A customer orders $30 of food from you on a delivery app. (US delivery tickets usually run $25–35, so I'll use $30 throughout.) By the time that order lands in your bank account, you keep somewhere between $18 and $21. Every owner knows the commission line. Far fewer have added up the charges around it, and that gap is what decides whether delivery earns you money or just keeps your kitchen busy.
Commission is only the starting number
US marketplace plans in 2026 come in tiers. As of mid-2026, DoorDash lists Basic at 15%, Plus at 25%, and Premier at 30%, with pickup at 6%. Uber Eats covers the same 15–30% range, roughly 20% on its cheapest plan, 25% standard, 30% at the top, and it raised those rates in March 2026. The tier is less optional than it looks. Pick the basic one and your restaurant sits below dozens of competitors who paid for placement, so most restaurants that live on marketplace volume end up at 25–30%.
On our $30 order at the 30% tier, commission comes to $9.00. That's the number everyone argues about. The charges that decide the order sit underneath it.
Four costs that don't appear in the pitch deck
Start with payment processing. Some platforms fold it into the commission; others add 2.9% + $0.30 on top. On a $30 order that's about $1.17.
Then the promotions you "choose" to join: free-delivery campaigns, 20% off a first order, sponsored placement. The platform runs the campaign and the restaurant funds the discount. Operators who track this say promo participation adds 5–10% of revenue in costs during the months they run. Averaged out, call it $1.80 on our order.
Refunds are the third line, and you don't control them. When a marketplace courier delivers cold food an hour late, the customer gets a refund that usually gets charged back to you. Industry surveys put refund leakage at 1–3% of marketplace revenue. Say $0.60.
One more that won't go into the math. The usual defense is to raise menu prices 15–20% on the apps. On paper it recovers the commission; in practice it costs you twice. Your conversion drops because your prices look expensive next to competitors who didn't mark up, and "overpriced" one-star reviews start following you around. That cost is real, but it hides inside lost volume, so I'll leave it out of the tally. Just don't count the markup as pure recovery.
Effective cost on this order: 41.9%. Put food cost at 30% and labor at 25%, and this marketplace order lost money before you even boxed it up. Even on the 15% tier with no promos, the platform takes more per order than your whole net margin on a typical ticket.
We'll walk through your numbers on a 30-minute call. No slides, just a calculator.
What the 30% actually buys, and when it's worth paying
Credit where it's due. Marketplaces are genuinely good at one thing: putting your food in front of people who have never heard of you. As a paid customer-acquisition channel, 30% of a first order isn't crazy. Restaurants pay similar acquisition costs through ads all the time.
The trouble starts with the second order, then the tenth. That repeat customer costs you the same 30% every time, forever, and the platform holds their name, phone, and order history. You can't message them or reward them for coming back. For the platform, that's the product working exactly as designed. For you, it's renting your own regulars back at full price. We covered what losing that customer data actually costs in a separate piece.
Where the math flips: acquire on the apps, keep them on your own channel
The restaurants that make delivery pay don't quit the apps. They change what the apps are for. Let the marketplace handle discovery, the first order from a stranger. Move the repeat orders onto a channel you own, an ordering website or a branded app, where the only per-order costs are payment processing and your own delivery.
The conversion numbers are what make the effort worth it. Across the 3M+ orders processed on our platform, branded restaurant apps convert visitors to buyers at up to 35% and well-built ordering sites reach 15%. A typical restaurant listing buried in a crowded marketplace converts at roughly 3%. Someone who installs your app has all but committed to the next order already. Add a cashback program and push campaigns and the repeat loop keeps running without ad spend.
Run the same $30 order through your own channel and the lines shrink: processing $1.17, delivery either your courier's real cost or a flat fee passed to the customer, commission $0. You keep about $28 instead of $17.43. Ten repeat orders from a single regular is a $100+ swing, per customer, per quarter.
Do this next
Pull last month's marketplace statements and work out one number: total deductions divided by gross marketplace sales. Not the headline commission rate, the whole ratio. If it comes in above 30%, then every marketing insert, every QR card in the bag, and every "order direct, get 10% cashback" flyer pays for itself on the second order. The step-by-step version is here: moving delivery-app customers to your own channel.
Branded app, ordering site, delivery management, and marketing — launched in about two weeks.