Australia lost two delivery platforms in three years. Deliveroo pulled out of the country in 2022, and Menulog followed in November 2025. What is left is a two-platform market, and neither survivor has much reason to compete on the number that matters to an operator.
The commission most Australian venues quote is somewhere between 15% and 30%, depending on tier. That is the headline. Once the service fee, payment processing, promotional co-funding and the cost of packaging a delivery order are counted, the effective take on delivery revenue lands north of 30%. This piece walks the arithmetic on a single order, then the part of the cost that never appears on an invoice.
The arithmetic on one order
Take a $40 delivery order at a 30% effective rate. The platform takes $12. You keep $28, and out of that $28 comes food cost, packaging, labour and rent. On a venue running a 65% gross margin, that $40 order leaves roughly $26 of gross profit before the platform is paid, and roughly $14 after. The same $40 order placed through your own channel leaves the whole $26.
Scale that. A venue doing $30,000 a month in delivery gives up around $9,000 of it. Over a year that is $108,000 — for most independent operators, a staff member, or the difference between opening a second site and not.
None of that arithmetic is an argument for switching delivery off. Delivery reaches people who were never going to walk in, and that reach is worth paying for. The argument is narrower and harder to dismiss: you are paying the same rate on the guest who already knows you as on the one who has never heard of you.
The cost that is not on the invoice
When an order arrives through a platform, the platform holds the relationship. You get an order ticket. You do not get a name you can contact, a visit history, or any way to tell a first-timer from somebody on their twentieth order. The next time that guest is hungry, the app they open is not yours.
That is the compounding part. Commission is a cost you can measure monthly. Not owning the guest is a cost that shows up as never being able to bring anyone back deliberately — no second-visit nudge, no knowing which regulars have gone quiet, no way to make the offer that would have worked.
Why Australia specifically
Two things make the direct channel unusually available here. Roughly 95% of in-person card payments are contactless, and 99.5% of transactions under a hundred dollars never involve cash. Paying by phone is not a behaviour you have to teach an Australian guest — it is already how they pay.
And the field is large and mostly independent: around 27,600 cafés trade across the country. Very few of them have an app of their own, which is exactly why the ones that do stand out rather than blend in.
What actually changes
A direct channel is not a replacement for delivery. It is a place for the orders that did not need a platform in the first place — the regular who orders the same thing every Tuesday, the office that orders for eight on Fridays, the guest who lives around the corner. Those orders carry no per-order commission and they tell you who placed them.
The practical test is not "can we leave the aggregators". It is "what share of last month's platform orders came from someone who has ordered before". If that number is meaningful, you are renting access to your own regulars.