Key takeaways
- Much of the recent growth in restaurant delivery comes from new sites and newly added delivery, not more orders per restaurant, so the apps are getting more crowded.
- A promotion is only worth running if the extra orders it creates earn more than the discount you give to customers who would have ordered anyway.
- Measure promotions against a baseline period and track new versus returning customers, not just total orders.
- Set a stop rule before you start: a cost per new customer you will not exceed.
Why everyone seems to be running an offer
Open any delivery app on a Friday and a large share of listings carry an offer: a percentage off, free delivery, a free side or a buy-one-get-one-free. The reason is visible in the sales data. In September 2026, restaurants' at-home revenue was 11.8% higher than a year earlier once new sites and newly added delivery were included, but like-for-like growth was only 2.3%[1]. In August, like-for-like at-home sales rose just 0.2%[2].
Source: NIQ (CGA by NIQ), 2026[1]
Show the data
| Item | Value |
|---|---|
| Total, including new sites and newly added delivery | 11.8% |
| Like-for-like | 2.3% |
In plain terms, more restaurants are on the apps, and each one is competing harder for a market that is growing slowly per site. Promotions are one of the few levers that move a listing up the screen. That does not make them profitable.
The types of promotion and what they really cost
| Promotion | Who pays | Hidden cost |
|---|---|---|
| Percentage off the order | Usually you, sometimes shared with the platform | Commission may still be charged on the pre-discount total; check your terms |
| Free delivery | You or the platform, depending on the scheme | Encourages small orders that may not cover packaging and handling |
| Free item over a spend threshold | You | Cost of the free item, but lifts average order value |
| Buy one get one free on a dish | You | Halves revenue on that dish; best on low food-cost items |
| Sponsored or promoted listing | You, per click, impression or order | Pays for attention whether or not orders follow |
| Platform-funded offers | The platform | Often tied to price or commitment conditions in your agreement |
Every one of these should be judged against the same question: how many orders did it bring that would not have happened anyway, and what did each of those orders earn after every cost?
A worked Example
Example: a fictional burger takeaway normally receives 300 app orders a week with an average order of £22. It runs 20% off all orders for two weeks and receives 390 orders a week. The owner's assumed figures are food and packaging at 32% of net sales and contribution of about £8 per full-price order after commission.
- Example: extra orders created are 90 a week (390 minus the 300 baseline).
- Example: the discount is about £4.40 per order, given on all 390 orders, including the 300 that would have happened anyway.
- Example: cost of discounting the baseline orders is 300 x £4.40, or about £1,320 a week.
- Example: contribution from the 90 extra orders at the discounted price, roughly £8 minus £4.40 plus any commission saving, is about £3.60 each, or about £324 a week.
- Example: net effect is roughly £1,000 a week worse off, unless enough of the new customers return at full price.
This Example is deliberately simple, and it shows why blanket discounts are rarely worth it on their own. The promotion only pays if it brings genuinely new customers and a meaningful share of them reorder later without a discount.
Questions to ask before opting in
Platform account managers often suggest promotions, particularly before busy periods or when a new competitor appears nearby. That advice can be useful, but the account manager's targets are not your margin. Get answers to a few questions in writing before you agree.
- Who funds the discount, and is any part of it shared or reimbursed by the platform?
- Is commission charged on the order value before or after the discount?
- Does opting in change your ranking, and for how long after the promotion ends?
- Can the offer be limited to new customers, certain days or certain dishes?
- Are there conditions attached, such as price parity with other channels or a minimum term?
- How will the platform report results, and can you see new and returning customers separately?
- Can you stop the promotion early, and how quickly does it switch off?
If the answers are vague, run a smaller test first: one dish, one quiet day, or new customers only. A small test that fails costs little; a blanket discount that fails can cost a month of profit.
Measure incremental orders, not total orders
Set a baseline
Record orders, average order value, new versus returning customers and contribution for the four weeks before the promotion, by day of week.
Run the promotion for a fixed period
Two to four weeks is enough to see a pattern. Avoid weeks with bank holidays, big matches or school holidays, which distort the baseline.
Track new customers
Most platforms report new and returning customers. The new-customer count is the number that justifies a promotion.
Measure the hangover
Check orders and new customers for four weeks after the promotion ends. A promotion that only pulls forward orders shows a dip afterwards.
Work out cost per new retained customer
Divide the net cost of the promotion by the number of new customers who reorder at full price within, say, eight weeks.
Decide your maximum acceptable cost per new retained customer before you start. If you would not pay that much to win a regular any other way, the promotion should stop.
Promotions that tend to work better
- Targeted at new customers only, where the platform allows it, so regulars are not discounted.
- A free side or dessert over a spend threshold, which lifts order value and costs less than a percentage off.
- Limited to quiet days or slots when the kitchen has spare capacity and the extra orders cost little to make.
- On high-margin dishes with low food cost, such as sides, drinks and some starters.
- Time-limited and tested, with a clear end date, rather than permanent.
Watch food costs as you plan. In Q3 2026, 84% of hospitality leaders were concerned about food and drink inflation[3]. A promotion priced on last quarter's ingredient costs can become loss-making without anyone noticing.
Sponsored listings and ads
Paying for a higher position works differently from a discount: you pay for visibility, not for the order. The same discipline applies. Track impressions, menu views, orders and new customers attributed to the ad, and calculate the cost per new customer. If the platform reports return on ad spend, check what it counts as an attributed order, and compare it with your own order data.
Strong ratings and good photos often lift conversion from views to orders more cheaply than buying extra views. Fix the listing before paying to send more people to it. Our guide to improving your ratings on delivery apps covers what to change first.
Spend some of the budget on your own channels
Money spent discounting on an app builds the app's customer relationship as much as yours. Customers who find you through a promotion are still the platform's customers until they order from you directly. It is worth setting aside part of any promotion budget for offers that bring people to your own phone line or website.
- A bag card in every app order with your direct number and a small direct-order perk, where your agreement allows it.
- A first-order offer on your own website, promoted on your social channels and printed menus.
- Rewards for regulars who order direct, such as a free side every few orders, recorded on the till.
- Local leaflet drops to postcodes where your app orders already come from.
Keep a simple promotion log with the dates, the offer, the cost, orders and new customers for every promotion you run, on any channel. After a few months you will know which offers work for your kitchen and which only move orders from one week to the next.
When to stop
- New customers are not rising compared with the baseline.
- Orders dip below baseline after the promotion ends.
- Cost per new retained customer is above the limit you set.
- The kitchen is overwhelmed and ratings start to fall.
- Most orders under the offer come from existing regulars.
Many operators have also been raising prices. In September 2026, 12% of accommodation and food businesses raised prices compared with August, while 72.9% kept them the same[4]. If you raise menu prices and run deeper discounts at the same time, customers notice; it is usually better to set a fair menu price and use small, targeted offers.
For the wider channel question, see phone orders vs app orders, and for running several platforms from one screen, too many delivery tablets.
Frequently asked questions
Are Deliveroo and Just Eat promotions worth it?
Only if they bring new customers who later reorder at full price. Measure orders and new customers against a baseline, track the weeks after the offer ends, and calculate the cost per new retained customer.
Who pays for delivery app discounts?
It depends on the promotion and your agreement. Many discounts are funded by the restaurant, some are shared, and some are platform-funded with conditions attached. Read the promotion terms before opting in.
How long should I run a delivery app promotion?
Long enough to see a pattern, usually two to four weeks, with a fixed end date. Then compare the four weeks after with the baseline before deciding whether to repeat it.
References
NIQ (CGA by NIQ) (2026). Restaurants' delivery sales strengthen in September after tough summer (Hospitality at Home Tracker) (opens in a new tab) ↑
NIQ (CGA by NIQ) (2026). Delivery rollouts boost restaurants' at-home sales but like-for-like growth slows (Hospitality at Home Tracker) (opens in a new tab) ↑
NIQ (CGA by NIQ) and Zonal (2026). Hospitality's confidence rises after solid Q3, but Budget support crucial for stability (Business Confidence Survey with Zonal) (opens in a new tab) ↑
Office for National Statistics (2026). Business insights and impact on the UK economy, wave 166 dataset (opens in a new tab) ↑