Delivery apps keep 15% to 30% of your sales (and your customer)
You sell more than last year and the bank account doesn't show it: commission, promotions and packaging take the margin. And the customer.
In this article
Last month you sold 320 orders through delivery apps. Revenue was up 18% on the previous year, and your bank account never noticed.
When you open the breakdown, the story makes sense: 28% commission, a free-delivery promotion you funded yourself, packaging that went up in price and three cancelled orders nobody paid you for. The dish that leaves 40% margin in the dining room leaves 12% on the app.
And the hardest part to accept: of those 320 customers, you don't have a single phone number.
What an app really charges you
The commission is the visible part. The real cost of selling through a delivery platform has five components:
| Line item | Example on a $30 order | Who pays |
|---|---|---|
| Platform commission (15–30%) | $7.50 | The restaurant |
| Delivery promotion or discount | $2.00 | The restaurant, in full or in part |
| Packaging and bags | $1.20 | The restaurant |
| Food and prep cost | $9.60 | The restaurant |
| Cancellation or incident fees | Varies | The restaurant |
All in, the $30 order leaves about $9.70 before staff, rent and utilities. If the commission were 30% and the promotion bigger, the margin gets close to zero.
Three details almost nobody checks in the contract:
- What the commission is calculated on. Some platforms apply it to the subtotal, others to the total including tax and delivery.
- Who pays for promotions. A free-delivery offer is usually split, and the restaurant's share isn't always spelled out.
- Incident fees. A cancelled or disputed order can cost you the dish and the commission.
The math on a single dish: app vs. direct
The same dish, three channels, three different results. With a selling price of $15 and an ingredient cost of $4.50 (if you don't have that number, start with the real food cost per dish):
| Line item | Dining room | Delivery app | Direct order |
|---|---|---|---|
| Selling price | 15.00 | 15.00 | 15.00 |
| Platform commission | — | −4.50 | −0.60 |
| Packaging | — | −0.80 | −0.80 |
| Food cost | −4.50 | −4.50 | −4.50 |
| Gross margin | 10.50 | 5.20 | 9.10 |
On the direct channel, the $0.60 commission covers the payment gateway and the ordering platform. The difference between $5.20 and $9.10 per order is the margin the app keeps for acting as the middleman.
At 300 orders a month, moving just 20% to the direct channel is 60 orders × $3.90: about $234 a month, nearly $2,800 a year. Illustrative figures, but the structure repeats in any restaurant with volume.
The real problem isn't the commission
The commission hurts, but it's a known cost. What you don't see is what you lose on every order:
- You don't have the contact. No phone number, no email, no purchase history.
- You can't reach them. You can't announce a new dish or invite them in on a slow Tuesday.
- The app competes with you inside your own order. At checkout, it suggests other restaurants.
- The customer belongs to the platform. If it offers them a discount at the place across the street tomorrow, they leave, and there's nothing you can do about it.
You own the kitchen, but not the audience. And a restaurant without an audience depends on whoever brings it customers, at whatever commission they set.
The three-layer strategy
This isn't about breaking with the apps: they're still a powerful discovery channel, especially for new customers. It's about no longer depending on them.
1. Capture at every touchpoint
- In the bag: an insert with a QR code and a welcome offer for the direct channel (a dessert, a drink, free delivery on the first order placed with you).
- On the dine-in receipt: the same invitation, with a concrete reason to sign up.
- In the order confirmation: a message that thanks them and offers the direct channel for next time.
- On Google and social media: the direct order link visible, not buried in the footer.
An honest warning: some delivery platform contracts prohibit including advertising for your own channels on the packaging. Check your agreement before you print a thousand inserts.
2. A frictionless direct channel
Direct ordering competes with an app that already has the card on file. If your process asks them to create an account, confirm by WhatsApp and pay cash, you lost before you started.
- A clear menu, with decent photos and final prices.
- Card or wallet payment, not just cash.
- Automatic order confirmation and estimated time.
- A pickup option: it removes the logistics problem and captures the customer anyway.
3. A customer list you can actually use
Collecting data without using it is just piling up forms. The list works when it's segmented:
- By frequency: those who order every week, those who ordered once and never came back.
- By day: those who order on Tuesdays, those who order on Fridays.
- By preference: family, vegetarian, office, celebration.
- By ticket size: those ordering for one and those ordering for four.
With that in place, a Tuesday promotion goes only to the people who usually order on Tuesdays, and the message stops being noise.
How to migrate without losing volume
- Don't switch the app off. Keep it as an acquisition channel while you build your own.
- Pick one product. Start with your most ordered and most profitable dish: the one people already search for by name.
- Give them a reason to switch. A lower final price, a free item, free delivery or direct service. Make the difference visible.
- Test it on slow days. Tuesdays and Wednesdays, when you need volume and your kitchen has capacity.
- Measure one thing. Direct orders per week and average ticket. If they don't grow in four weeks, the problem is your ordering process, not the customer.
What to automate and what not to
The repetitive part has three moments: capturing the contact, tagging it and reaching out again at the right time. That's what a CRM does.
With GoHighLevel, for example, you can create a QR form for the bag, tag contacts automatically by dish or by day, and launch SMS, WhatsApp or email campaigns with order links. It also lets you measure which campaign generated orders.
What it doesn't do, and it's worth saying: it is not a POS, it doesn't manage delivery drivers or delivery logistics, it doesn't sync automatically with Uber Eats, Rappi or Glovo, and it doesn't replace your ordering site if you already have one that works. It's the customer relationship layer, not the operations layer.
What to measure
| Metric | How it's calculated | What it tells you |
|---|---|---|
| % of sales through apps | App sales / total sales | Your level of dependency |
| Effective commission | Commissions paid / app sales | More real than the nominal rate |
| Margin by channel | (Sales − variable costs) / sales | Where to push volume |
| Direct orders per week | Count | The real progress of your own channel |
| % of the list captured | Contacts with consent / orders | Whether capture is working |
| Repeat purchase at 60 days | Customers with 2+ orders / customers | Whether the relationship exists or only the transaction |
Common mistakes
- Competing with the app on price. If your direct channel is cheaper because you took margin out of the dish, you didn't win: you traded one problem for another.
- Permanent discounts. They teach customers to wait for the promotion and destroy your reference price.
- Capturing without permission. Without explicit consent, the list is useless and your campaigns land you in legal trouble.
- Ignoring the operations of your own channel. A direct order handled badly costs more than a commission: you lose the customer for good.
- Not measuring the effective commission. The contract rate isn't what you pay: promotions and fees push it higher.
- Sending the same promotion to your whole list. It's the fastest way to get people to unsubscribe.
Frequently asked questions
Should I leave the delivery apps?
Not all at once. They're a discovery channel, and for many restaurants they account for a significant share of volume. What's worth doing is lowering your dependency: a growing share of orders coming through a channel you own.
How long does it take to build a direct channel?
The first version — menu, card payment, automatic confirmation — can be ready in one or two weeks. Adoption is what takes time: three to six months for a significant share of customers to change their habits.
Is it legal to ask app customers to order direct?
It depends on your contract with each platform. Many prohibit including your own advertising on the packaging or contacting the customer outside the app. The safe route is to capture through the channels you do control: your dining room, your website, your social media and your own loyalty program.
Do I need my own ordering system?
To start, no: an order form with payment and an automatic confirmation is enough. When volume grows, a dedicated ordering system will give you control over timing, inventory and the kitchen.
What about delivery logistics?
Start with pickup: it removes the problem and still captures the customer. Then evaluate your own drivers or third-party couriers by zone. Logistics is a separate business, and it's not worth improvising.
Does this work if my restaurant doesn't do delivery?
Yes, the logic is the same: build your own customer list that you can contact. In that case, the focus is reservations and repeat visits, not orders.
Four-week plan
- Week 1: calculate how much you paid in commissions last month and what your real margin was by channel.
- Week 2: set up capture: QR code, welcome offer and a form with consent.
- Week 3: launch the direct channel in its minimum version (menu, payment, confirmation) and test it with ten customers.
- Week 4: launch the first campaign on your slowest day and measure direct orders and average ticket.
The commission is paid on every order. A customer list is built once and used many times.
