Tourism Industry Insight: Turnaway Data Can Improve Hotel Forecasts

Thirty days before a holiday weekend, a 40-villa resort sells its final room. The booking curve stops rising, and the revenue report records demand of 40 villas. Yet during the following week, twelve travellers search for unavailable dates, three advisers request longer stays, and one family abandons the booking engine because the required room sequence cannot be offered. The hotel sold exactly 40 villas, but demand may have been considerably higher. Next year’s forecast will be biased if it learns only from what the property managed to sell.

Sales can become a ceiling rather than a demand signal

Statistics calls this censoring. The outcome is observed only up to a limit—in this case, available inventory.

A simple relationship is:

Observed bookings = minimum of underlying demand and available inventory.

If 28 travellers want a room and 40 are available, bookings reveal demand reasonably well. If 55 travellers want one but only 40 are available, the same report shows 40 bookings. The data cannot distinguish between exactly full and substantially oversubscribed without another signal.

Restrictions create subtler censoring. A villa may be physically empty but unavailable to someone requesting a two-night stay because a minimum-stay rule is active. A family may disappear from the data because no connected configuration is offered. A rate plan may close before lower-paying demand completes its normal booking window.

This extends The Stay-Pattern Advantage: protecting valuable multi-night patterns can be rational, but rejected requests still contain information about what the market wanted. They should not be mistaken for demand that never existed.

Hotel revenue-management research describes recovering this missing quantity as demand unconstraining. Comparative research has shown that the method chosen to estimate censored demand can materially affect forecasts and revenue decisions. More recent accommodation research makes the wider point that observed bookings are constrained by available supply wherever inventory binds.

Record why the booking disappeared

Hotels do not need a complicated model to begin. They need to preserve evidence at the moment demand becomes invisible.

Useful signals include searches returning no availability, booking-engine exits after an unavailable result, denied telephone or adviser requests, waitlist additions, closed-channel dates and enquiries that fail because of room type, stay restriction, occupancy limit or transfer availability. These are not equivalent. A casual search has weaker intent than a traveller who submits dates, party size and contact details.

Timing matters as well. Hidden Clock: How Lead Time Can Help Operators Plan Better shows that bookings arriving 120 days before travel carry different information from requests appearing two days out. Turnaways should therefore be recorded by arrival date, request date, channel, room type, length of stay, party size and reason for rejection.

Cancellations complicate the picture. A hotel may reject demand because it appears full, only for rooms to return later. As The Cancellation Policy That Makes Forecasts Lie explains, booked inventory has different probabilities of surviving until arrival. High turnaways combined with high cancellation risk may support a carefully tested overbooking or waitlist policy—not an automatic increase in prices.

Estimate a range, then test the decision

Not every unsuccessful search would have converted. Adding all turnaways to bookings would exaggerate demand. A practical approach assigns different conversion assumptions to different signals.

Suppose the resort sold 40 villas, received ten unavailable booking-engine searches and rejected four qualified adviser requests. If historical evidence suggests 10–20% of unavailable searches and 50–75% of qualified requests would have converted, estimated unconstrained demand might be approximately 43–45 villas. That is an illustrative range, not an observed fact.

This is where The Confidence Band becomes important. Missing demand should be expressed as a range and tested against later outcomes, not converted into a falsely precise number.

Channel evidence matters too. How Hotels Can Make OTA Commissions Work Harder asks whether intermediaries create incremental demand. A channel showing repeated unavailable searches on particular dates may be revealing genuine unmet demand—or merely comparison activity. Subsequent bookings, waitlist conversion and reopened-inventory performance help distinguish them.

Return to the 40-villa resort. Its operating report still records 40 rooms sold, because that is what happened. But its planning record should also preserve the twelve searches, four qualified requests and reasons those travellers could not book.

That second record changes next year’s decision. It may justify holding rate, adjusting stay controls, reopening a channel, redesigning room combinations or expanding capacity. More importantly, it prevents a full hotel from teaching the forecasting system that demand stopped precisely when inventory did.

Bookings show captured demand. Turnaways reveal the market that remained outside the door.