A cafe owner told us she had spent four months on delivery and had finally worked out that her best-selling dish did not survive the journey. The reviews had told her before the numbers did.
Collection and delivery look like the same decision with a different radius. They are not. They have different economics, different failure modes, and one of them is a logistics business you did not intend to start.
What each one actually costs
Collection costs you almost nothing extra. The food is made the way it is made, the customer comes to you, and the only new thing is a way to take the order and a place to put the bag. Your margin is the normal margin minus the payment fee.
Delivery costs you the food, the driver and the risk. Whether you use your own driver or a courier, someone is paid per drop, and that cost lands on every order regardless of its size. It is why a delivery business with a low average order value can be busy and lose money at the same time, and why the arithmetic in our restaurant commission calculator carries a per-order fulfilment cost rather than percentages alone.
The risk is the part nobody prices. A cold or late order is your reputation, not the courier's, and the customer reviews you.
Start with collection, and mean it
Most independent places should switch collection on first and leave it running alone for a month. It is the version that cannot lose money, it teaches you what your online ordering actually does to the kitchen, and the customers who use it are usually local regulars.
If collection is quiet, delivery will not save it. That is worth knowing before you add drivers to the problem.
The settings that decide whether either works
These are the ones that get set once and then quietly govern everything:
- Preparation time. The honest number, not the optimistic one. It sets the customer's expectation and it is the single biggest driver of complaints when it is wrong.
- Minimum order for delivery. The number below which a drop cannot pay for itself. Work it out from your fulfilment cost rather than copying the place down the road.
- Delivery fee. Charging it is not the deterrent people fear. Hiding it until checkout is.
- Opening hours, and whether you are open now. An ordering page that takes an order at 22:40 for a kitchen that closed at 22:30 has created a refund and an angry customer.
- A pause switch. Saturday at eight, the kitchen is drowning, and the ability to stop taking orders for twenty minutes without pulling the page down is worth more than any feature on this list.
- Scheduled orders. Letting somebody order at four for collection at seven flattens your peak, which is the cheapest capacity you will ever add.
The menu is different online
A menu built for a table and a menu built for a phone are not the same document. Photographs matter more, descriptions matter more, and the choices a waiter would have asked about have to be on the screen: size, sides, spice level, extras that carry a price. Our guide to putting your menu online covers structuring it, including the dietary tags and the option groups.
One thing worth doing before you switch anything on: order your own food. Both ways. It takes an evening and it finds the things no configuration screen tells you.
And then take the direct orders seriously
The point of running your own ordering is that the order arrives without a commission attached. That only pays if people use it, which means the delivery apps cannot be the only place your regulars know to look - the argument in stop paying delivery marketplace commission, and the reason our own ordering system exists.
Frequently asked questions
Should a restaurant offer collection or delivery first?
Collection, in almost every case. It adds no fulfilment cost, it cannot lose money on a small order, and it teaches you what online ordering does to your kitchen before drivers are involved. If collection is quiet, adding delivery will not fix that - it will add cost to the same problem.
What does delivery actually cost a small restaurant?
The food, plus a flat cost per drop for whoever delivers it, plus the reputational risk of a late or cold order arriving with your name on it. That per-order cost is what makes a low average order value unprofitable no matter how busy you are, which is why it belongs in the arithmetic rather than being treated as a percentage.
What is a sensible minimum order for delivery?
Work it out rather than copying it: the point where the order still makes money after the food cost and the cost of the drop. Setting it too low is how a busy delivery service loses money on every third order, and setting it honestly filters those orders towards collection instead.
How do I stop taking online orders when the kitchen is overwhelmed?
Use a pause rather than taking the page down. On a Saturday night the ability to stop new orders for twenty minutes and then resume is worth more than most features, because the alternative is accepting orders you cannot cook on time and refunding them afterwards.