Delivery operations: zones, fees and drivers

Delivery looks like free growth: extra orders without extra tables. Then, a year in, many restaurants discover that delivery is 40% of revenue and 5% of profit.

The problem is not delivery. It is running delivery as if it were an extension of the dining room.

First: zones are drawn in minutes, not kilometres

A 5 km radius on a map is 12 minutes in one direction and 35 in another, where a railway or a ring road gets in the way.

Draw your zones with three questions:

1. How many minutes does it take at peak, not at three in the morning?
2. Does the food survive the trip? Fried items lose their quality in 15–20 minutes.
3. Does the order’s revenue cover the cost of the trip?

Then set three numbers per zone: delivery fee, minimum order, and expected time. More importantly, have the system block out-of-range orders automatically, instead of a staff member apologising after the customer has paid.

Second: assignment timing

The costliest mistake in delivery is assigning at the wrong moment:

• Too early: the driver waits twenty minutes at the door — paid time producing nothing.
• Too late: the food is ready and cooling while it waits for someone to carry it.

The working rule:

assign at = expected ready time − driver travel time to the restaurant

Which requires the delivery side to know the ticket’s status in the kitchen. When the kitchen display and driver dispatch sit in one system, that calculation becomes automatic rather than a human guess under pressure.

Third: batching

A driver carrying two orders in one direction nearly doubles their output. But batching has limits:

• Do not batch more than two unless they are in the same building or street.
• Do not batch a hot order with one that needs a long stop.
• Never delay the first order by more than 7–10 minutes.

The second order must not fund itself out of the first order’s rating.

Fourth: cost delivery separately

This is the number that changes decisions. A delivery order is not a dine-in order with an address.

A 49 difference per order. At 100 delivery orders a day that is around 147,000 a month. This does not mean stopping delivery — it means pricing and running it deliberately.

Fifth: delivery metrics

Orders per driver hour is the metric that exposes scheduling problems: too many drivers in a quiet hour is fixed cost buying nothing.

Sixth: customers are waiting for information, not only food

Most delivery complaints are not about time. They are about uncertainty. A customer who knows their order arrives in 35 minutes is calmer than one waiting 25 minutes with no information at all.

Live order tracking, a notification when the driver leaves, and an honest ETA cut complaints more than adding drivers does.

In GoSufra the customer follows their order live in your own branded app, the driver is tracked by GPS with pickup and drop-off timestamps and a delivery rating — all of which comes back to you as reports rather than impressions.

An execution list

1. Redraw your zones using peak-hour travel time.
2. Set a fee, a minimum and an expected time per zone.
3. Tie driver assignment to kitchen status.
4. Calculate delivery margin after packaging and commission.
5. Track orders per driver hour weekly.
6. Give customers real tracking instead of “your order is being prepared”.