A weekly review brief is written for an owner deciding where to put attention. Read out verbatim to a team about to work a Saturday, it lands as an accusation with statistics attached — and the reliable result is a defensive floor rather than a different service.
The translation is a real piece of work, and it is almost always skipped. What usually happens instead is "reviews are saying we're slow, so let's pick it up tonight", which is a sentence with no content: nobody knows which part of slow, nobody knows what to do differently, and everybody has been told off.
Ninety seconds, two items
Item one, and it goes first: one positive, read out, with the name in it.
"Table on Thursday wrote in about Maria — said she noticed they were in a hurry and got them out in time for the theatre. Nice one."
Not decoration and not morale management. Three reasons it is first.
It makes the whole exercise about what the team did, so item two arrives as coaching rather than as complaint. It closes the loop for anyone who has been asking guests for reviews, which is the single strongest thing you can do to keep that habit alive past week three. And it demonstrates that reviews are read, which changes how the team treats the ones that are not read out.
Use a name. A positive about "the team" achieves roughly nothing.
Item two: one theme, stated as a mechanism, with one behaviour attached.
"Three people this week mentioned the gap between starters and mains. Not the kitchen — the plates were up. So tonight: if you're waiting on a table's mains, tell them where it is before they wonder. That's the whole thing."
Four components in that: the frequency, the specific step, an explicit absolution of the group not at fault, and one behaviour a person can execute tonight.
What each component is doing
The frequency. "Three people this week" is why it matters. Without a number it is one bad review, and teams correctly discount one bad review.
The mechanism, not the mood. "We're slow" is a mood. "The gap between starters and mains" is a place in the service where something is happening. Only the second is coachable — and it is why the brief has to have done the theme counting properly, because you cannot deliver a specific mechanism you have not identified.
Who is not at fault. Skip this and the kitchen assumes the complaint is theirs, the floor assumes it is the kitchen's, and the actual behaviour change gets lost in a jurisdictional argument. Naming the exonerated party costs four words.
One behaviour. Not a standard, not a principle. A thing to say or do. "Tell them where it is before they wonder" can be done by someone who was not listening carefully.
One theme. Not five.
The brief may contain several drifts. Pre-shift gets one.
Five things to work on is zero things to work on — this is the most reliable failure in shift briefings and it happens because the person delivering it wants to be complete. Completeness is the enemy here in exactly the way it is the enemy of a written report.
If a second theme genuinely matters, it is next week's.
What never gets read out
Individual negative reviews, verbatim. Reading a guest's angry words to the people it is about produces defensiveness and nothing else. Summarise; the judgement is the useful part.
Anything naming a member of staff negatively. Ever, in front of the group. That is a conversation with one person, in private, and doing it at pre-shift will cost you the person.
Star ratings and averages. Nobody on the floor can act on 4.1, and the number invites an argument about whether it is fair.
Anything alleging harm. That is an incident, not a briefing item, and it has its own process.
Why it stops happening
The same reason everything else does: it needs somebody to have assembled two items before every shift. On the third difficult week the manager arrives four minutes before doors and does the version with no preparation — "reviews say we're slow" — which is worse than saying nothing, because it spends the team's patience without changing anything.
Ninety seconds of delivery needs about ten minutes of preparation it does not have. That is the gap OMMU covers: the new reviews arrive already sorted with the names and the mechanisms pulled out, so the two items are picked rather than found.