"Slow service" is the least useful complaint you can receive, and the most common. You can't coach it, you can't rota against it, and reading it more carefully doesn't help because the guest genuinely doesn't know what was slow. They only know that the evening had gaps in it.
But a table changes hands about five times between the door and the card machine, and the guest's own wording — not their conclusion, their wording — usually indicates which handoff dropped it. That's fifteen minutes of work and it turns an uncoachable complaint into a specific thing one named person can do differently on Friday.
The five handoffs
Door to table. Host takes the booking, seats them, and someone has to know they've arrived.
Table to bar. Drinks order leaves the table and has to come back.
Kitchen to table. Food is ready and has to be carried by someone who isn't the person who cooked it.
Server to server. Shift change, break cover, section swap — the table's owner changes mid-meal.
Table to till. The bill, and whoever runs it.
Every one of these has a characteristic complaint, and they don't sound like each other once you know what you're looking for.
The phrases, and what they point at
| The guest wrote | Likely handoff | What to look at |
|---|---|---|
| "Sat for ten minutes before anyone came" | Door to table | Whether seating notifies the section, or the server has to notice |
| "Took ages to get a drink" / "still waiting when the food came" | Table to bar | Order sequence, and who fetches |
| "Food was lukewarm" / "sat under the lamp" | Kitchen to table | Seconds on the pass; whether a runner exists at that hour |
| "Had to ask three people" / "our waitress vanished" | Server to server | Break rota, shift change time, table ownership at changeover |
| "Waited 20 minutes for the bill" / "had to go up to pay" | Table to till | Who's authorised on the till at that hour |
| "Nobody checked on us" | None of them — no owner at all | Whether the section was covered or split |
The last row is the important one. "Nobody checked on us" is not a slow handoff, it's an absent owner, and it usually means a section was informally split between two people who each assumed the other had it.
The fifteen minutes
Minutes 0–6: pull twenty-five reviews and tag them. Anything mentioning timing, waiting, coldness or attention, going back three months. You want the guest's words in one column and a handoff in the next. Guess when it's ambiguous — a rough tag is fine, because you're looking for a lump rather than precision.
Minutes 6–9: count and note the hour. Which handoff has the most, and what time do those reviews describe? A handoff problem with no time concentration is usually a process gap. One that only appears between eight and nine is a coverage gap. That distinction is the whole point of the exercise.
Minutes 9–12: ask three questions of the winning handoff.
- Who is responsible for it by name, at that hour, on that day?
- How do they know it needs doing — do they see it, or does someone tell them?
- What are they doing at that same moment that could stop them?
Most handoff failures answer question two with "they see it". A handoff that depends on somebody noticing works fine at 40 covers and fails at 90.
Minutes 12–15: pick the one you'll go and watch. Because the tagging is a hypothesis, not a finding. Which brings us to the part that isn't optional.
Then go and stand there at the hour in question
Fifteen minutes of tagging tells you where to look. It cannot tell you what's happening, and if you act on the spreadsheet alone you will reorganise something that was working.
Go at the hour the reviews describe, stand where you can see the handoff, and count seconds. Plates on the pass: count them. Time from seating to first contact: count it. Don't help, don't explain why you're standing there, and don't do it on a Wednesday if the complaints are about Saturday. The rest of how to make that observation useful is here.
Coordination gap or staffing gap?
This is where handoff audits go wrong, in both directions.
It's a coordination gap if the same number of people, doing the same jobs in a different order, would fix it. Drinks going in before the food discussion. The host putting the docket on the section rather than expecting the server to spot the table. Someone calling the plate rather than assuming the runner is looking. These are free, and they hold if you brief them consistently.
It's a staffing gap if the handoff fails specifically and only at the busiest hour, and the person responsible for it is demonstrably doing something else at that moment. If your 8pm cold-food problem is a runner who is also the only person on the floor's coffee station, no amount of reordering fixes it — you're asking one person to be in two places at the hour it matters most.
The test is simple: does the failure appear at 40 covers as well as 90? If it appears at both, coordination. If only at 90, coverage.
And the honest bit — most operators, including us when we were doing this badly, default to the coordination explanation, because it's the one that costs nothing. We wrote up a case where it took three weeks to accept it was a rota gap, and three weeks of process briefings had already been tried by then.
Keep the cadence light
Once a month, not weekly. Review volumes for a single site are too small for a weekly tag to show anything but noise, and a monthly audit that actually happens beats a weekly one that survives until the middle of October.
Track the same tag counts each month so you can tell whether the handoff you fixed stayed fixed. That's the entire measurement: one number per handoff, twelve rows a year. If the lump moves from "table to bar" to "kitchen to table", you didn't fail — you fixed one and found the next one.
We tag the handoff phrases automatically so the six minutes of reading becomes a count you already have. The standing-and-counting part, unfortunately, is not automatable, and it's the part that tells you what's actually wrong.