Direct Ordering vs Delivery Apps: Which Orders Make Money?

Direct ordering can leave a restaurant with more money from a repeat customer. It can also cost more than a delivery marketplace once you pay to acquire that customer, take the payment and get the food to their door.

Compare channels using contribution per fulfilled order, then ask whether the order was genuinely additional and whether the kitchen had capacity. Use the actual Just Eat or Deliveroo agreement and settlement statement. A headline commission rate does not describe every service, promotion, refund or operating cost.

By David Black, founder of PureRapid. Editorial analysis; sources checked 21 October 2026.

Start with the order, not the settlement deposit

A platform settlement is a cash transfer after certain deductions. It is not the order’s profit. Ingredients, packaging, labour and other costs may sit elsewhere in your records. The same distinction applies to money arriving from a direct ordering provider.

For each channel, reconcile completed orders, sales, discounts, refunds, fees and payments over the same period. Separate delivery from collection. Keep customer charges that the restaurant does not receive out of restaurant revenue.

Use a consistent VAT basis. For a VAT-registered business that can recover relevant input VAT, a net-of-recoverable-VAT comparison differs from the cash movement on a fee invoice. If VAT is not recoverable, it remains a cost. Have your finance lead confirm the treatment rather than treating every deduction as equivalent. HMRC explains the general VAT recovery rules and exceptions, including restrictions under the Flat Rate Scheme.

Deliveroo’s published partner terms describe service fees calculated on VAT-inclusive GMV, with VAT added to the fee, and settlement adjustments. This illustrates why the percentage and its calculation base both matter. Check your own current agreement; this article does not quote a standard fee for either marketplace.

How to compare Just Eat, Deliveroo and your own site

There is no universal platform winner in this calculation. A marketplace supplying the courier is a different service from one taking orders for your own drivers. Compare the actual package at each location, then compare customers of the same type and orders you can fulfil to the same standard.

Keep a column for the evidence behind each number: agreement, settlement line, supplier invoice or documented allocation. If a charge is unclear, resolve it before projecting a channel-wide saving. A lower fee matters only after the comparison covers the same work.

Count what replaces the marketplace fee

A marketplace may supply discovery, ordering infrastructure, payment handling and delivery, depending on the service purchased. Direct ordering gives you another route to the customer, but those functions still have to be supplied.

• Food and fulfilment: ingredients, packaging, wastage attributable to the order and any genuinely incremental labour.
• Channel costs: marketplace fees or direct ordering charges, payment processing, courier costs and support.
• Demand costs: funded discounts, advertising, allocated creative or management costs and customer communications.
• Failure costs: refunds, remakes, rejected orders and delivery problems, allocated on a consistent basis.

Count each item once. If a contracted fee already covers payment handling or delivery, do not add an invented second charge. Equally, “zero commission” does not mean zero acquisition or fulfilment cost.

A direct order can be better on the second purchase

Illustrative example, not a PureRapid client result. Consider three hypothetical delivered orders at £25 net restaurant sales each. All figures are GBP per completed order, excluding recoverable VAT. They are modelling assumptions, not Just Eat, Deliveroo or courier prices.

Each order uses £8.75 of ingredients, £1.25 of packaging and £2.50 of incremental preparation labour, leaving £12.50 before channel and demand costs. Assume no refund or remake on these particular orders; a real forecast needs an allowance based on actual experience.

The direct new-customer order uses £5 of complete acquisition cost instead of the repeat order’s £0.50 communication cost. In this example, it leaves £1 less than the marketplace order. The direct repeat order leaves £3.50 more. Those are different decisions hidden inside one channel label.

The £5 acquisition figure must be allocated across genuinely acquired customers, not every order reported by an advertising platform. If it rises to £8, first-order contribution falls to −£0.50. A hoped-for repeat purchase does not pay that bill today.

Will switching the order improve the business?

Suppose a direct ordering system also costs £120 per month, excluded from the per-order figures above. If repeat orders genuinely move from the illustrated marketplace route to the direct route, each saves £3.50 before that subscription. You need 35 such orders to exceed £120: 35 × £3.50 = £122.50.

That calculation only covers the subscription. It excludes setup, extra support and additional fixed staffing. Nor does it prove those customers will switch. A discount used to encourage the move reduces the £3.50 benefit.

An additional order is different again. A £3.50 marketplace contribution can be worthwhile when there is spare capacity and the customer would otherwise have bought elsewhere. If it displaces a £7 direct order at peak time, the opportunity cost changes the decision.

Measure the total contribution across channels, not just direct revenue. A rising direct share can reflect useful retention, expensive discounting or simple movement of existing demand. The channel mix alone cannot tell you which happened.

Price the Friday-night constraint

A restaurant with spare kitchen capacity and one at its operating limit should not use the same rule. At the limit, preparation minutes, courier availability and accuracy become scarce resources.

Compare contribution against the actual bottleneck. A complicated order that leaves more money may consume enough preparation time to block several simpler orders. If service quality falls, refunds and lost repeat custom can reduce the apparent gain further.

Review busy and quiet periods separately. Adjust menus, preparation estimates, promotions and channel availability within your operational capabilities and partner terms. Keep the customer promise realistic; an unmanageable delivery time is not a marketing success.

Use contribution per scarce minute when the kitchen is full

Illustrative example, not a PureRapid client result. At the contributions above, a £7 direct repeat order using ten minutes of the constrained preparation station leaves £0.70 per minute. A £3.50 marketplace order using four minutes leaves £0.875 per minute. The larger contribution per order is not automatically the better use of that station.

This only helps if those minutes really limit output and other demand is available. Batch preparation, delivery slots, menu dependencies and customer experience may change the choice. When the kitchen has spare capacity, a positive additional contribution can be useful even at the lower rate.

Run a controlled direct-ordering test

1. Build a channel ledger. Take one representative period and reconcile completed orders, net sales, refunds and all order-related costs.
2. Separate new and repeat demand. Allocate acquisition and retention costs deliberately, with no double counting.
3. Test a small change. Improve the direct journey or promote it through your own authorised channels. Compare total contribution and service quality against a comparable period.
4. Track what happens next. Follow repeat orders, support workload, delivery reliability and the time taken to recover acquisition spend.

Use only customer data and communication routes you are entitled to use. A marketplace order does not by itself give permission to add that customer to your marketing list or ignore partner terms.

Keep useful marketplace demand while improving the direct experience for customers who choose it. A reliable ordering journey, accurate menu and clear collection or delivery promise can be worth more than another temporary discount.

Choose the mix that earns its place

Do not make “leave the apps” the objective. Set a contribution requirement, protect service capacity and test whether direct retention improves the total result. Review by location as well as channel; catchments, demand and fulfilment costs differ.

For several sites, use the same ledger alongside the decision about which location gets the next marketing investment. For a new opening, plan how first-time demand becomes repeat visits before assuming a mature direct-ordering mix.