The True Cost of Third-Party Delivery: A 2026 Margin Audit

There’s a number that gets thrown around in restaurant trade press: third-party delivery commissions are about 30%. It’s not wrong, but it’s also not the whole story, and the story is the interesting part. We talked to twelve independent restaurant owners across five states and pulled six months of their aggregator settlement statements apart. Here’s what we found.

The visible costs

The line items most owners can quote from memory:

• Marketplace commission — the headline rate. DoorDash’s public tiers run roughly 15%, 25%, and 30% depending on plan. Uber Eats and Grubhub structure things similarly. In our sample, the effective rate (after promotions and fees) ranged from 22% to 31%.
• Delivery commission — when the platform handles courier dispatch, this is bundled into the marketplace fee. When you do your own delivery via the platform’s “self-delivery” mode, it drops dramatically — 6%–12% in our sample.
• Customer-facing fees — service fees, small-order fees, and delivery fees that the customer pays. These don’t come out of the restaurant’s revenue, but they do dampen demand.
• Tablet rental — $0–$30/month. Often $0 in year one and a small line item thereafter.

The invisible costs

This is where the calculation gets uncomfortable. Below are the costs that owners we interviewed had to be prompted to remember, but which showed up clearly when we did the math.

Promotional spend you can’t opt out of

Most platforms run a continuous loyalty or promotional engine — “DashPass,” Uber One, “free delivery” banners — and those discounts are funded, in part, by the restaurant. If you opt out, your listing gets de-ranked in the app. Three of our twelve owners had tried opting out. All three opted back in within a quarter.

Refund and chargeback risk

When a courier hands a customer the wrong bag, the platform refunds the customer and clawback comes out of the restaurant in roughly 60% of cases. Across our sample, refund clawbacks averaged 1.4% of gross sales — small, but real, and largely outside the restaurant’s control.

Tip leakage

When a customer tips on the app, the courier gets it. When a customer tips on the bag at pickup, your staff gets it. The fraction of orders that come through aggregators are the fraction of tips your staff doesn’t see. One of our owners mentioned that her front-of-house turnover doubled the year she went heavy on DoorDash, and she thinks tips were 60% of the reason.

Customer ownership

This is the cost most owners don’t price because it’s not on a statement. When a customer orders through DoorDash, they are DoorDash’s customer, not yours. You don’t get their phone number, you can’t text them about your new menu, and if DoorDash adjusts its algorithm, you can’t reach them to defend yourself.

A working number

If we add up the visible and invisible costs, our twelve-owner sample suggests that the all-in cost of third-party marketplace orders is about 33%–38% of gross — call it a third, plus or minus a few points.

At a typical 20% food-cost, 30% labor-cost restaurant, you can do the arithmetic in your head: a third of every aggregator dollar is gone before you start making the food. That doesn’t mean aggregators are a bad deal. Customer acquisition is real. Discoverability is real. But it does mean you should know what you’re paying for, and it does mean the math changes a lot if you can move some of that volume direct.

What “direct” actually costs

Here’s the part of this conversation that’s usually missing. When restaurant owners hear “cut out the middleman,” they sometimes hear “build your own app,” which is a great way to spend $80,000 and get nothing. The realistic alternatives in 2026 are:

• A direct web ordering page with payment processing. Stripe, Toast, Square, and a dozen others can power this for processing fees of 2.9% + 30¢ plus a small monthly platform fee. All-in cost: ~5%–7% of gross on the orders that flow through it.
• SMS / chat ordering. A flat monthly platform fee plus carrier costs. All-in cost: ~6%–10% on a typical small operator.
• Phone calls. Still the highest-margin channel by a wide margin, and still the channel small operators are worst at staffing.

The interesting strategic question isn’t whether to leave the aggregators — most operators can’t, and shouldn’t. It’s how to gradually move the channel mix so that the all-in blended cost drops a few points. Going from 0% direct to 30% direct, on the same revenue, is the difference between a restaurant that breaks even and one that puts money in the bank.

The honest counterargument

There are restaurants for whom heavy aggregator dependence is the right answer. Ghost kitchens that have no physical foot traffic. New openings without a customer base. Operators in dense urban markets where DoorDash is the discovery layer. If your fixed costs are low and the aggregators bring you orders you would not otherwise get, the 33% is fine.

The trap is the opposite case: a restaurant with a loyal local following, in a walkable neighborhood, that has gradually let aggregators become 60% of revenue because each individual order seemed worth taking. That restaurant has effectively rented its customer relationships back from a platform, and the rent goes up every year.

The action item, if you want one

Pull your last six months of aggregator statements. Add up the headline commission. Then add 4–5 points for the invisible costs we discussed. Then ask: of the orders that came through that channel, what percentage of those customers do I have any way to reach directly? If the answer is below 20%, you have a channel-mix problem, not a restaurant problem. The good news is, the channel mix is fixable.