Turning the weekly POS export into a plain-English shift report
The POS produces a wall of numbers; what a manager and staff actually want is the story — what sold, what is slipping, and where labor or food cost ran hot. Instead of building a spreadsheet, managers paste the export and ask for a short, plain summary they can act on and share.
You are a restaurant analyst. Turn the numbers below into a short, plain-English weekly summary. This week's POS and labor figures, with last week's for comparison: {{pos_numbers}} Anything unusual that week (weather, event, closure): {{context_notes}} Write: 1. A 150-word summary. Bottom line first (net sales and the single biggest change from last week). Then the largest driver explained in plain terms. Then three items trending up and three trending down. Close with one thing to test next week. 2. A three-bullet "at a glance" block: net sales, labor %, food cost %. Rules: use ONLY the numbers I paste. Always name the comparison ("down 8% vs. last week"). If a figure I referenced is missing, write [MISSING] rather than estimating it. Do not forecast future revenue or make any guarantee. Do not recompute percentages I already provided — quote mine.
Fill in your details and the prompt updates live — then copy.
Net sales were $48,200, down 7.5% vs. last week ($52,100). The whole drop tracks to the weekend rain — the patio was closed Friday and Saturday, and covers fell to 1,940 from 2,110. Average check held at $24.85, so this was a traffic problem, not a spending one. Labor climbed to 31% from 28% because we were staffed for the patio that never opened. Up: burgers, wings, and the IPA. Down: the fish special and weekend brunch. Test next week: a rain-day call to pull one server off the floor by 6 if the patio is closed. At a glance — net sales $48,200; labor 31%; food cost 33%.
The full workflow
- Export the week's sales, labor, and cost numbers with the prior-week column.
- Strip guest payment data and swap employee names for roles or initials.
- Generate the summary, then reconcile every figure against the POS report.
- Share the plain-English version at your next manager or staff meeting.
Watch out for
Language models miscalculate and quietly fill in blank cells — reconcile every number against the actual POS report before you act on it or share it.
Employee sales, void, comp, and discount data is sensitive performance information — do not tie it to names in a consumer AI tool; use roles or initials.
Strip customer payment card data and any personal guest details from exports before pasting — that data is covered by PCI rules and does not belong in a chatbot.
Where this comes from
Every use case on this site is grounded in real reports from working restaurant managers — not invented by us.