GrabFood and foodpanda commissions in Malaysia: what delivery really costs your restaurant

Ask ten Malaysian F&B operators what GrabFood or foodpanda takes per order and you will get ten different answers, because there is no published rate card: every commission is negotiated, tiered and confidential. What exists instead is a decade of operator testimony, third-party guides and one very public complaint to the government, and they all point the same way: the headline percentage is painful, and the real cost is higher than the headline.

This article puts the reported numbers in one place, with their sources and dates, looks at what the Uber takeover of foodpanda's parent company means for Malaysian restaurants, and then does the arithmetic the platforms would rather you skipped: the contrast between a delivery order that gives away a quarter of the bill and a dine-in booking that gives away nothing.

The short version:

• no delivery platform in Malaysia publishes its commission rates; everything below is reported, negotiated and dated;
• a May 2026 third-party guide puts GrabFood at roughly 15-22%, foodpanda at 12-20% and ShopeeFood at 10-18%, depending on tier and volume;
• operators interviewed across 2025-2026 describe quoted rates of 25-30%, and a true cost of 28-35% of delivery revenue once packaging and paid visibility are counted;
• in 2021, a restaurant association formally asked the government to intervene, saying the big two demanded 30% or more;
• Uber agreed in July 2026 to buy foodpanda's parent, Delivery Hero, with closing expected in the second half of 2027: your delivery partner's owner is changing again;
• dine-in bookings taken directly carry 0% commission, forever, and the guest data stays yours.

What GrabFood, foodpanda and ShopeeFood reportedly charge

Start with the most recent third-party figures. A guide to delivery commission fees in Malaysia published by klikit on 1 May 2026 reports these ranges:

• GrabFood: about 15-22%, depending on the subscription tier the outlet signs;
• foodpanda: about 12-20%, with volume-based discounts;
• ShopeeFood: about 10-18%, positioned as the cheaper challenger.

Treat all three as reported estimates, not prices you can hold anyone to. The platforms negotiate per merchant, chains get better terms than a single kopitiam, and the rate you are offered can change with your tier, your exclusivity and the promotions you accept. The only number that matters is the one in your own merchant agreement, read carefully, including the clauses about marketing contributions.

And operator testimony runs higher than the guide ranges. The MenuBase State of Malaysian F&B 2026 report, built on operator observations gathered across 2025 and 2026, says platform commissions are "typically quoted at 25-30% of order value". Both sets of numbers can be true at once: the lower ranges describe negotiated base rates on paid tiers, the higher ones describe what a small independent is actually quoted when it signs up without leverage.

The true cost is higher than the commission line

The commission is only the visible part. The same MenuBase operator interviews itemise what sits around it:

• packaging: RM1.50 to RM3.50 per order, paid by you, worthless to you the moment the rider leaves;
• paid visibility: RM300 to RM1,000 a month in boosting and promotional spend that operators describe as effectively mandatory to stay visible in the app's listings;
• discount dilution: platform campaigns funded partly or wholly from your margin.

Add it up and MenuBase's estimate is that the effective cost of aggregator sales runs 28-35% of delivery revenue, with outlets whose average delivery order sits below roughly RM32-RM35 structurally losing money on those orders. Those are one firm's interview-based estimates, not an audit, but they match what operators have been saying publicly for years.

The public record goes back to the lockdowns. In July 2021, the Restaurant and Bistro Owners Association (PPRB) asked the government to intervene on commission fees, with its vice-president telling Malay Mail that Malaysia's two biggest delivery platforms demanded "30 per cent or upwards" and calling for rates to be brought down to 15% while dine-in was banned. The Malaysia Competition Commission said it was closely monitoring the platforms later that month. Five years on, no commission cap exists, and the rates are still unpublished.

Uber is buying foodpanda's parent: platform risk is real

Here is the 2026 development worth planning around. On 16 July 2026, Uber agreed to acquire Delivery Hero, foodpanda's German parent, for about US$14.8 billion, in a deal that includes foodpanda Malaysia and is expected to close in the second half of 2027, subject to shareholder and regulatory approval.

One clarification, because the two deals are constantly conflated: Grab did not buy foodpanda Malaysia. Grab bought foodpanda's Taiwan business only, for US$600 million in March 2026. In Malaysia, foodpanda's future owner is Uber.

Nobody knows yet what Uber will do with Malaysian commission structures, merchant contracts or the foodpanda brand, and anyone who claims to is guessing. The lesson is not a prediction, it is a principle: when your main sales channel can change owner, pricing and policy overnight, and you hold none of the customer relationships flowing through it, that is not a channel you control. It is shelf space you rent.

Let's be honest: delivery earns its place

None of this means you should quit the platforms, and we would be the wrong people to pretend otherwise. GrabFood and foodpanda put your menu in front of hundreds of thousands of Malaysians you could not reach on your own, with riders you do not have to hire, insure or schedule. For a new outlet, the apps are discovery; for a mamak-style late-night operation, they monetise hours when the dining room is quiet; for many kitchens, a delivery order at 28-35% true cost is still incremental revenue on top of a rent and payroll that were already paid for by dine-in.

The honest framing is portfolio, not boycott: delivery is a high-commission channel that buys reach, and it stops being dangerous the moment it stops being the only channel you have. The operators MenuBase interviewed are already acting on that logic: it estimates aggregators at about 22% of F&B revenue in its 2026 channel mix, with established venues deliberately shifting volume toward channels they own.

The channel the platforms cannot touch: direct dine-in bookings

Now run the comparison that actually matters for a restaurant with tables. A RM200 dinner for four pax ordered through a delivery app leaves you roughly RM130-RM145 after the true costs above. The same four pax seated in your dining room, having booked directly, leave you RM200, plus drinks, plus dessert, plus the chance to turn them into regulars, and the booking itself cost you nothing.

That is not an accident of pricing, it is structural: a reservation is not a marketplace transaction, so there is nothing to take a percentage of. With a reservation system whose free version takes 0% commission, the economics are permanent: unlimited direct bookings, no per-cover fee, no percentage of the bill, ever. The free version of ViteUneTable is a full booking system, not a trial. The paid tier, the Pack Standard at 29 € a month excluding tax (about RM134 at early-October 2026 rates; the price is in euros, so factor in the exchange rate as you would with any USD-billed platform), adds automatic email confirmations and reminders, and the Anti No-Show pack at 49 € adds card guarantees for the bookings you cannot afford to lose.

The deeper difference is data. A delivery platform's customer is the platform's customer: you see an order number, the app owns the relationship, and it will happily show your regulars a competitor's promo tomorrow. A direct booking gives you the guest's name and email, their visit history and their preferences, a list you keep even if you change systems. If you are weighing up the options, our comparison of restaurant reservation systems in Malaysia covers the whole field, from UMAI to Oddle to the marketplaces.

Practical ways to depend on the platforms less

No preaching, just mechanics. Each of these moves volume from a 28-35% channel toward a 0% one without touching your delivery listings:

1. Claim and arm your Google Business Profile. Most Malaysians find restaurants on Google and Maps before any app. A complete Google Business Profile with photos, hours and a booking link turns that discovery into direct bookings instead of app orders.
2. Put a booking link everywhere the platforms are not. Instagram bio, TikTok, WhatsApp auto-reply, the printed receipt. Every booking that bypasses an aggregator is margin recovered.
3. Convert delivery customers into dine-in guests. A card in the bag inviting them to book a table directly costs sen and attacks the commission at its root: the platforms monetise distance, your dining room monetises presence.
4. Measure your channel mix monthly. Revenue per channel, true cost per channel. Operators who audit their stack recover real money; MenuBase puts typical annual recoveries from stack audits at RM2,000-RM6,000.
5. Be careful with discount marketplaces too. Commission thinking applies beyond delivery: dine-in marketplaces take their cut differently, as our look at Eatigo's per-diner fee model for Malaysian restaurants shows. The question is always the same: who owns the guest, and what does the channel really cost?

Frequently asked questions

What commission does GrabFood charge restaurants in Malaysia?

GrabFood publishes no rate card. A third-party guide from klikit dated 1 May 2026 reports roughly 15-22% depending on subscription tier, while operators interviewed by MenuBase across 2025-2026 describe quoted rates of 25-30%. Rates are negotiated per merchant, so the only authoritative number is the one in your own agreement.

What commission does foodpanda charge restaurants in Malaysia?

Also unpublished. The same klikit guide reports roughly 12-20% with volume-based discounts, and the same operator interviews put quoted rates at 25-30%. foodpanda Malaysia's parent, Delivery Hero, agreed in July 2026 to be acquired by Uber, with closing expected in the second half of 2027, so terms may evolve under new ownership.

Is ShopeeFood cheaper than GrabFood and foodpanda?

Reportedly yes at the headline level: klikit's May 2026 guide puts ShopeeFood at roughly 10-18%, the lowest of the three. The true cost still depends on packaging, promotional spend and your average order value, and a lower percentage of a smaller order volume may not beat the bigger platforms' reach.

Did Grab buy foodpanda in Malaysia?

No. Grab bought foodpanda's Taiwan business only, for US$600 million in March 2026. foodpanda Malaysia is part of Delivery Hero, which Uber agreed to acquire on 16 July 2026, in a deal expected to close in the second half of 2027.

Should my restaurant leave the delivery platforms?

For most outlets, no. Delivery buys reach and incremental revenue that a dining room alone cannot, especially off-peak. The sensible goal is balance: keep delivery for what it does well, measure its true cost honestly, and grow direct dine-in bookings, which carry 0% commission and leave the guest relationship with you.

How much commission do restaurants pay on direct reservations?

Zero, if the booking system is commission-free. A reservation is not a marketplace order, so there is no bill to take a percentage of. ViteUneTable's free version takes unlimited direct bookings with 0% commission and no per-cover fees; paid packs add email reminders and no-show protection, priced in euros at 29 € and 49 € a month excluding tax.