Own your online ordering

Delivery apps solved a real problem. They brought customers who would never have found you, they handled payment and logistics, and during the years when nobody walked in they kept a great many restaurants alive. That is worth remembering before anyone calls them the enemy.

But the commission on an order can reach around 30% once delivery and service fees are counted, and it is charged on every order — including the regular who has eaten with you for six years and only uses the app because it is the easiest way to order.

The point is not to leave the platforms

It is to stop paying discovery rates for customers you have already discovered.

Somebody searching your restaurant by name is not being introduced to you. That introduction happened years ago, in your dining room. Routing that person through a third party costs you close to a third of the ticket in exchange for something you did not need.

First-party ordering is where those customers should land. Keep the platforms for reach. Keep your regulars for margin. Most operators who do this well are still on the apps a year later — they have simply moved the people who were never being acquired.

What first-party ordering has to get right

It has to be as easy as the app. On a phone, in under a minute, with no account required to complete a first order. Forced registration is where more first-time orders die than anywhere else.

It has to know when you are closed, and when an item has run out. Nothing burns goodwill faster than taking money for something you cannot make, and nothing generates more phone calls.

It has to reach the kitchen the way your staff already work — printed, or on the screen they already watch. An ordering system that requires someone to remember to check a laptop will be forgotten on the first busy Friday.

And it has to take payment reliably, because a failed payment at eight on a Friday is worse than having no online ordering at all. The customer does not conclude that the payment provider had a problem. They conclude that you are unreliable.

Give people a reason to switch

Habit is the hard part, not the software. Someone who has ordered through an app forty times will not change because you would prefer it.

Put the link on every receipt, every bag, every table, every reply to a review. Make a small discount available only when ordering directly — smaller than the commission, so it still pays. And tell people plainly that ordering from your site keeps more of the money in the restaurant. Customers respond to that far more than most owners expect, particularly locally, and particularly in a city where people have opinions about their restaurants.

The objection worth taking seriously

The usual one is delivery: the platforms have drivers and you do not.

That is true, and it is why leaving the platforms outright is usually a mistake. But a large share of orders in most restaurants are collection, and collection needs no driver at all. Start there. It is the part of the business you can move without solving logistics, and it is often a third or more of the online volume.

If delivery matters, the platforms increasingly sell it separately — you keep the customer relationship and pay for the driver, rather than paying a commission on the whole ticket for both. The rates are not trivial, but they are a different number from the one you are paying now, and worth pricing rather than assuming.

The second objection is that customers will not change. Some will not. You are not trying to move everyone — you are trying to move the regulars, who are the ones costing you the most in commission and the least in acquisition.

What it costs to run yourself

Direct ordering is not free, and it is worth seeing the real number rather than only the commission you are avoiding.

Card processing is the main one, typically under 3% including the fixed per-transaction fee. The ordering software itself is usually a monthly figure in the low hundreds, or bundled into a managed site. Someone has to keep the menu current, which is minutes a week rather than hours, and someone has to watch for orders during service, which is the part that actually needs a decision about who.

Against a 30% commission, the arithmetic is not close at volume — the fixed monthly cost stops mattering somewhere around the first few hundred dollars of direct orders. Below that it is roughly a wash, which is the honest answer for a very small operation and the reason not to promise anyone this is free money.

The first month decides it

Direct ordering does not fail because the software is bad. It fails because it launches quietly and nobody notices it exists.

Every regular has a habit, and the habit is the app. Breaking it takes a specific prompt at a specific moment: a card in the bag with the next order, a line on the receipt, the person at the counter mentioning it while handing over the food. Whatever you choose, it has to happen on every order for several weeks, not on the launch day and then whenever somebody remembers.

It also has to be someone's actual job. "We will mention it" is not a plan, and during service nobody mentions anything. Pick the person, pick the moment, and check after a fortnight whether it is still happening — it usually is not, and that is the whole reason these launches stall.

The first month is also when the ordering flow gets its only honest test. Order from your own site as a customer would, on a phone, on a busy evening. Most of the friction that costs you orders is found in the four minutes that takes.

What to measure

Two numbers, monthly. The share of orders coming direct rather than through a platform, and your average direct ticket against your average platform ticket.

The second one surprises people. Direct orders often run higher, because your own menu is not competing with four hundred other restaurants on the same screen, and because upsells work better when the customer is already committed to you rather than still browsing.

If neither number moves after three months of promoting it properly, the problem is the ordering flow, not the idea. Sit with a customer and watch them use it.