AI Demand Forecasting
The AI Demand Forecasting module uses machine learning models trained on your historical data to predict future sales volume, recommend staffing levels, and optimize inventory purchasing.
How It Works
The forecasting engine analyzes multiple data signals to produce predictions at 15-minute, hourly, and daily granularities:
- Historical sales data — same day of week, seasonal trends, year-over-year growth
- Local events — concerts, sports games, conventions sourced from event APIs
- Weather forecasts — temperature, precipitation, and severe weather alerts
- Holiday calendar — national holidays, school schedules, and local observances
- Promotional activity — active discounts, LTOs, and marketing campaigns
Sales Predictions
View forecasted revenue and item-level demand for up to 14 days ahead. Predictions include confidence intervals so managers understand the range of likely outcomes.
| Forecast Horizon | Granularity | Typical Accuracy |
|---|---|---|
| Next 24 hours | 15-minute intervals | 92-95% |
| 2-7 days out | Hourly intervals | 85-90% |
| 8-14 days out | Daily totals | 78-85% |
Item-level forecasts show predicted quantities for each menu item, enabling prep planning and reducing over-production waste.
Staffing Recommendations
Based on predicted customer volume and your configured service-level targets, the system recommends staff counts by role and by 30-minute block.
- Front-of-house — servers, hosts, bussers, bartenders
- Back-of-house — line cooks, prep cooks, dishwashers
- Drive-thru — order takers, payment window, expeditors
- Recommendations account for labor cost targets and overtime thresholds
Managers can accept recommendations directly into the scheduling module or adjust them before publishing.
Inventory Planning
Demand forecasts feed directly into the inventory ordering engine. For each ingredient, the system calculates:
- Projected usage based on forecasted menu item sales and recipe mappings
- Current on-hand stock from real-time inventory
- Recommended order quantity accounting for vendor lead times and par levels
- Suggested order date to ensure delivery before stock runs out
Weather Impact Analysis
The weather module quantifies how local conditions affect your business. Over time, the model learns location-specific patterns such as:
- Rain reducing dine-in traffic but increasing delivery orders
- Extreme heat boosting cold beverage and dessert sales
- Snow events causing early closures and staff call-outs
- Pleasant weekend weather driving patio and drive-thru volume
A weather impact score is displayed alongside every forecast to explain deviations from baseline predictions.
Reviewing Forecast Accuracy
A dedicated accuracy dashboard compares past predictions against actual results. Use this to:
- Build confidence in the model's recommendations
- Identify categories where the model underperforms
- Provide feedback that improves future predictions