This gets published as a seven-day plan and it is not one. It is about an hour, once, and then twenty minutes a month afterwards. Spreading it across a week is how it gets abandoned on day three.
Worked below on a composite six-site group so the arithmetic is visible.
Step 1 — Covers by site (10 minutes)
From the till, last 90 days, weekly average. This is the denominator and nothing downstream works without it.
| Site | Covers/week |
|---|---|
| Kingsway | 310 |
| Northgate | 285 |
| Harbour | 240 |
| Milton | 190 |
| Deacon St | 165 |
| Old Mill | 95 |
Step 2 — Pull 90 days of reviews and keep the negatives (15 minutes)
Everything at three stars or below, across every platform, per site. Do not read them properly yet. You are counting.
| Site | Negatives, 90 days |
|---|---|
| Northgate | 26 |
| Kingsway | 24 |
| Old Mill | 13 |
| Harbour | 11 |
| Milton | 9 |
| Deacon St | 7 |
Northgate looks like the problem, with Kingsway just behind. Hold that thought.
Step 3 — Divide (5 minutes)
Negatives per hundred covers, over 90 days. Roughly 13 weeks, so the denominator is weekly covers × 13, divided by 100.
| Site | Covers (90 days) | Rate per 100 covers | vs group |
|---|---|---|---|
| Old Mill | 1,235 | 1.05 | 1.9× group |
| Northgate | 3,705 | 0.70 | 1.3× |
| Kingsway | 4,030 | 0.60 | 1.1× |
| Milton | 2,470 | 0.36 | 0.7× |
| Harbour | 3,120 | 0.35 | 0.7× |
| Deacon St | 2,145 | 0.33 | 0.6× |
Group rate: 90 negatives across 16,705 covers, or 0.54 per hundred.
Old Mill — the smallest site, thirteen complaints, the one nobody was worried about — has by some distance the worst rate in the group, at nearly twice the average. Northgate is genuinely elevated. And Kingsway, which had the second highest raw count and would have been the first stop on a list sorted by complaints, is within about ten per cent of the group and probably fine.
This step is the audit. Everything before it is data collection and everything after it is interpretation. The reasoning is in the normalisation piece.
Step 4 — Now read them, tagging two things only (20 minutes)
For each negative at the two or three flagged sites: a theme, and a daypart.
Keep the theme list short — six or seven categories. Speed, temperature, attention, cleanliness, money, staff conduct, other. Long taxonomies produce inconsistent tagging and inconsistent tagging destroys the comparison.
Old Mill, 13 negatives:
- Attention: 7 (all Friday and Saturday evening)
- Money: 3
- Speed: 3
Seven of thirteen complaints, at one site, about one thing, in one daypart. That is a finding, and it is specific enough to act on — attention complaints concentrated in an evening service are usually a section-size or break-scheduling problem rather than a staff quality one, which is exactly the case in the Friday rota story.
Step 5 — Check whether it is only yours (5 minutes)
Before anyone drives to Old Mill: does the same theme appear at other sites?
Money complaints, in this composite, turned out to run at every site since a menu change — a group problem that would have been "fixed" locally six times. Distinguishing the two is a piece in itself, and this five-minute check is what stops you sending an area manager to solve a head-office decision.
Step 6 — Write four lines and stop (5 minutes)
Old Mill: 1.05 negatives per 100 covers, 1.9× group. Seven of thirteen about attention, all Friday/Saturday evening. Likely section sizes or break cover.
Northgate: 0.70, high volume and elevated rate. Mixed themes, no single cause. Needs a longer look.
Money complaints at all six sites since the July menu. Group issue, not local.
Kingsway, Harbour, Deacon St: normal.
Then act on one thing. An audit producing six actions produces none.
The two traps in this procedure
Old Mill's thirteen complaints is a small number. At 95 covers a week, 90 days is about 1,235 covers, and rates on small denominators are volatile — three complaints either way moves Old Mill's rate by a quarter. Seven attention complaints in one daypart is a strong enough signal to go and look. It is not strong enough to restructure a rota before you have.
The tagging is where consistency dies. Done by two people, or by one person in two sittings, the categories drift and next quarter's comparison is worthless. If you do nothing else, write the six categories down with an example of each.
The twenty minutes a month
Repeat steps 2 and 3 only. Rates by site, compared to last month. That is the whole ongoing commitment, and it is what makes the original hour worth anything — a single snapshot cannot tell you what is moving.
The monthly twenty minutes is also the part that stops in November, which is why we built OMMU to hold the rates and the tagging continuously. What arrives is the four lines, with the comparison already made.