Your price change "worked" on the day a food critic showed up. Did it, though? This week you learn to tell your own decisions apart from the luck of the draw.
Learning objectives. After this week you can (1) use the Daily Report as a KPI dashboard; (2) separate signal from noise; and (3) design a fair before-and-after test.
Remember this: one day proves almost nothing. Read the event line before you believe a result.
1) The report as a dashboard
A KPI (key performance indicator) is a number chosen in advance to answer one question. Your report has three zones: Results (what happened), Mechanics (why), and Lesson + Streak + Alerts.
| KPI | Question it answers |
|---|---|
| Utilization (units served ÷ capacity) | Is my crew sized to my demand? |
| Conversion | What did walk-ups do when they saw my price? |
| Stockout | Did food or crew cap my best day? |
| Foot traffic | Is demand my problem, or is serving it? |
| Contribution per day | Is the whole machine making money? |
A weak day breaks down into foot traffic × conversion × capacity: find which one moved before deciding anything. Profit is a lagging indicator: it reports what already happened. Conversion, utilization and stockouts are leading: they move first, while you can still act.
2) Signal vs. noise
Signal is repeatable: a district's weekday pattern, your reputation, your price. Noise is the random event each day draws, printed on your report as Today's event. Judge a change on several comparable days, not one. Remember that correlation is not causation, and smooth the noise with a moving average:
3-Day Moving Average = (Day 1 + Day 2 + Day 3) ÷ 3
Profits of $410, $455 and $390 average $418. If tomorrow prints $340 with a slow-weekday event, ask whether the average is moving before asking what went wrong.
3) A fair test across days
Change one lever, compare the same district on the same weekday, and note every day's event. A price test over four Wednesdays: old price $410 (normal day) and $520 (busy-lunch event); new price $455 (normal) and $390 (slow-weekday event). The averages say the old price won, $465 to $422.50. The two normal days say the new price won, $455 to $410. The averages were bent by one lucky and one unlucky draw. Write your decision rule before the data arrives, so the data can't be bent to fit the decision you wanted.
Quick self-check
(a) Capacity 380, served 342: utilization? (b) Profits of $300, $360 and $330: 3-day average? (c) You raised price on a festival day and revenue jumped. What can you conclude? (Answers: 90%; $330; nothing yet, the festival may explain it.)
Common mistakes
- Reacting to one day. One event draw can undo a good strategy in your head.
- Cherry-picking. Quoting only your festival day proves nothing.
- A dashboard without decisions. A KPI you never act on is decoration.
This week in the arena
Run one honest experiment. Change one lever, hold everything else, and log one KPI with each day's event on comparable days before and after. Report your verdict: signal or noise.