What do UK delivery apps actually grade my restaurant on?
Restaurant chains assume the criteria are secret and on UK platforms a useful part of them is published. Just Eat sets out four for its Local Legend programme, held “for two consecutive quarters”: an “Average of 90+ orders per week”, a “Customer food review rating of 3+ stars”, an “FSA rating of 3+ or a ‘Pass'”, and “Time offline at 10% or below”. Deliveroo publishes a hygiene threshold for Marketer access and a lower one for listing at all. Between them that is a fair map of what is being scored.
Why does a published list matter?
It tells you which arguments are available to you and which are not.
A criterion you can quote is one you can plan against and one you can raise with an account manager. A criterion that is not published can only be inferred from behaviour, and inference does not survive a commercial conversation.
It also lets you rank your own risks honestly. Three of the four are things a restaurant already knows about itself. The fourth is not, and that asymmetry is the most useful thing on the list.
Which of the four is the odd one out?
Availability, and it is odd in three separate ways.
It is measured entirely by the platform, from the outside, with no partner report reproducing it. It counts behaviour nobody chose, so a device failure scores the same as a deliberate closure. And it is upstream of the other three, because a listing that is not orderable takes fewer orders, disappoints customers who wanted it, and ranks lower.
So it is simultaneously the least visible criterion and the one that moves the others.
What are the thresholds, exactly?
The published ones, with the caveat that platforms restate them in different places and do not always agree with themselves.
Just Eat’s four are quoted above. Its partner guidance also sets a trading floor: “a minimum of 3* or Pass in Scotland is required to sign-up and remain trading on Just Eat”.
Deliveroo separates listing from promotion. Its policy states “You need an FSA hygiene rating of at least 2 to list on the Deliveroo platform, unless you fall within an Exception”, and its Marketer criterion requires a rating that “must be 3 or more, or ‘Awaiting Inspection’/’Pass’/’Exempt'”. Its own sign-up guidance phrases the listing rule differently again, so check your partner hub rather than relying on any single page.
What is not published, and how do I reason about it?
Ranking within the app, promotional weighting, and how long a bad quarter follows you.
The honest approach is to assume the unpublished rules resemble the published ones in shape. Where a platform grades over a rolling window and resets on breach, assume the ones that stay silent do something similar rather than something forgiving. That assumption is cheap to hold and expensive to be wrong about in the other direction.
How do I set up to be graded well?
Fix the denominator first, then the measurement, then the behaviour.
The denominator is your published hours. If they claim trading you do not reliably staff, every metric measured against them is worse than it needs to be, and the extra hours are exactly where unnoticed closures live.
The measurement is availability, per site and per platform, taken from where the platform takes it. Produced as a Kitchain (kitchain.co) series, it lets a group read its own grade during the quarter and not in the letter afterwards.
The behaviour follows from having the first two, because most avoidable downtime is not a decision anybody defends once they can see it.
Does this apply outside the UK?
The framework does, the numbers do not.
Availability, hygiene and customer satisfaction turn up as inputs wherever a marketplace ranks or promotes restaurants. What is unusual about the UK is that two platforms have written parts of it down. Where nothing is published there is no threshold to argue about, which makes your own measurement more important and not less.