By SuperApp Team — who we are
Marketplace Business Models, Explained
How a marketplace actually makes money — commission, delivery fees, subscriptions and placement — what each costs to run, and where each one breaks.
A marketplace business model is the arrangement by which an operator who owns neither the supply nor the delivery capacity gets paid for connecting them — and the choice between the available arrangements decides more about whether the marketplace works than the software underneath it does.
There are only five ways a local commerce marketplace takes money, and most operators run three of them at once without having decided to. Setting them out separately is the first useful step, because they fail in different places.
The five, and who actually pays
Commission on the order value. The merchant gives up a percentage of every order. Scales automatically with basket size, requires no negotiation per merchant, and is the model merchants resent most — because the amount grows while the service does not.
A delivery fee from the customer. The customer pays for the drop. Honest, visible, and the most price-sensitive line in the whole model: it is the number that makes a customer close the app, and it is why so many operators end up subsidising it.
A merchant subscription. A fixed monthly fee per location regardless of volume. Predictable both ways, easy to forecast, and the model that most obviously breaks the alignment between your revenue and the merchant’s — you get paid the same whether they had a good month or a bad one.
Customer-side fees. Service fees, small-basket fees, priority fees. Individually small, collectively material, and the fastest way to lose the trust you spent your acquisition budget buying if they arrive late in the checkout.
Placement and advertising. Merchants pay for position. The highest-margin line in the model and the last one that becomes available, because it needs enough demand that position is worth buying. Operators who plan for this in year one are usually planning for revenue they cannot yet sell.
Most working marketplaces are commission plus a delivery fee, drifting toward subscriptions for large merchants and placement once density arrives.
Why take rate is not the number that matters
Take rate — your revenue as a share of order value — is the figure operators quote and investors ask for, and on its own it says very little.
What matters is contribution per order after the drop. A 30% take rate on a £12 order that costs £4.50 to deliver is worse than a 15% take rate on a £45 basket delivered for the same £4.50. The take rate looks twice as good and earns half as much.
Three consequences follow, and they are the actual strategy:
Basket size is a lever on the same footing as take rate, and it is usually easier to move. Minimum order values, bundles, and merchant mix change it directly.
Delivery cost per drop falls with density, not with volume. A thousand orders spread over a city is a worse business than four hundred in three postcodes. This is why the city-by-city sequence exists rather than a national launch — we set out the operational version of that in launching a regional aggregator city by city.
Batching is where the margin actually is. Two drops on one trip changes the arithmetic more than a percentage point of commission ever will, and it depends on density rather than on pricing.
What each model breaks on
Commission breaks on merchant churn. The best merchants — highest volume, lowest support cost — are the ones for whom commission is most expensive, so they are the first to build their own ordering channel. Losing them is the standard failure and the reason a subscription tier for large merchants exists.
Delivery fees break on the competitor who subsidises. If a funded rival sets the fee to zero in your city, your honest fee reads as expensive. Being the cheapest is not a defensible position; being reliable in a defined area sometimes is.
Subscriptions break on the small merchant. A fixed fee is a large number to a site doing five orders a day, and those merchants leave first — which thins your catalogue at exactly the point where catalogue breadth is what makes customers open the app.
Placement breaks on trust. Sell too much of it and the ranking stops reflecting quality, customers notice, and the asset you were selling stops being worth buying.
What to work out before choosing
Five questions that decide the model more reliably than a competitive survey:
- What is your average basket, and what does one drop cost you? Everything else is downstream of those two numbers. If you cannot state both, the model is not yet a decision.
- Which merchants would you lose first, and can you afford to? Run it specifically: name the top five and imagine each one leaving.
- Are you charging the side that has the alternative? Merchants can build their own ordering; customers can walk to the shop. The side with the weaker alternative can bear more.
- What happens to this model at a tenth of your target density, and at ten times? Models that only work at the target are not plans.
- Which of the five are you running by accident? Most operators are running more than they think, and a fee nobody decided to add is a fee nobody is defending.
When a marketplace is the wrong structure
Marketplaces are a demand-aggregation business, and the model only makes sense if aggregating demand is genuinely hard for the merchants themselves.
If you have a handful of merchants who each already have their own customers, you are not aggregating demand — you are running a shared delivery operation, and a per-delivery fee describes that honestly while a commission does not.
If your category has very low purchase frequency, the customer account never compounds, and the acquisition cost has to be repaid on one or two orders rather than amortised across a year. That is a much harder business than the same model in food.
And if a single merchant would account for most of your volume, you are building their ordering channel with extra steps. That is a white-label engagement, not a marketplace, and it is better priced as one — the distinction is set out in what a white-label super app is.
For the platform-selection question rather than the business-model one — build versus buy, single versus multi-vertical, and how vendors themselves charge — see how to choose a marketplace platform in 2026.