Take the order on your own site, at your own prices.

You already know the apps take a cut. Most owners have done the sensible thing about it and raised the price on the app so the cut comes out of the order instead of out of the kitchen. Nothing wrong with that. It is the obvious move and I would make it too.

Here is the part that catches people. The commission is charged on the subtotal as it appears on the app, and that subtotal is the price you raised. The markup does not sit outside the fee. It sits inside it, and it drags the fee up along with it.

So the same markup can be a win or a loss, and which one it is comes down to a single number you agreed to when you signed up and have probably not looked at since.

Now run the same plate down the other lane.

The customer paid six dollars less and you kept thirty-eight cents more.

That gap is the obvious argument for taking your own orders, and it is not the biggest one. The biggest one is further down this page and it has nothing to do with the money on the night.

This is what I go through before I build one, and most of it has nothing to do with the customer. It is about the twenty minutes on a Friday when everything goes wrong at once.

• It reaches the kitchenA thermal printer on the pass, a screen by the phone, or both. If your cooks have to learn something new in the middle of a rush, it has already lost.
• You accept it, and you set the timeThe customer gets twenty-five minutes instead of silence, and you get a second to look at the board before you commit to it.
• You can turn it offOne tap pauses ordering when the kitchen is buried, and the page says so instead of quietly taking food you cannot cook. It is the first thing an owner asks me about and the last thing most order buttons have.
• You can sell out of somethingOne tap marks an item gone until tomorrow, gone until you say otherwise, or off the menu entirely. Without that, the site keeps selling the thing you ran out of at seven.
• The menu knows what time it isLunch that appears at eleven and goes away at three. A weekend section. A catering list that only exists where somebody asked about catering.
• It takes money the way people payCard, Apple Pay, Google Pay, and a tip line if you take tips. It pays out to your account on your processor's normal schedule. Your account, in your name, and it stays yours the day you decide you are done with me.
• It can hold an order for SaturdayThe catering order somebody places on a Tuesday for the weekend is the biggest ticket of your week, and it is the exact one a plain order button cannot take.
• The ticket reads at arm's lengthItems, add-ons, and a line for whoever bags it to check against. Every wrong bag is a review you have to work back off later.

None of that is clever. It is the difference between a button that takes orders and something that survives a Saturday.

On a delivery order you get a first name and a masked number that stops working once the driver is gone. No email. Nothing you can text on a slow Tuesday.

An order placed on your own site arrives with a name, a number, an email if they gave you one, and a record of what they ordered and when. That is not a marketing feature. It is the difference between a customer and a transaction, and it is the one thing you cannot buy back later at any price.

What it lets you do is small and boring and it works. The message that goes to somebody who used to order every week and has not ordered in six. The review ask that goes out while they are still happy rather than three days later. The note about Saturday that reaches two hundred people who have already handed you money once.