Cloud Kitchen Margins per Brand: Costing After Delivery Commission

Start with the average order subtotal on the app, then subtract commission, packaging, promotions you fund, error charges and food cost. What is left is contribution per order; divide it by the subtotal for the margin. Do this for each brand separately, because packaging and error costs barely change with price, so cheaper baskets lose more margin points.

Two things usually surprise owners when they first work out a cloud kitchen profit margin brand by brand. The brand with the best food-cost percentage is not always the best earner, because packaging and error charges cost about the same on a $9 basket as on a $22 one. And the commission plan moves the result more than most recipe tweaks: every five points of commission costs $1.10 on a $22 order.

Follow me on Instagram@sagnikteaches

Six lines between the app price and what you keep

Every order on a marketplace app passes through the same lines, whatever the platform calls them. Your sheet needs one row for each, and the platform's statement tells you where to find four of the six. Here are the lines, with the headings DoorDash uses on its payout and monthly statement:

Connect on LinkedInSagnik Bhattacharya

Other platforms use different headings, so match each heading on yours to one of the six lines and add a seventh only when something genuinely does not fit. Packaging and food cost are the two lines no platform can fill in for you, which is why they are the ones most often left out of a first margin figure. If food cost is the weak spot in your own sheet, building a recipe costing spreadsheet with ChatGPT covers it.

Subscribe on YouTube@sagnikteaches

The commission itself is a contract matter. DoorDash's pricing page lists three plans for delivery orders: Basic at 15% of the order subtotal, Plus at 25% and Premier at 30%, with pickup at 6% on every plan. It also says there is no separate payment processing fee on its marketplace orders. Uber Eats' pricing page says its Lite plan starts at a 15% marketplace fee on delivery orders, shows that rate only for selected markets and prints no rate for Plus or Premium. Treat those as list rates in USD. Your contract, any introductory period and your plan's extras decide what you really pay, and so does whether card processing sits inside the commission or is charged on top.

Three brands from one kitchen: the inputs

The kitchen here has four people, one shared line and three virtual brands (smash burgers, rice bowls and dessert pots), all on one app at the 25% Plus rate. Before the owner built a brand sheet, the only margin figure in sight was the blended one: $85,860 of monthly sales across 5,000 orders and a vague sense that things were "about 30%". The inputs below are illustrative, but each one comes from somewhere you can look up in your own account.

• Subtotal: the last four to eight weeks of sales for that brand divided by its orders. Eight weeks smooths out one odd promotion.
• Food cost: each dish's recipe cost, weighted by the mix of dishes that actually sold, not by the menu's top three.
• Packaging: lay out one complete pack for each brand and cost every piece, including the cutlery, napkin and tamper-evident seal that never make it onto the menu costing.
• Promotions: the customer-discount and marketing-fee lines from the statement, divided by that brand's orders.
• Error charges: the Error charges and Adjustments lines divided by orders. DoorDash describes error charges as covering reimbursements issued to customers for missing or incorrect items.
• Prep minutes: time ten orders with a stopwatch, or use ticket-to-ready times if you have a kitchen display.

If the brands share a single store listing, split the statement by item first. The sheet only works when each brand's averages describe that brand alone.

The margin waterfall for each brand at a 25% commission

Look at the last two rows against the margin. On food cost alone the dessert pots are the best brand and the rice bowls the worst. After the other lines the order changes: burgers lead, dessert pots drop to second because packaging eats 11.5% of a small basket, and rice bowls stay last on a double hit of high food cost and a packaging bill of 8%. The commission is the same 25% on all three. What differs is how much is left once the fixed-ish costs per order have been paid.

Multiply by monthly orders and a second pattern appears. The three brands together bring in $85,860 and $26,030 of contribution, a blended 30.3%. Rice bowls provide 30% of the orders but only 21.6% of the contribution, and burgers provide 48% of the orders and 65.9% of the contribution. The owner's feeling that the kitchen ran at about 30% turned out to be right for the blend, which is exactly why it hid the problem. Judged by order count, rice bowls look like the kitchen's second pillar; judged by margin, they are its weakest brand.

Prompt: build the sheet with formulas you can audit

Give the assistant the definitions along with the numbers. Without them it picks its own meaning of "margin", and the answer looks tidy either way. Ask for formulas as well as results, because a formula can be audited and a pasted number cannot. For the spreadsheet side, the habits in using ChatGPT to write Excel formulas carry over to any sheet.

I run a cloud kitchen with three brands on one delivery app. Build a per-order margin table for each brand.
Definitions:
- subtotal = average order value before tax and tips
- contribution = subtotal - commission - food cost - packaging - promotions I fund - error charges
- margin = contribution / subtotal (not contribution / cost)
Commission is a percentage of subtotal. Calculate it at 15%, 25% and 30%.
Inputs per order, USD:
burgers: subtotal 22.00, food 6.60, packaging 1.30, promotions 1.10, errors 0.35
rice bowls: subtotal 15.00, food 5.10, packaging 1.20, promotions 0.90, errors 0.30
dessert pots: subtotal 9.60, food 2.20, packaging 1.10, promotions 0.70, errors 0.25
Output: (1) one table per brand showing every line; (2) the spreadsheet formula for each line,
assuming the five inputs sit in B2:B6 and the commission rate in B7; (3) a list of every assumption.
Do not round until the last line. If an input is ambiguous, ask me instead of guessing.

Illustrative output (contains a slip)
Burgers at 25% commission
  Subtotal              22.00
  Commission            -5.50
  Food cost             -6.60
  Packaging             -1.30
  Promotions            -1.10
  Contribution           7.50
  Margin                34.1%
Contribution formula: =B2-B2*B7-B3-B4-B5
Assumptions: commission applies to subtotal only; tax and tips excluded;
promotions are an average per order, including orders with none.

Add up the lines yourself: 22.00 - 5.50 - 6.60 - 1.30 - 1.10 is 7.50, and the error-charge line is missing from both the table and the formula, so contribution is 35 cents too high and the margin should read 32.5%, not 34.1%. That is the cheapest check in the whole process: the lines must sum to the result. A second slip to watch for is dividing contribution by total costs instead of by subtotal, which would show the burger brand at 48.1%. That is a markup, and it flatters every brand. Third, read the assumptions list. The one that says "tax and tips excluded" should match how your platform actually calculates commission.

What the commission plan does to each brand

Each cell shows contribution per order in dollars and the margin on subtotal. Five points of commission is always 5% of the subtotal, so it costs $1.10 on a burger order, $0.75 on a rice-bowl order and $0.48 on a dessert order. The margin points move equally across the brands; the dollars do not, and dollars are what pay for labour and rent. At 30% the rice bowls leave $3.00 an order, so a single fully reimbursed $15 order cancels the contribution of five normal ones.

A higher plan is sold on reach (DoorDash's page lists a wider delivery radius on Plus and the widest range on Premier), and the test is arithmetic. Moving the burger brand from 25% to 30% costs $1.10 on each of its 2,400 monthly orders, $2,640 in total. At $6.05 contribution per extra order, the move needs 437 more orders a month, an 18% increase, just to break even. If the platform cannot credibly promise that, the plan does not pay. The same sum works in reverse for a cheaper plan: giving up reach is worth it when the saved commission on your existing orders outweighs the orders you lose.

A $5-off offer, costed line by line

Suppose you fund a threshold offer: $5 off orders of $25 or more. The typical qualifying burger order has $26.00 of items, food at 30% ($7.80), packaging of $1.30 and expected error charges of $0.35, on the 25% plan.

DoorDash's promotions page says the commission rate is applied to the order subtotal less free items and discounts, which is why commission drops to $5.25 in the right-hand column. Read how your own contract treats it. If commission were charged on the full $26.00, contribution on the offer would fall to $5.05. The same page gives a promotion's cost as the customer discount plus a DoorDash marketing fee but does not price the fee, so this table leaves it out and the real figure is lower still.

To earn the same total contribution as before, you would need 59.5% more qualifying orders (10.05 divided by 6.30). If commission is charged on the full subtotal, you would need almost twice as many (10.05 divided by 5.05). The bar is that high because a $5 offer on a $26 basket is a 19% discount, and it lands on every order that uses it, including orders from customers who would have bought anyway. A smaller offer, or one that only applies in your quietest daypart, costs far less per order. The forecasting side of that choice is in forecasting prep by delivery platform, where promotions are flagged and their uplift measured.

Error charges: the line nobody budgets

An error charge appears when a customer is reimbursed for something missing or wrong, and it is easy to ignore because each one is small. Across the three brands the kitchen pays $1,565 a month (2,400 × $0.35 plus 1,500 × $0.30 plus 1,100 × $0.25). Halving the rate saves about $780 a month and needs no change of supplier.

Start by reading the reasons behind each charge for a month. If you can paste them into an assistant with customer details removed, ask it to group them by cause and by item; you may find that a handful of items, such as sauces, drinks or extra sides, account for most of them. Then fix the cause at the pass: a one-line checklist taped above the bagging station (items, sauces, cutlery, seal), a photo of the sealed bag on larger orders, and separate bags for hot and cold items. None of that costs more than seconds per order.

Raising app prices on the weakest brand

The rice bowls are the obvious candidate. Raise their app prices by 8% and track what happens to the same lines. Promotions are assumed to stay at 6% of the subtotal, because they are percentage discounts.

If the 1,500 monthly orders hold, contribution rises by $1,245. The orders can fall by 18% (one minus 3.75 divided by 4.58) before the kitchen is worse off, which is a generous cushion, but customers do not always respond smoothly. Run the change on its own for a few weeks and judge it against the forecast baseline for that brand and app, not against last month's gut feel. Whether app prices should sit above prices elsewhere is a decision with its own trade-offs, set out in whether takeaway menu prices should rise on apps; read the pricing section of your platform contract first.

Margin per prep minute can reverse the ranking

So far the sheet has ignored the people cooking. Add labour by allocating it on prep minutes. At an assumed $24 an hour fully loaded, a minute costs 40 cents; use your own rate. The table adds one more row that matters at peak: contribution per prep minute, before labour.

After labour the rice bowls keep only $1.95 an order, 13.0% of the subtotal. And when the line is saturated, say on a Friday at 19:00, each minute spent on a rice bowl earns 83 cents while a minute on a dessert pot earns $1.48. That does not mean dropping the rice bowls. It tells you which brand deserves the promotion slot, and the next free pair of hands, when two brands compete for the same minutes. If prep minutes are the constraint, rank by margin per minute; on a quiet Tuesday, rank by margin per order.

Reconcile one week against a real statement

A sheet that has never met a statement is a guess with good manners. Pick one brand and one week, count its orders, run the lines forward and compare the predicted net total with the statement's. DoorDash defines its net total as sales minus commission and fees and marketing spend, plus amendments. Here is the burger brand for a week of 120 orders, with the statement figures filled in:

In this illustration the statement agrees with the sheet to the dollar, and the contribution of $7.15 an order matches, which real statements rarely do on the first attempt. If your statement's net total differs from what the model predicts for that week, there is a line you have not mapped: ad spend, an adjustment, a fee on a pickup order or a tax line. A gap of more than 1% or 2% of sales is worth chasing, starting with Adjustments and Marketing fees. DoorDash's statement also shows tax on commission and on merchant fees; whether that tax is recoverable is a question for your accountant, and until you know, model it as a cost.

Four weeks from first export to a price decision

1. Week one: download four to eight weeks of statements and order data, split them by brand and strip customer details. Work out the averages in the inputs table.
2. Week two: cost the packs and the recipes, build the sheet with the assistant, check that the lines sum, and reconcile one week against the statement.
3. Week three: choose one lever for the weakest brand. Options include an app price rise, a cheaper or lighter pack, a smaller promotion, a recipe adjustment, a check at the pass to cut error charges, or a different commission plan.
4. Week four: compare orders and contribution with the baseline you recorded in week one. Keep the change, reverse it, or try the next lever.

Change one thing per brand at a time, otherwise you will not know which lever worked. And keep the sheet alive: the next time a supplier raises a price or the platform changes a fee, you will be changing one input rather than starting again.

Questions about brand margins and commission

Is commission charged on tax and tips?

Platform pages describe commission as a percentage of the order subtotal and do not always spell out how tax and tips are treated. Test it on your own numbers: divide one week's commission by that week's sales. If the ratio comes out above your plan's rate, something else is being charged, and your account contact should be able to explain what.

Should brand margin include kitchen labour and rent?

Do both. Contribution margin, before labour and rent, tells you which brand to fix or promote. A margin after labour, using prep minutes per order, tells you whether the kitchen pays its way. Share rent by prep minutes or by orders, and never let a weak brand hide inside a blended figure for the whole kitchen.

What margin should a cloud kitchen brand aim for?

There is no universal figure. Work backwards from your own costs: add up monthly rent, labour not already allocated, software and the profit you want, then divide by monthly sales. That percentage is the contribution margin your brands must average. A brand that sits below it needs a price, recipe, packaging or plan change.

How often should I redo the brand sheet?

Redo it whenever the commission plan, a packaging supplier, a major ingredient price or a menu changes, and check it against a real statement once a month. Between changes the sheet only needs new averages from the latest four to eight weeks of orders, which takes about twenty minutes with an assistant.

Can AI read my delivery statement for me?

It can map the line names on a statement you paste in to the lines in your sheet, once customer details are removed. It can also misread columns, so always compare its totals with the statement's own net total. Use it for the tedious sorting and the formulas, and keep the final reconciliation as a human check.

Further reads

• Are Ghost Kitchens Still Worth It? A Cash-Flow Test You Run With AI — Run a cash-flow test on whether the whole kitchen still pays.
• Restaurant Menu Price Increase: Test Each Change by Dish With AI — Test a price rise dish by dish before rolling it out.
• Restaurant Labour Cost Percentage: Test Rota Changes With AI First — Test rota changes against labour cost before you commit.
• AI in Hospitality Management: Catch Supplier Price Creep on Invoices — Catch supplier price creep that quietly erodes food cost.
• How to Check Your Margins Product by Product With AI — A general method for checking margins product by product.
• What Is a Ghost Kitchen? How an Existing Restaurant Can Test One — How an existing restaurant can test a ghost kitchen first.
• AI for a Cloud Kitchen: Keep Every Brand's Menu Consistent on Apps — Keep every brand's menu and prices consistent on each app.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: DoorDash merchant pricing page, promotions page and payout and monthly statement learning-centre page; Uber Eats merchant pricing page. All brand figures in the examples are illustrative.