A delivery order is a different product that happens to share a name
Food retailers in Singapore and Malaysia still price and plan delivery as if it were the in-store meal carried a little further. It has its own costs and its own ways of failing, and it should be priced, ranged and judged on its own account.
Walk into almost any café in Singapore at breakfast and you will see two businesses running in one room. One serves people at the counter and at the tables. The other serves riders who wait by the door with insulated bags, collecting orders placed on a phone somewhere else. The two share a kitchen, a menu and a brand. In most companies I know they also share a price list and a set of measures, and that is where the trouble begins.
At Rumbia Kopi, our café brand, delivery grew from almost nothing in 2018 to 21 per cent of sales by 2022. For a while we treated that growth as good news without qualification. Every delivered order looked like revenue we would not otherwise have earned, from customers who would not otherwise have come in. It took a careful allocation of costs to show that the average delivered order was losing us money.
What the commission actually pays for
The visible cost is the commission. Across the platforms we use, it was around 30 per cent of the menu price in 2022, and higher when we paid for promoted placement. A commission at that level is not unreasonable for what a platform provides: the rider, the app, the payment, many of the disputes and a stream of customers we could not have found alone. The mistake is to book it as a marketing cost that lives somewhere in head office, when it is a variable cost of the product, incurred on every order.
Then come the costs that seldom reach a delivery report. Packaging that keeps toast crisp and a drink sealed cost us about S$0.38 an order. Packing takes labour, and in a café it takes labour at the worst moment, during the breakfast and lunch peaks, when the same people are serving a queue. When we timed it, a delivery order took about 40 per cent longer to assemble than the same items on a tray. Refunds for missing or damaged items, usually granted by the platform and charged back to us, added a further cost that our finance team had been recording as a reduction in revenue rather than as a cost of delivery.
Once all of that was allocated, delivery contributed minus 1.8 per cent of its own sales. The in-store business was subsidising it. Nobody had chosen that outcome. It accumulated one sensible-looking decision at a time, because delivery was measured on sales growth and in-store trade was measured on margin, and nobody had been asked to look at both.
Food that survives the journey
The second reason to treat delivery as its own product is that some food does not travel. Our best-selling breakfast set, kaya toast with soft-boiled eggs and coffee, is excellent at a table and disappointing after twenty minutes in a bag. The toast steams itself soft, the eggs keep cooking and the coffee cools. Customers who receive it like that do not blame the rider. They blame us, and our complaint data showed it: 2.9 complaints per thousand delivered orders, several times the rate we saw at the counter.
In January 2023 we cut the delivery menu from 64 items to 31. We kept products that arrive the way they left: buns, pastries, sandwiches, sealed drinks and set meals designed for the journey, including a toast set with the kaya packed separately so that the customer assembles it at home. We removed items that depended on minutes, temperature or a particular texture. Colleagues who saw each removed item as lost revenue resisted the cut. I saw each one as a complaint we would no longer be paid to create.
The food halls needed a different answer. Laman Food Halls house stalls we run ourselves alongside stalls run by tenants. Several of our own stalls were small enough that a run of delivery orders stalled the queue for people standing in front of them. We withdrew our own stalls in 11 halls from delivery altogether, while tenants kept the right to list their stalls as they chose. A food hall's first customer is the person who has walked into it.
Parity is a choice, not a duty
The most contentious part of the change was price. We set delivery prices on average 14 per cent above in-store prices and stated the difference clearly on our listings, so that no customer could be surprised. Many operators keep the two prices the same, believing that customers will punish any difference or that a platform's ranking rewards equal pricing.
I do not think price parity is a duty owed to customers. It is a decision about who pays for delivery. When the prices are equal, the customer sitting in the café pays part of the cost of the customer who stays at home. That may be a sensible strategy for a company that wants to build delivery quickly, but it should be a decision taken with the numbers in view, not an assumption carried forward from the years when delivery was small.
The cost of our choice was real. Delivery sales fell from S$46 million in 2022 to S$37 million in 2023, and café revenue fell 3.4 per cent that year. One platform moved us down its rankings for about three months, a reaction I had underestimated, and we lost orders to competitors who had left their prices alone. Our head of brands had warned of exactly that, and she was right to.
By 2024 delivery contributed 8.6 per cent of its own sales, complaints had fallen to 1.1 per thousand orders, and our own order-ahead pickup service, which carries no commission, accounted for 7 per cent of café transactions. The smaller delivery business is the better one. It is also clearer about who pays for what.
There are limits to the argument. A brand built mainly for delivery, a kitchen with no dining room, has nothing to separate and must make delivery pay on its own terms from the start. A company entering a new market may reasonably accept delivery losses for a time to become known. And platforms are entitled to run their rankings as they see fit; an operator who prices differently has to accept the consequences.
For boards the practical point is narrow. Ask to see delivery contribution after commission, packaging, labour and refunds, separately from in-store margin, for each brand. Ask which menu items generate most of the complaints. Ask whether the delivery price was set deliberately or simply inherited. A management team that can answer those questions is running two products. One that cannot is running one product and an unexamined subsidy.