Online Ordering System: Cut Grab and Foodpanda Fees

Every F&B owner in Klang Valley has done the same sum. An order shows RM48 on the tablet, the payout report says RM33, and nobody can explain where the rest went. It is commission, promo co-funding and payment charges stacked together, and it repeats on every order for as long as you stay on that channel.

An online ordering system of your own does not make aggregators disappear. It gives you somewhere to send the customers who already know you, so the second and the twentieth order stop carrying a discovery fee. This guide covers food and retail ordering specifically: menu variants, delivery zones, kitchen tickets, e-wallets and the maths of running both channels. For appointments and time slots, our guide to an online booking system in Malaysia covers that different problem. ZenWeb builds these channels for Malaysian outlets, and gets called in when the first attempt sat unused behind a QR nobody scanned.

The video below covers setting up direct ordering before we put Malaysian numbers, payment rails and kitchen realities on it.

1. What a Direct Ordering Channel Replaces, and What It Does Not

Quick Answer: A direct ordering system replaces the transaction, not the discovery. Aggregators are still the cheapest way for a stranger to find you. Your own channel is where a customer who already knows your name should place their next order, without a third party charging you for the introduction again.

Most write-ups argue that restaurants should leave the apps. That advice loses money for a new outlet with no mailing list and no foot traffic. Separate the two jobs the apps do for you, then replace only the one you are overpaying for.

• Discovery — worth paying commission for. Someone who has never heard of you, browsing at 8pm between forty listings. That impression has real value and you will not buy it cheaper elsewhere.
• Repeat transactions — not worth paying commission for. A regular who orders the same set lunch twice a week. You already earned that relationship, so a discovery fee on order forty is the part that quietly eats the year.
• Customer data — you cannot buy it back. The apps keep the phone number, the order history and the delivery address. Without them, you cannot bring anyone back on purpose.

That split decides the whole build. Our web development services team scopes a first online ordering system around repeat customers only, because their economics change immediately.

Key takeaway: Do not build to leave the apps. Build to stop paying a discovery fee on customers you have already discovered, and let the aggregators keep doing the introductions.

2. The Commission Maths on a RM100 Order

Quick Answer: On an aggregator, a RM100 order commonly settles somewhere between RM65 and RM78 after commission, promo co-funding and payment charges. On a direct channel the same order settles around RM97 to RM98, because the only deduction is the payment gateway. The gap is the whole business case.

Owners usually know their headline commission rate and stop there. That rate is rarely what lands in the bank, so compare the full deduction stack. Menu mark-ups complicate it: raise app prices to cover commission and you raise the amount commission is charged on.

Net settlement on a RM100 food order across four Malaysian ordering channels, showing the typical total deduction, what the deduction is made up of, and the net amount received by the outlet.

Channel	Net received	Net (RM)	Deducted	What the deduction is
Aggregator, on promo		65	35%	Commission plus promo co-funding
Aggregator, standard listing		73	27%	Commission and platform charges
Aggregator, self-delivery plan		78	22%	Lower rate, you ride or arrange it
Own site, card or e-wallet		97	3%	Gateway percentage only
Own site, FPX or DuitNow		98–99	Flat fee	Fixed sen per transaction

Bars show the share of a RM100 order that reaches the outlet. Source: ZenWeb operational data, settlement reports reviewed with Malaysian F&B and retail clients, 2024–2026. Bands vary by outlet, negotiated rate and promo participation; treat them as the shape of the stack, not a quoted rate. Licence.

Run your own version on a real week: gross order value, minus what actually landed, divided by order count. The annualised figure usually exceeds the entire build cost of an online ordering system. The paid side of those channels is covered in Foodpanda in-app promos and Grab in-app advertising.

Key takeaway: Compare net settlement per order, not headline commission. The gap between roughly 22–35% deducted and roughly 1–3% deducted is what funds the build, and it repeats every order.

3. Building the Menu: Items, Variants and Add-Ons

Quick Answer: The menu is the hardest part of the build, not the checkout. Variants, add-ons, portion sizes, spice levels and time-limited items each change the price and the kitchen ticket. Model them properly on day one, because retrofitting a variant structure means re-entering the whole menu.

A restaurant menu is not a product list. One dish can carry four decisions before it reaches the kitchen, and each has to survive the trip from phone to printer. Get the structure right and updates take five minutes; get it wrong and staff go back to WhatsApp.

• Variants change the price. Regular and large, single and set, half chicken and whole. One item with priced options, never separate listings, or your reporting stops making sense.
• Add-ons stack. Extra egg, upgrade to iced, add cheese. Each needs its own price, its own line on the ticket, and a rule for how many can be chosen.
• Free choices still matter. Spice level, no onion, less ice. No price change, but the kitchen must see it, and it is what customers complain about when it goes missing.
• Availability changes hourly. Breakfast items, weekend sets, sold-out dishes. Staff need a one-tap toggle on a phone, not a login to a content system.

This is where a purpose-built ordering module beats a general e-commerce cart, which expects one product at a fixed price. If you also sell packaged goods, see our guide to an e-commerce website in Malaysia and the trade-off in online store versus marketplace. Operators with very large menus across several brands sometimes split menu data from the front end, the case for a headless CMS instead of standard WordPress.

Key takeaway: Model variants, add-ons and free choices before anyone designs a page. The menu structure decides both the customer experience and whether kitchen staff trust the tickets.

4. Pickup, Delivery Zones and Who Actually Rides

Quick Answer: Start with pickup. It needs no rider, no zone map and no delivery fee argument, and it converts the customers who already live nearby. Add delivery once volume justifies it, using postcode zones with a fee per zone and a minimum order that reflects the real cost.

Delivery is where first attempts stall. Owners promise island-wide delivery, find two orders a night cost RM18 each to fulfil, and quietly switch it off. Choosing the fulfilment model per outlet, in advance, avoids that.

Fulfilment models available to a Malaysian outlet running its own ordering channel, showing who delivers, the typical cost per order to the outlet, the setup effort, and the situation each model suits best.

Model	Who delivers	Cost per order	Setup effort	Suits
Pickup only	Customer	RM0	Low	Office lunch crowds, cafés
Own rider, fixed zones	Your staff	RM5–9	Medium	Dense residential catchments
On-demand courier booking	Third-party rider	RM8–18	Medium	Irregular or long-distance orders
Scheduled batch delivery	Your staff, one run	RM3–6	Medium	Meal plans, catering, tingkat
Aggregator self-delivery	You, listed there	Commission	Low	Keeping discovery while cutting fees

Source: ZenWeb operational data, fulfilment setups implemented and reviewed with Malaysian F&B clients, 2024–2026. Courier costs vary by distance, time of day and city. Licence.

Two rules save a lot of grief. Set the minimum order above your delivery cost, so a RM12 order never triggers a RM14 ride. And draw zones by postcode, not radius: a five-kilometre circle in Kuala Lumpur can contain a river, a highway and forty minutes of traffic. Point your Google Business Profile at the ordering page too; the local play is in restaurant marketing in Kuala Lumpur and F&B marketing across Klang Valley.

Key takeaway: Launch pickup first and add delivery once the order volume covers a rider. Zones by postcode, a fee per zone and a minimum order above your delivery cost keep the feature from being switched off in month two.

5. Payments: FPX, E-Wallets and Cards on Your Own Checkout

Quick Answer: Offer FPX or DuitNow, at least one major e-wallet, and cards. Malaysian food orders are small, so a flat-fee bank transfer often costs less than a percentage card fee, while e-wallets win on speed. Offering only cards is the single most common reason a direct checkout underperforms.

Payment choice is not cosmetic. A RM32 dinner order paid by card carries a percentage fee; the same order over FPX carries a fixed charge in sen. Across a few thousand orders a year that gap is real money, and it decides who finishes checkout at all.

Payment methods used on Malaysian direct food ordering checkouts, showing the share of completed orders by method, whether the fee is a flat charge or a percentage, the usual settlement timing, and the main operational drawback of each.

Method	Share of orders	Fee shape	Settlement	Watch out for
E-wallet	41%	Percentage	1–3 working days	App switching drops some users
FPX online banking	28%	Flat sen per txn	1–2 working days	Bank downtime at night
Debit or credit card	19%	Percentage	2–3 working days	3-D Secure drops on weak signal
DuitNow QR	8%	Flat or low	Same or next day	Manual matching if not integrated
Cash on pickup	4%	None	Immediate	No-shows on prepared food

Source: ZenWeb operational data, completed direct food orders across Malaysian F&B client checkouts, 2024–2026. Shares are of paid orders and vary by outlet type and average basket size. Licence.

The plumbing behind those rows is its own project, covered in payment gateway integration in Malaysia, with the provider comparison in choosing a payment gateway and the basics in what a payment gateway is. Speed matters too: a slow checkout page loses hungry customers faster than a missing wallet, with wider fixes in checkout optimisation. If invoices come out of your online ordering system, LHDN's e-invoicing rollout reached taxpayers turning over up to RM5 million on 1 January 2026, with businesses under RM3 million exempted, so store the required fields from the start.

Key takeaway: Offer e-wallets, FPX and cards together, and check the fee shape against your average basket. On small food orders a flat-fee bank transfer usually beats a percentage card fee, and the difference compounds across a year.

6. Getting Orders Into the Kitchen Without Chaos

Quick Answer: An order that arrives as an email nobody opens is a lost order. Route direct orders to the same place aggregator orders land: a printed ticket or a screen the kitchen already watches, with an audible alert and an acceptance step that starts the prep clock.

Most articles on direct ordering stop at the customer's screen. The kitchen is where these projects fail, usually on the first busy Friday, and the fix is operational rather than technical.

How to route direct orders into a working kitchen

1. Pick one landing point. A thermal printer at the pass, or one tablet on the wall. Two places means a missed order the day someone is on leave.
2. Make it audible. A silent notification is invisible during service. A printer that chirps or a tablet that chimes is what staff respond to.
3. Add an accept step. Someone taps accept and the customer sees a prep time. That tap is also your record that the order was seen.
4. Print the modifiers on the ticket. Less spicy, no onion, extra rice. If the variant never reaches the ticket, the customer complains about a system the kitchen cannot see.
5. Set a pre-order cut-off. Scheduled orders need a lock time, or a nasi lemak order for 8am arrives at 7:52am.

Stock is the neighbouring problem. If an item sells out at the counter and the site keeps selling it, refunds follow, so tie availability to whatever your team already updates. Multi-outlet operators formalise this with a custom inventory system, priced alongside the connection work in system integration costs in Malaysia.

Key takeaway: Decide the kitchen landing point before launch day. One audible destination, an accept step and modifiers printed on the ticket are what separate a used system from an abandoned one.

7. The Customer Data You Finally Own

Quick Answer: Aggregators keep the customer relationship. A direct channel gives you the phone number, the order history and the address, which turns a quiet Tuesday into something you can act on instead of something you watch happen.

Commission savings get the attention, but the data compounds. Order history shows who ordered three times and stopped, which item pulls the biggest basket, and which postcode deserves a delivery zone next. Without it you promote to nobody in particular.

• Bring back lapsed regulars. A short message to customers who last ordered six weeks ago is the highest-return campaign most outlets have, and it needs a list you own.
• Lift the basket deliberately. Order history shows which add-on pairs with which main, so prompts stop being guesswork. The metric behind it is average order value.
• Reward frequency, not discounts. A tenth-order reward costs less than permanent price cuts, as covered in loyalty marketing.
• Message where people read. In Malaysia that is chat, so plug the list into WhatsApp marketing rather than email alone.

Owning data brings duties. Collect only what an order needs, say what you use it for, and keep a deletion path; our PDPA compliance checklist covers the obligations. If corporate customers need a login to reorder and see past invoices, that is customer portal development, not a checkout. Measurement matters too, since broken e-commerce tracking in GA4 hides the revenue this was built to prove.

Key takeaway: The customer list outlasts the commission saving. Order history lets you bring lapsed regulars back on purpose, and that is something no aggregator listing will ever hand over.

8. Running Hybrid: How Fast Orders Actually Shift Across

Quick Answer: Direct orders do not arrive on launch day. With packaging inserts, a QR at the counter and staff mentioning it, outlets typically move a fifth of orders direct by month three and around a third by month six. Doing nothing after launch moves almost nobody.

This is the number owners most want and least often see, because it decides whether the build pays back in one year or five. Shifting orders is a marketing job that runs for months after the developer finishes.

Share of all orders placed through an outlet's own ordering channel rather than an aggregator, at each month after launch, compared across three promotion approaches: link on the website only, QR code and packaging inserts, and a full push adding staff prompts and a first-order incentive.

Month after launch	Website link only	QR code and inserts	Full push with staff prompts
Month 1	2%	6%	11%
Month 2	3%	11%	17%
Month 3	4%	15%	22%
Month 4	5%	19%	27%
Month 5	6%	22%	31%
Month 6	7%	25%	34%

Source: ZenWeb operational data, direct-channel launches tracked with Malaysian F&B and retail clients, 2024–2026. Share is of total orders across all channels for the same outlet. Licence.

The gap between the first column and the last is not technology. It is whether anyone told the customer. Print the QR on the packaging, mention it at the counter, and give the first direct order a small reason to try. Keep the aggregator listing running throughout, because discovery still has to come from somewhere. DOSM reports that 72.7% of Malaysian establishments had a web presence in 2023, up from 71.4% the year before. What most of them lack is an online ordering system to take the order once someone arrives.

Key takeaway: Budget for the shift, not just the build. Packaging inserts, a counter QR and staff prompts are what turn a single-digit direct share into roughly a third within six months.

9. What It Costs, and How to Start Small

Quick Answer: Build an online ordering system malaysia customers will actually use in stages. Menu and pickup checkout first, then delivery zones, then loyalty. Most single-outlet builds land in the same band as a small e-commerce site, and the commission saved on repeat orders is what funds the next stage.

Start narrower than feels satisfying. One outlet, the twenty items that sell, pickup only, one payment method your customers already use. That version is live in weeks, and it produces the number every later decision depends on: how many people order direct when asked.

Budget bands sit in our e-commerce website cost guide for Malaysia, and the sequence from discovery to launch in our web development process. If your flow needs logic no plugin covers, such as subscription meal plans, that becomes custom web application development. Most food orders arrive on a phone, so mobile conversion and page speed matter more here than on a brochure site, with benchmarks in e-commerce conversion rate and the long game in customer retention. If social is where orders start today, see selling on TikTok Shop, Shopee and Instagram and e-commerce marketing tools. Our web development services in Malaysia cover the build, the payment rails and the kitchen routing, and we will say plainly when your volume does not justify an online ordering system yet.

10. Frequently Asked Questions

1. What is an online ordering system?

It is your own menu, cart and checkout on your own domain, so customers order directly instead of through a delivery app. It handles items with variants and add-ons, pickup or delivery, payment, and sending the order to your kitchen. With no third party in between, the only deduction is the payment fee.

2. How much can a Malaysian restaurant save by taking orders directly?

On an aggregator, a RM100 order commonly settles at RM65 to RM78 once commission, promo co-funding and platform charges are taken. On your own checkout the same order settles at roughly RM97 to RM99, depending on the payment method. The saving repeats for as long as the customer keeps ordering direct.

3. Should I remove my outlet from Grab and Foodpanda?

Usually no. Aggregators remain the cheapest way for someone who has never heard of you to find you. The better approach is hybrid: keep the listing for new customers, then use packaging inserts, a counter QR and staff prompts to move repeat customers across.

4. Do customers actually order direct once the system is live?

Only if you tell them. A link sitting quietly on a website moves around 7% of orders by month six. A QR code plus packaging inserts moves roughly a quarter, and adding staff prompts and a first-order incentive moves about a third. The build is the easy half.

5. Which payment methods should a Malaysian ordering system accept?

E-wallets, FPX online banking and cards, at minimum. E-wallets take the largest share of direct food orders and are fastest on a phone, while FPX carries a flat charge that beats a percentage fee on small baskets. Cards alone is the most common reason a checkout underperforms.

6. How is this different from an online booking system?

A booking system sells time: slots, staff availability and appointment reminders. An ordering system sells items: menus with variants and add-ons, fulfilment choices, and a kitchen ticket at the end. Some businesses need both, but they are separate builds and should be scoped separately.