How to cost your delivery by channel and build a profitable direct channel (step-by-step 2026)

Step 0: understand what the channel takes before costing anything

Before costing a single dish, you must see the full hole. Platforms charge between 15% and 30% per order: on a $25 order at 25% that is $6.25 before food and labor. But the contract commission is not everything the channel takes. Add packaging of $0.60 to $1.40 per order, refunds on incomplete orders of 3% to 6%, and a VAT many owners miscalculate on gross instead of net. The real app channel cost runs 33-38% of the ticket, not the 25% you think you pay. This Masterestaurant guide starts here because costing on the contract commission instead of the effective one is the mistake that ruins half of all delivery redesigns. Diego F. Parra says it bluntly: you cannot patch a hole you have not measured, and the hole is almost always 10 points bigger than the owner imagines.

Which commission should I cost my app price on?

You must cost on the real effective commission, not the contract one: take the last 90 days, add commission paid, packaging, and refunds, and divide by gross channel sales.

That percentage — typically 32-38%, not the nominal 25% — is the correct base. If you cost on the 25% contract figure, you leave 11 points of cost uncovered on every order and sell below cost without knowing it. Do it per platform, because each app has its own effective commission and that decides where you push volume. In practice, a group that believed it paid 24% discovers it pays 35% once packaging and a 4% refund rate are added. Masterestaurant lets no client move forward without this number measured per app: it is the foundation of all per-channel costing, and skipping it guarantees the week-two reprice falls short and delivery keeps draining cash.

Cost every dish twice: once for dine-in, once for the app

The heart of this guide is costing each dish per channel, not by average. A dish with 30% food cost leaves 70% gross margin in the dining room, but the same dish on the app at 33% effective commission leaves barely 37% before labor. That is why the app price rises 15% to 30% over the dine-in price to keep equivalent margin, according to DoorDash / Uber Eats (2025). Respect the Masterestaurant hard rule: 32% food cost is the maximum, never call it recommended, and it measures ingredients only. Commission is subtracted when costing the delivery channel because it is a direct variable cost of the order; payroll and rent are not, because they belong to the break-even point of the business. Reprice only the dishes that can absorb the increase, using per-dish profitability analysis to choose. This step alone usually moves the channel margin from -$1.80 to $4.10 per $25 order, without touching the kitchen or hiring anyone.

Why isn't payroll charged to the delivery plate?

Payroll is not charged to the plate because it is a fixed cost of the business, not a variable cost of the order: it belongs to the monthly break-even point, calculated separately.

This is the most common costing error and the one that most confuses owners redesigning delivery. To the delivery plate you subtract only what varies with each order: food cost, effective platform commission, and packaging. Payroll, rent, and utilities are covered by the accumulated contribution margin of all orders, not by loading each plate with an arbitrary share. Confusing these two layers leads to raising app prices wrongly — inflating dishes that did not need it — or subsidizing delivery with the dining room without noticing. Masterestaurant always separates the two layers: commission to the plate costing, payroll to the channel break-even. That discipline is what keeps delivery from looking profitable per order while the group loses cash at month-end.

Build the delivery P&L apart or the dining room hides the hole

The third step pulls delivery out of the dining-room average with its own income statement: gross sales, total effective commission, packaging, specific food cost, and contribution margin per order, down to a separate break-even point. When delivery lives hidden in the global P&L, a channel losing 7 margin points disappears and the owner believes all is well because total cash rises. In Diego F. Parra's experience, delivery tends to destroy margin precisely when nobody measures it in isolation from the rest of the business. Calculate how many orders the channel needs to cover the payroll, rent, and riders it consumes: if your delivery does 3,000 orders at $4.10 contribution margin, it contributes $12,300 monthly to fixed costs. If that does not cover what the channel uses, the dining room subsidizes it. Seeing the channel alone is what lets you decide with data before growing further.

How do I build a profitable direct delivery channel?

You build a direct channel with an ordering website or WhatsApp connected to a payment gateway that charges 4-6%, versus the platforms' 25%, and you migrate the recurring customer with an incentive.

Do not abandon the apps: they bring reach and new customers you cannot win alone. The key is turning the recurring customer into a direct one with something simple, like 10% off their second order through your channel; even giving away that 10%, you still win against the app's 25%. Every migrated order recovers nearly 20 points of commission. Growing the direct channel's share of delivery in a matter of months is achievable with disciplined pricing and data, and at volumes of thousands of orders per month the effect shows up in money that used to vanish in invisible commission. The technical build is the easy part today: the hard part, and where the value sits, is designing the migration incentive so the customer prefers your channel without feeling less convenience.

The role of AI: dynamic per-channel pricing in the final phase

Artificial intelligence enters at the end of the guide, not the start: first you measure and split the P&L, then AI optimizes. Dynamic per-channel pricing cross-references demand by hour, commission by platform, and price elasticity by dish to decide which app prices to raise without scaring off orders. It is not raising everything 18% blindly; it is raising the dish whose demand can take it by 24% and leaving the price-sensitive one untouched. AI also runs real-time profitability analysis per platform and builds a delivery-optimized menu that prioritizes the highest-margin dishes after commission. Groups Masterestaurant supports with per-channel pricing raise the average app ticket 11% while losing under 3% of orders, a trade that almost always leaves net positive cash. But AI enhances per-channel costing; it does not replace it. Without the separate P&L from the earlier steps, the best algorithm optimizes on a wrong base.

Sustain the redesign: channel mix is a monthly board KPI

The redesign holds by turning channel mix into a monthly review KPI at the board level, not an adjustment you make once and forget. Measure what percentage of delivery goes through the expensive app at 25% and what percentage through the direct channel at 5%, the margin per order of each, and their month-over-month evolution. The mistake I see over and over is owners who build the direct channel, let it stall at 8%, and never push it; migrated volume stops and commission eats the cash again. Diego F. Parra repeats it in every engagement: profitable delivery is a monthly per-channel costing habit, not a 90-day project that ends. This week's action is a single one: measure your real effective commission per platform and start step one.