Every restaurant operator has had this argument, usually at midnight with a calculator. The delivery apps bring orders you would not otherwise have, and take a quarter to a third of the value. Your own ordering page takes nothing, and also brings nothing unless somebody sends people to it. Both statements are true, which is why the argument never resolves.
Run the actual numbers on one order
Take a typical delivery order and work down. Suppose it is 2,000 in your local currency, with food cost at 30%, packaging at 4%, and a marketplace commission of 28% plus payment processing.
| Line | Via aggregator | Direct order |
|---|---|---|
| Order value | 2,000 | 2,000 |
| Commission | −560 | 0 |
| Payment processing | −40 | −40 |
| Food cost (30%) | −600 | −600 |
| Packaging | −80 | −80 |
| Delivery cost | included | −150 |
| Contribution | 720 | 1,130 |
Even after paying your own rider, the direct order contributes over fifty percent more. Two direct orders are worth roughly three aggregator orders. That is the whole argument in one line, and it is why the direct channel is worth building even if it starts slowly.
But the aggregator brings the customer
Yes - the first time. The mistake is treating aggregator orders as a permanent arrangement rather than an acquisition cost. Think of that 28% as marketing spend for a first order, and then work deliberately to move the repeat business to your own channel.
- Put a card in every aggregator delivery bag with a QR code and a genuine reason to order direct next time.
- Make the direct price a little better, or the portion a little more generous. Customers notice.
- Capture the phone number on the first direct order and use it - sparingly and usefully.
- Give direct customers loyalty points that aggregator orders do not earn.
The in-house QR menu is the easy win
Table-side QR ordering is where most restaurants see an immediate return, and it has nothing to do with delivery. Guests order when they are ready rather than when a server is free, which reliably increases the number of items ordered - particularly the second drink and the dessert that a table would otherwise skip because catching someone's eye felt like too much effort.
“We introduced QR ordering to save on printing menus. The average check went up fourteen percent and we did not change a single price.”
Do it properly or do not bother
A bad QR menu is worse than a paper one. A PDF that opens at 20% zoom and requires pinching is a guaranteed complaint. The bar for a digital menu in 2026 is genuinely high:
- 1It must load in under two seconds on a mid-range phone with a weak signal.
- 2It must not require an app, an account or a login.
- 3Photography must be real and current - stock imagery reads as dishonest.
- 4Sold-out items must disappear the moment the kitchen says so.
- 5Reordering must not require rescanning the code.
- 6Splitting the bill has to work, because the alternative is nine people doing mental arithmetic.
Keep both, on purpose
The mature position is not abandoning the aggregators. It is running both channels through one kitchen queue so your staff work from a single screen, treating the marketplace as paid acquisition, and moving repeat customers to a direct channel that costs you nothing per order. Over a year, shifting even a third of your delivery volume to direct changes your profitability more than any menu price rise you would dare to make.
Want this handled by software rather than a spreadsheet?
SumPOS does most of what this article describes automatically - and the free plan is genuinely free, so you can test the idea before committing to anything.