The operator in this piece had already checked the obvious thing. Ticket times for the fortnight averaged 14:20 against a 16:00 target, the line was staffed to plan, and no chef on the rota thought anything had changed. On paper the kitchen was performing better than it had in spring.
Reviews said otherwise, in a very specific way. Eleven mentions of temperature in fourteen days — "burger wasn't hot", "chips were lukewarm", "food had clearly been sitting" — and every one of them from a table of four or more.
That last detail is the whole case study.
A composite, drawn from a pattern we see often enough to be worth writing down. The numbers are representative rather than one venue's audited figures.
Ticket time measures the wrong end of the journey
Ticket time stops when the food leaves the pass. The guest's experience of temperature begins when the plate lands in front of them. Everything in between — the wait for a runner, the walk, the pause while the runner works out which seat ordered the fish — is invisible to the kitchen display system and is usually where the heat goes.
Once you look at it that way, the "table of four or more" pattern stops being a curiosity and becomes the answer. A two-cover order goes out in one trip. A six-cover order goes out in two or three, and someone decides whether to hold the first plates until the rest are up. Hold them and they cool. Send them and the table sits watching two people eat.
Nobody on the floor experienced this as a failure. They experienced it as juggling, which is what service feels like anyway.
What the three changes actually were
None of them touched a recipe, a supplier or a piece of equipment. All three were announced in a pre-shift huddle and cost nothing.
One: large tables get a named runner before the ticket is fired. The expo called out table sizes at the start of each wave, and the manager assigned a runner to anything over four. The point was not extra labour — it was removing the two-second hesitation of whose table is that, which is what created the pause at the pass.
Two: plates go out in the order the kitchen finishes them, and the runner says so. "Your fish is thirty seconds behind, I didn't want to sit on this." A sentence that converts a service failure into evident care. This was the change the floor resisted most and the one that showed up fastest in the reviews.
Three: the pass got a hard rule — nothing waits more than ninety seconds. If a plate hit ninety seconds under the lamp, expo either sent it or remade it. That forced the earlier two changes to be real rather than aspirational, because the only way to hit ninety seconds consistently was to have a runner already standing there.
The ten days, honestly read
| Metric | Prior 14 days | Following 10 days |
|---|---|---|
| Temperature mentions in reviews | 11 | 2 |
| Remakes logged | 19 | 7 |
| Average ticket time | 14:20 | 14:55 |
| Average rating, new reviews | 3.9 | 4.4 |
Note the third row. Ticket time got worse by thirty-five seconds, because expo started holding plates rather than pushing them out to an empty floor. This is the part that matters if you are going to try it: the number the kitchen is measured on moved in the wrong direction while the guest experience improved. If your GM is bonused on ticket time, you have to say this out loud before you start, or the change will be quietly reversed inside a month.
Ten days and thirteen reviews is also not proof. Two temperature mentions instead of eleven could be noise; the remake count is the more trustworthy figure because it is a count of every occurrence rather than a count of people who chose to write something down. We wrote about what a single week of fixes can and cannot demonstrate separately, and the honest position here is that this is a strong signal worth keeping, not a proven causal result.
Why reviews found it and the reports did not
The reports were measuring the kitchen, and the kitchen was fine. No internal metric in this business covered the gap between the pass and the table, because no system owns that gap — it belongs to expo and to the floor jointly, which in practice means to nobody.
Guest reviews do not respect the org chart. They describe the plate as it arrived, which is the only measurement that includes every step. That is genuinely useful, and it is also why review complaints so often feel unfair to the department they land on: "cold food" sounds like a kitchen accusation and was, in this case, a floor scheduling finding.
The practical move is to read temperature complaints as a question rather than a verdict. Where between the lamp and the table did this go wrong? There are only about four candidate answers, and the reviews usually tell you which one if you sort them by table size, daypart and section before you read the text.
If you want to run the same check
Pull your last sixty reviews, keep only the ones mentioning temperature, and tag each with the party size if you can reconstruct it from the booking system. If they skew to large tables, the finding is a handoff problem and the three changes above are a reasonable place to start. If they skew to one section, it is a distance or a section-size problem. If they are evenly spread, it probably is the kitchen, and this piece is not the one you need.
Doing that sort once is an hour of work. Doing it every month is the thing almost nobody manages, which is what OMMU exists to cover — the same read done continuously, arriving as one short email about what shifted this week and what is worth fixing before Friday.