How to calculate your delivery refund rate

“We get a lot of refunds.”

That statement doesn’t tell you whether the restaurant actually has a problem.

A restaurant handling 3,000 orders a month will naturally get more complaints than one handling 300.

To analyse the situation properly, you need ratios.

Metric 1: refunded order rate

The simplest formula is:

Number of orders that led to a refund / Total number of orders × 100

Example:

30 refunded orders out of 1,500.

30 / 1,500 × 100 = 2%

The refunded order rate is therefore 2%.

Metric 2: financial refund rate

The number of cases isn’t enough.

Two restaurants can have the same complaint rate but very different amounts.

So use:

Total refunds / Delivery revenue × 100

Example:

€600 refunded on €40,000 of delivery revenue.

600 / 40,000 × 100 = 1.5%

Metric 3: average refund amount

The formula:

Total amount refunded / Number of refunds

shows the average size of cases.

This metric is especially useful for working out how much time it’s worth spending on handling them.

Metric 4: dispute rate

Once cases have been checked:

Number of refunds disputed / Total number of refunds × 100

A very high rate isn’t necessarily good news.

It may mean you’re disputing almost everything without checking cases properly.

The aim isn’t to maximise this rate.

Metric 5: recovery rate

It can be calculated by number of cases:

Disputes accepted / Disputes processed × 100

or by value:

Amount recovered / Amount disputed × 100

Measuring in euros is usually the most meaningful financially.

Metric 6: share of confirmed errors

Traceability lets you add a much more operational metric:

the number of complaints where the evidence confirms a preparation error.

You can then sort them by reason:

• drinks
• desserts
• sauces
• sides
• wrong bag
• multi-bag
• other

This analysis can reveal very practical areas for improvement.

Segmenting the results

An overall rate can hide major differences.

If possible, analyse refunds:

• by platform
• by restaurant
• by day
• by time slot
• by order type
• by product
• by reason

You might find, for example, that most incidents happen between 8pm and 9.30pm during the production peak.

That finding is far more actionable than an overall monthly rate.

Tracking changes over time

A metric is only really valuable when tracked regularly.

A monthly table might show:

You can then measure the impact of a new procedure or improved checks.