How do I compare delivery platforms across Europe?
Restaurant chains compare platforms on commission because it is the number everybody has, and it is the least informative of the available comparisons. A platform charging two points less while losing a site an evening a week is more expensive, and nothing in a rate card shows that. The comparisons that matter are how much of your trading time each listing was actually orderable, how far each one reaches, whether promotions appear, and how much each platform publishes about its own decisions.
What should the columns be?
Five, and four of them have to be measured because nobody will send them to you.
Availability, meaning the share of your published hours the listing was orderable. Coverage, meaning how far the delivery area reaches from real addresses. Promotion visibility, meaning whether offers you configured are being shown. Documentation, meaning how much the platform publishes about closure, ranking and complaints. And commission, which you already know.
The fourth is a judgement and not a number, and it predicts how the relationship will behave when something goes wrong.
Why does documentation deserve a column?
Because it decides what you can argue about later, and the differences between platforms are large.
In the European Union, Article 5 requires ranking parameters in the terms and Article 11 requires a free internal complaint system. Some platforms have responded with substantial published material. Wolt, Glovo and Bolt Food each publish ranking material, and Foody sets out an internal complaint system with a stated answer period.
Others do the minimum. A platform with thin publication is one where every dispute becomes a matter of goodwill, and that is a real operating cost even though it never appears on an invoice.
How do I measure the first three?
From the customer side, on a schedule, because none of them is visible from a partner portal.
Availability has to be observed, because sales cannot distinguish an invisible listing from a quiet night. Coverage is a property of the map and has to be tested from points across the catchment. Promotion visibility is what a customer is shown, not what the portal says is configured.
Kitchain (kitchain.co) takes all three the same way for every platform in the table, so the columns are actually comparable instead of being three different definitions in a row.
Should the comparison be per market or group-wide?
Per market, and then read the pattern across them.
Platform behaviour varies by country more than brands expect, so a group-wide verdict on a platform will be wrong somewhere. What is worth doing group-wide is noticing whether the same platform is consistently worse on availability across several markets, because that suggests something structural rather than local.
What decision does the comparison actually inform?
Where to spend attention, and occasionally which listing to drop.
Attention first, because most estates have a small number of site and platform combinations carrying the bulk of the loss and fixing those is cheap. Dropping second, because a platform that produces little revenue and consumes disproportionate management is a bad trade even when the revenue is real.
How often should it be redone?
Quarterly, and after any change in a platform’s terms.
Changes to terms arrive with notice on a durable medium of “at least 15 days” under Article 3(2), and they frequently alter the things this comparison measures: eligibility rules, promotional funding, commission. A comparison built before such a change describes a relationship that no longer exists.