Delivery refunds: what 4 months of real data reveal
Between 1 January and 30 April 2026, KonkiCam tracked more than 11,000 delivery orders in a single restaurant, representing €319,390 in revenue.
The aim: to understand what really lies behind customer refunds and measure the value of photo evidence at the moment of preparation.
The results show two distinct phenomena: genuine preparation errors, but also a majority of complaints that don’t match what the available evidence shows.
74% of the disputes analysed don’t match a visible error by the restaurant
Of the 209 disputes recorded, the analysis produces two categories:
- 26% match an identified preparation error
- 74% instead show a correct order in the available evidence
It would be going too far to label all of these 74% as fraud: a photo proves the state of the order at the restaurant, not what happened afterwards.
It does, however, establish one essential point: the order was correct when it was sent.
That’s why we prefer to speak of disputable complaints rather than proven fraud.
93% of disputes succeeded
Where the evidence could challenge the complaint, 155 cases were disputed.
144 were successful – a 93% success rate.
These cases represent €1,266.74 recovered over the period.
So the point of the photo isn’t to dispute every refund. Instead, it lets you quickly sort between:
- a genuine error, which there’s no point disputing
- an apparently correct order, for which a dispute can be documented
Preparation errors fall sharply over four months
The monthly breakdown reveals something else.
The preparation error rate thus falls from 0.85% in January to 0.33% in April – a drop of around 61%.
Meanwhile, the number of errors identified falls from 24 to 8 a month.
Internal errors fall much more sharply than disputable complaints.
This data shows that traceability isn’t just useful for disputes: it also helps identify genuine errors and tackle their causes.
Refund amounts fall too
Recorded refunds fall from €670 in January to €291 in April.
That’s a reduction of 57%, while monthly volume remains at several thousand orders.
Over the same period, the dispute success rate rises from 89% in January to 97% in April.
Evidence quality remains decisive
Over the period, 92% of disputes had a photo and 92% of the photos taken were usable.
Conversely, missing evidence accounted for €134.85 of avoidable losses.
So a photo is only valuable if it quickly answers three questions:
- 1Which order is it?
- 2When was it prepared?
- 3Is the item in question visible?
It’s this link between the order and its evidence that turns a simple photo into a traceability tool.
What we take away from these four months
This analysis highlights three lessons.
- 1Not every refund is abusive. About a quarter of the disputes analysed correspond to genuine errors by the restaurant
- 2Most, however, are worth checking. In 74% of cases, the available evidence did not support the error attributed to the restaurant
- 3Traceability is as much about improving operations as recovering refunds. Between January and April, the observed error rate was more than halved
KonkiCam thus quickly answers a question that has until now been hard to settle:
“Was this order correct when it left the restaurant?”