Operating question · Demand forecasting
Stop prepping and staffing for last week's service.
Before Service turns POS history, weather and local events into a 15-minute interval forecast of covers and item mix — so the kitchen and the schedule are aligned with the day that's actually coming.
Focused output
IN REVIEWDemand forecast
System generated — operator review required.
What it answers
Gut-feel forecasts cost you twice — every shift.
Most restaurants project demand from a manager's memory of last week. The result is a kitchen that's either over-prepped on a slow night or 86'ing the hero dish at peak. Both end up on the P&L.
Tuesday slow, Friday packed — and the prep list looks the same.
Weather changes covers by ±15% and no one re-staffs.
Delivery channel mix moves and the POS doesn't flag it.
Hero dishes sell out before the rush ends.
Inputs
Signals used to prepare the brief
- Covers, items and channel mix forecast by 15-minute interval
- Weather and local event signals built in, not bolted on
- Confidence score with the reason, never a fake number
- One-click override that the model learns from
Output artifact
A confidence-scored forecast you can read in 30 seconds.
- Covers, items and channel mix forecast by 15-minute interval
- Weather and local event signals built in, not bolted on
- Confidence score with the reason, never a fake number
Confidence + rationale
Readable evidence, not a black box
- Sales history, reservations, weather, and operator overrides
- Confidence shown beside the recommendation
- Source signals stay available for review
Operator boundary
The system prepares the call. Your team confirms it.
Realistic pre-service scenario
Rain softens patio demand. The forecast lowers expected covers and explains which signal moved.
No automatic action. A human operator reviews every decision.
Service sequence
How a GM uses the forecast every morning.
- 06:00
Forecast finishes overnight using yesterday's actuals plus this morning's weather and reservations.
- 06:30
Chef reviews item-level demand on the prep screen and adjusts pull-forwards.
- 06:45
GM checks covers by daypart and confirms the labor calls in the daily brief.
- Mid-shift
If real covers diverge from forecast, Before Service re-runs and flags the next decision.
- Post-shift
Forecast vs actual reconciles automatically and feeds tomorrow's model.
Connected workflows
Forecasting is one input. Here's where it lands.
Forecast the day before you cook for it.
Import 4 weeks of POS history and your first confidence-scored forecast is ready in under 10 minutes.
FAQ
Questions operators actually ask.
No. Before Service sits on top of the POS, schedule and vendor invoices you already use. We turn that data into the daily decisions a POS won't tell you — what to prep, who to staff, what to re-price, what to 86.