1Tonight's numbers
Rooms you can sell, what is booked, and how often guests cancel or fail to show. Use your PMS history for the same day of week and season.
2Cost of walking a guest
"Walking" a guest means sending someone with a confirmed booking to another hotel because you are full. List every cost of doing it once; the lines add up.
| Cost item | Amount ($) | |
|---|---|---|
| Total cost per walked guest |
Tip: "future business at risk" is the profit you expect to lose because some walked guests never book again. A common estimate is the chance they don't return times what they would spend over the next few years.
3Your risk appetite
The best average result sometimes means walking a guest now and then. If your brand standards say that should be rare, set a cap.
Tonight's recommendation
Based on the numbers above. "Expected" means the average over many nights like this one.
Expected net value by overbooking level
Net value = profit from rooms filled minus the cost of guests walked. The orange line is the chance of walking at least one guest.
Recommended level
Every level compared
| Overbook by | Booking limit | Guests expected to arrive | Expected empty rooms | Expected walked guests | Chance of walking anyone | Expected net value | Gain vs no overbooking |
|---|
Green row: recommended. The booking limit is rooms available plus the overbooking level: stop selling when bookings reach it.
414-night planner
One row per night. Type rooms available and bookings on the books; leave the rate columns blank to use the rates from step 1, or enter a night's own rate (for example a holiday or a big event). Revenue, walk cost and risk appetite come from steps 1 to 3.
| Night | Rooms available | On the books | Cancel % (optional) | No-show % (optional) | Overbook by | Booking limit | What to do | Chance of walking anyone | Gain vs no overbooking |
|---|
CSV columns: date, rooms_available, on_the_books, cancel_pct, no_show_pct. The last two can be left blank. Up to 14 nights.
How it's calculated
Show-up chance. Each booking turns into an arriving guest with probability p = (1 - cancellation rate) × (1 - no-show rate). Guests are assumed to decide independently of each other.
Arrivals. If you accept L bookings (the booking limit = rooms available C + overbooking level k), the number who arrive, S, follows a binomial distribution with L trials and chance p. The tool works out the full distribution exactly (in log space so large hotels don't overflow). Above 3,000 bookings it switches to a normal approximation with a continuity correction, which is accurate at that size.
For each level:
Expected empty rooms = E[max(C - S, 0)] Expected walked guests = E[max(S - C, 0)] Chance of walking any = P(S > C) Value of a filled room = room rate + extras - servicing cost Expected net value = value of a filled room × E[min(S, C)] - walk cost × E[walked]
Recommendation. The level with the highest expected net value. With a risk cap, only levels whose chance of walking anyone is at or below the cap are considered. Levels are tested from 0 up to well past the point where extra bookings stop paying.
Assumptions and limits. The model assumes you will sell up to the booking limit (overbooking only matters when demand is strong enough). It does not model walk-ins, early departures, stay-overs that extend, or guests on multi-night stays affecting the next night; no-show fees you collect are not counted, which makes it slightly cautious. Cancellations and no-shows often run together in groups (a storm, a cancelled concert), which this model treats as independent, so on unusual nights a lower cap is wise. Rates are best taken from at least 8 weeks of history for the same day of week and booking channel.
Your inputs are saved only in this browser (local storage). Nothing is sent anywhere.