FIELD NOTE · HONG KONG F&B · DELIVERY-PLATFORM COMMISSION
← THE FIELD NOTES
// FIELD NOTE :: PLATFORM COMMISSION

How to avoid delivery app commission in Hong Kong

Keeta and Foodpanda take a quarter to a third of every order. You cannot always beat that on price. You can stop letting it decide your margin.

How much commission do delivery apps take in Hong Kong?

In Hong Kong, delivery platforms take roughly a quarter to over a third of every order. Keeta raised its rate from 25% to 28% for restaurants that would not sign an exclusive deal, and Foodpanda sits in the same band.

The exact number depends on your deal, but the Competition Commission put the range on record: commissions run from about one-quarter to over one-third of the order value, and restaurants that stay off exclusive contracts pay the higher end (Competition Commission of Hong Kong, 2023). That gap is real money: Keeta pushed one Kwun Tong restaurant from 25% to 28% for refusing to sign an exclusive deal (Hong Kong Free Press, Apr 2025).

That is before you count the rest. The Consumer Council checked six ordering apps and found the same dishes priced up to 85% higher than the in-store takeaway price, once you add in the markups and fees (SCMP, Feb 2024). Platforms pass some of that to the customer. A large slice still comes out of your kitchen.

One Shau Kei Wan owner put it to the South China Morning Post as his restaurant wound down: takeaways already account for 30 per cent of our business, and after the high commissions we are left to rely on thin profits and quick turnover. (SCMP, 2026) That is the trap. More volume, less margin, and no relationship with the customer at the end of it.

Why did delivery commissions get worse after Deliveroo left?

Deliveroo closed its Hong Kong business in early April 2025, leaving Keeta and Foodpanda as the two real options. With one fewer platform competing for your restaurant, the Federation of Restaurants warned commissions would climb.

Keeta, run by mainland giant Meituan, launched in Hong Kong in 2023 and was the number one platform by order volume inside a year (TechNode, May 2024). When Deliveroo pulled out in April 2025 (SCMP, Mar 2025), the market went from three players to two.

Fewer players means less reason for any one of them to keep your rate low. The president of the Hong Kong Federation of Restaurants said as much at the time: with a competitor gone, the remaining platforms have room to push commissions up (RTHK, Mar 2025).

You cannot control that. What you can control is how much of your business depends on it.

How do you actually avoid delivery app commission?

You do not have to leave the apps. Keep them for reach, and move your regulars onto a channel you own, where there is no commission on the order.

The apps are good at one thing: finding you a customer you would never have reached. That is worth paying for, once. The problem is that they charge you the same cut every time that customer comes back, forever, and they keep the customer’s details so you can never reach them directly.

So split the job. Let the platforms do discovery. When someone orders, you have a chance to bring them across to a channel that costs you nothing per order: your own WhatsApp. A regular who used to cost you 28% now costs you nothing to serve, and you can message them on a slow Tuesday instead of hoping the app surfaces you.

Most kitchens keep the apps running for reach and quietly shift the regulars over. Every order that moves is margin you keep. You are not fighting the platforms. You are just refusing to rent your best customers from them.

Is there a Keeta commission alternative?

The real alternative to a platform is not another platform. It is your own ordering channel, usually WhatsApp, where the customer, the phone number, and the order all belong to you instead of to Keeta or Foodpanda.

Swapping Keeta for Foodpanda just moves the same cut, a quarter to a third of the ticket, to a different logo. The alternative that changes your margin is an owned channel: a WhatsApp number where guests order direct, their details stay yours, and no one takes a percentage of the ticket.

There is a legal tailwind here, and it is recent enough that most menus have not caught up to it yet. When the Competition Commission investigated the platforms, it forced them to drop the clauses that used to stop restaurants offering cheaper prices on their own direct channels (SCMP, Nov 2025). You are now free to make ordering direct from you the better deal for the customer, and the platform cannot contractually punish you for it.

That is the whole game. Same food, a door you own, and a price you set.

What does moving off the apps look like for a small restaurant?

You keep the apps for discovery and build a WhatsApp ordering channel for your regulars. Most of the escape is just giving people who already like your food an easier, cheaper way to order direct.

It is not a rip-and-replace. It is a second door, next to the platform door, that keeps the customers you already earned. The owned channel answers enquiries, takes the order, and remembers who the customer is, so the next order comes straight to you.

Before I build any of it, the numbers have to be worth it. That is what an AI Assessment is for. In two business days I read your own platform payout statements back to you, put a dollar figure on what commission is actually costing you this year, and rank the fix against everything else leaking money. If the gap is too small to bother with, I will tell you, and it will not go on the plan.

If it is worth building, the build is priced on the value it brings back, measured on numbers we both read off documents you already own. You can see the full picture of how this works for a Hong Kong kitchen on the restaurants and cafes page, or book the free 15-minute Mini and I will point you at one tool worth trying first.

NEXT STEP · BOOK A CALL OR WRITE TO ME
// NEXT STEP :: PICK YOUR DOOR

Start with the Mini AI Assessment.

Four or five plain-English questions about how your week actually runs, and you leave with one tool worth trying, named and explained. Fifteen minutes, free. If nothing fits yet, I'll say so.

  • · 15 minutes, free, and the tool is yours whether we work together or not
  • · No deck, no homework, nothing to prepare
  • · "Not yet" is a fine answer, anytime