The obvious way to report six sites is a row per site. It is also the way that guarantees nobody reads it, because a six-row table asks the reader to do the comparison — and the comparison is the only difficult part of the job.
A table of six rows with four columns each contains twenty-four numbers and one finding. Presenting it as twenty-four numbers is a decision to hand the analysis back to the busiest person in the group.
Exception reporting, which is the whole answer
Report the sites that are unusual. Say explicitly that the others are not.
Five of six sites normal this week.
Kingsway: complaint rate 2.9 per 100 covers against a group average of 1.6. Fourth consecutive week above group. Wait times and staff attitude.
Northgate: value complaints 5 this week, none in the previous six. All mention the new set menu.
Two named sites, one screen, no table. An area manager reads it in twenty seconds and knows exactly where Monday goes.
The opening line matters as much as the exceptions. Without "five of six normal", the reader wonders about the four unnamed sites and mentally reconstructs the table you were trying to avoid.
The normalisation this depends on
Exception reporting is only honest if the comparison is fair, and raw counts across sites of different sizes are not fair. A 300-cover site will always generate more complaints than a 90-cover site, and reporting counts guarantees your largest location is permanently flagged.
So everything in the brief is complaints per hundred covers. This single change is what makes multi-site review reporting work, and skipping it is the most common reason these briefs get dismissed by the operators receiving them — a GM whose site is flagged every week for being busy learns to ignore the brief.
The mechanics of getting that comparison right are their own piece.
The three exception types worth reporting
Above group, persistently. Kingsway above average for four consecutive weeks. The word consecutive is the finding — one week above average is expected for roughly half your sites.
Changed sharply, regardless of level. Northgate going from zero value complaints to five. Still below group average in absolute terms, and much more interesting than Kingsway, because something happened recently and recent causes are findable. A site can be your best performer and still be the most urgent thing in the brief.
Shared across sites. The category most easily missed by per-site reporting. Three locations getting the same complaint in the same week is not three local problems, and telling three GMs to fix it separately produces three different answers, none of which addresses a supplier or a recipe or a policy. Worth its own line:
Group-wide: portion complaints on the chicken dish at four of six sites since the spec change on the 12th.
Distinguishing this from local variation is the harder analytical question.
What to leave out
Per-site ratings. They move too slowly to be weekly and they invite league tables. A league table changes the incentive from fixing problems to not being last, and the fastest route to not being last is gating your review requests.
Reply rates by site. Monthly, if at all.
The good news. One line, if there is something genuinely worth passing on. Not a section.
Any site that is normal. The point of the format is that silence about a site is information. If normal sites get a line each, the reader has to check every line, and you have written a table with extra words.
The ranking question the format does not answer
The brief names Kingsway and Northgate. It does not say which one gets the visit on Monday, and it should not — that is a judgement involving things reviews do not know: which GM is new, which site had a refit, where you were last week.
But the inputs to that judgement can be made explicit, and a priority score is a reasonable way to make the reasoning visible rather than intuitive. What the brief owes the reader is the two candidates and the reason each is flagged.
Where the work actually is
The format takes four minutes to write once the numbers exist. The numbers require covers by site by week, complaint themes tagged consistently across six sets of reviews, and a group baseline that goes back far enough for "four consecutive weeks" to mean anything.
That is the entire difficulty, and it is why multi-site review reporting is usually either a raw feed of reviews or a monthly deck nobody opens. The assembly is a job, and it is a job that competes with running restaurants.
Ours arrives in roughly the format above; the sample report is a real one sent to a multi-location owner.