Aggregator fleet |
|---|---|---|---|
| Cost structure | Fixed: salary plus fuel | Variable: per trip by distance | Percentage of order value |
| Cost per delivery at 600 orders | About ₹35 to ₹60 with two riders | About ₹50 to ₹80 per trip | ₹65 to ₹105 of commission on a ₹350 order |
| Monthly delivery cost | ₹30,000 to ₹40,000 | ₹30,000 to ₹48,000 | ₹40,000 to ₹63,000, plus the lost customer relationship |
| Pickup wait | None, rider is on site | 10 to 20 minutes typical | Set by the app |
| Peak-hour capacity | Capped by headcount | Scales, at surge prices | Scales, app controls assignment |
| Multi-order trips | Yes, you batch | Usually one order per booking | App batches, sometimes with other restaurants |
| Who the customer sees | Your brand | The courier's brand | The app's brand |
| Works for your own direct orders | Yes | Yes | No |
| Management effort | High: hiring, rosters, incidents | Low: book and track | None |
The pattern is clear. Below roughly 10 to 15 deliveries a day, on-demand riders are cheaper and far less effort. Above 25 to 30 a day within a tight radius, an in-house rider or two becomes cheaper per order and gives you batching and a known face at the door. In between, most kitchens run a hybrid.
The hybrid most kitchens end up with
A common shape for a kitchen doing 20 to 40 deliveries a day looks like this.
- One or two in-house riders cover the predictable lunch and dinner core within 3 km, batching two orders per trip where addresses line up.
- On-demand bookings handle overflow at peak, far addresses beyond the core radius, and the rider's weekly off.
- Aggregator orders continue for discovery, with every delivered order carrying a reason to order direct next time.
The hybrid depends on knowing which orders are near and which are far before you assign a rider, which is why delivery zones matter so much. Set them by distance with charges that reflect what the trip costs you, as described in how to set delivery zones and charges by distance. And know what a small order actually costs to deliver before you promise free delivery on it; see what a 200 rupee order really costs to deliver.
Five questions before you choose
- How many deliveries a day, and how bunched? Count last month's orders by hour. Two riders idle for twenty hours a day is the most common way kitchens overspend.
- What is your real radius? Map the last 200 addresses. If 80 percent are inside 3 km, build your model around that core and treat the rest as on-demand.
- Who takes the call when a delivery fails? With in-house riders it is you and you can fix it in minutes. With on-demand it is a support chat. With aggregators it is largely out of your hands.
- Can your ordering system talk to the courier? Rebooking every order by hand in a courier app costs staff time at exactly the busiest moment. An API connection or a dispatch screen removes that.
- Where do the orders come from? If most are direct, on your own number or store page, you need your own delivery answer anyway. If most are from apps, start by growing the direct share first.
One way to run the direct side is a WhatsApp ordering platform such as Tezzu, where the order, the address, the delivery zone charge and the status updates sit in one dashboard so your rider or courier booking starts from a complete order rather than a chat screenshot. The setup steps are in the setup playbook.
Set up delivery zones before you set up riders
The playbook walks through zones, charges, free-delivery thresholds and order statuses, so whichever rider model you pick starts from clean orders.
Read the setup playbookFrequently asked questions
What is hyperlocal delivery in the restaurant business?
Hyperlocal delivery is delivering orders to customers within a short radius of the kitchen, usually a few kilometres, within an hour of ordering. It is the model behind every food delivery app and every restaurant that sends out its own rider. The word distinguishes it from courier shipping, where packaged goods travel across cities over days.
Is it cheaper to have your own delivery boy or use on-demand delivery services?
It depends on volume. An in-house rider is a fixed monthly cost, so the cost per delivery falls as orders rise and climbs when they fall. On-demand services charge per trip, so they are cheaper below roughly 10 to 15 deliveries a day and more expensive above 25 to 30 a day within a tight radius. Many kitchens run both.
How far should a restaurant deliver?
For hot food, most kitchens cap delivery at 5 to 7 kilometres, with the bulk of orders inside 3 km. Beyond that the food arrives cooler, the rider's round trip eats into the next order, and the delivery cost rises faster than the charge customers will accept. Set distance-based zones and charge more for the outer ring.
Can I use delivery app riders for my own WhatsApp orders?
Not through the regular restaurant partnership; their fleets deliver orders placed on their apps. For orders that come through your own number or store page you need your own riders or an on-demand courier. That is the main reason restaurants moving orders to a direct channel also set up a delivery model of their own.
How do I manage deliveries without a delivery app?
You need three things: a clean order with a confirmed address, a way to assign it to a rider or book a courier, and status updates the customer can see. A WhatsApp ordering platform with delivery zones and order statuses covers the first and third; the rider or courier covers the middle. A notebook and phone calls work at five orders a day and break at thirty.