Two orders come in for the same $30 meal. One arrives through a delivery app. One arrives on your phone line. They look identical on the kitchen ticket. They are not identical on your bank statement.
That gap is the whole story. It's worth doing the math once, properly, so you know exactly what you're trading away every time an order lands on an app instead of your own line.
What the app actually takes
Delivery platforms charge a commission on every order that runs through them. The exact number varies by platform, by plan tier, and by whether you're paying for marketing placement on top, but restaurants commonly report commissions landing somewhere in the 20% to 30% range of the order total. On a $30 order, that's $6 to $9 gone before you've paid for food cost, labour, or a box.
Run that across a month of orders and it's not a rounding error. It's a second rent payment, except this one scales with your sales instead of staying fixed.
None of this makes the apps predatory. They built real infrastructure: a marketplace of hungry people browsing on their phones, delivery drivers, a checkout flow. That infrastructure costs something to run, and the commission is how they get paid for it. The question isn't whether the fee is fair. It's whether every order needs to pay it.
The cost that doesn't show up on the invoice
The commission is the visible cost. The bigger one is invisible.
When someone orders through an app, the app owns the relationship. It has their name, their order history, their payment method, their delivery address. You get an order number and a ticket. You don't get a customer you can text a promotion to, remember by their usual order, or win back with a call when they haven't ordered in a month, because you were never given their contact information in the first place.
That customer thinks of the app as the thing they used to order from you, and next time they're hungry, the app is what opens on their phone. The app decides which restaurants they see first. It can feature you today and bury you tomorrow, and there's very little you can do about it, because you never built a direct line to that person.
A phone order flips all of that. You get the number. You get to recognize a repeat caller. You get to be the thing they think of directly, not the thing an algorithm surfaces for them.
Apps aren't the enemy, they're the front door
This isn't an argument to quit the apps. For a lot of restaurants, they're genuinely the best tool available for one specific job: getting found by someone who has never ordered from you before. That's discovery, and it's worth paying for.
The mistake is letting every order, forever, run through the discovery channel. A first-time customer finding you on an app makes sense. That same customer ordering their fifth meal from you through the same app, still paying the same commission, still leaving all their data with a platform that isn't you, doesn't make sense. By the fifth order they know your name. They should be calling it.
The play is simple to state and takes real discipline to run: let the apps do what they're good at, which is bringing new people in the door, and actively move repeat customers to a channel you own. A card in the delivery bag with your number on it. A receipt line that says "call direct next time and skip the wait." A short mention when they pick up in person. None of it is complicated. It just has to happen consistently.
What you keep, channel by channel
Here's the same order, roughly, on each channel.
| App order | Phone order | |
|---|---|---|
| Commission | 20-30% typically leaves with the platform | None |
| Customer contact info | Kept by the app | Kept by you |
| Repeat order likelihood | Controlled by app placement | Controlled by your relationship |
| Who gets remembered next time | The app | You |
The food cost and labour are the same either way. The difference is entirely in what leaves the building and who owns the next order.
The catch: the phone only works if someone answers
Here's where the plan usually breaks. You print the cards, you tell customers to call direct, and then the phone rings during a Friday rush and nobody's free to grab it. 85% of restaurant calls go unanswered during peak hours, which is exactly when you most want that direct-channel math working in your favour.
A caller who gets no answer doesn't wait around loyal to your brand. They open the app they already have installed and order there instead, commission and all. You built the direct channel and then handed the order right back to the platform, because the one thing a direct line needs to actually work is someone on the other end of it, every time, including 8 p.m. on a Friday.
Restaurants that fix this see it show up in the numbers. Voice AI has been shown to cut missed calls by 87%, and businesses using it report revenue increases of up to 22%, partly from calls that used to go nowhere and partly from the ticket itself: on-call upsells lift the average order by as much as 25% when whoever's answering the phone always remembers to ask about the extra topping or the side.
Try it on your own line
Sempera answers your phone line instantly, every call, day or night, so the direct channel you're trying to build actually holds up under pressure. It takes the order, asks the upsell question every time, and warm-transfers anything tricky straight to your team. A flat monthly fee, never a per-order cut.
Try it on your own line: 7-day free trial with all features. Cancel anytime. No credit card required. sempera.org