Reservierungen
No-show prediction
A model that scores each upcoming reservation with a no-show probability so you can manage overbooking responsibly.
Every upcoming reservation gets a no-show probability, surfaced as a colored badge in the reservation list. The score is meant to inform decisions, not make them — a host can always decide to keep a high-risk booking on the books unchanged.
Inputs
The model considers:
- Guest history — past confirmed, completed, no-show, and cancellation rates.
- Booking lead time — how far in advance the booking was made.
- Time of day and day of week.
- Party size.
- Local weather forecast on the day of the reservation.
At least 30 days of history
Predictions only appear once the restaurant has 30 days of reservation data. Without enough history the model would produce noisy, misleading scores. The badge is simply hidden until then.
Confidence buckets
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
| Low | badge | — | — | Predicted no-show probability below 10%. Treat as a normal booking. |
| Medium | badge | — | — | 10% – 25% probability. Consider sending an extra reminder closer to the time. |
| High | badge | — | — | Above 25% probability. Often a candidate for a confirmation phone call or a deposit-request flow. |
Suggested overbooking
When you ask DineOS to fit an extra party into a service that is already at capacity, the assistant checks the aggregate no-show probability of the existing bookings in that service window and tells you how many seats are likely to actually free up. This is a suggestion — the host approves or rejects the overbook.
Use overbooking responsibly
A predicted no-show is still a guest who might show up. Never overbook beyond what you can recover from with grace — a quick free drink at the bar, or a credit on a future visit. The audit log records every approved overbook for accountability.