Deliberate overbooking: how many rooms to oversell, from your own data

Deliberate overbooking means confirming more rooms than you hold, because some confirmed guests will cancel late or never arrive. Set the number per date from your own measured late-cancellation and no-show rates by segment and channel, cap it at the walks you can absorb, and have a person sign off.

It is 23:40, the last room went this morning, and a family is standing at the desk holding a confirmation. That is a walk. The question worth asking is whether the room was sold twice by accident or oversold by a decision someone signed off, and whether that decision was ever priced. This article gives you the measurement, the cost model, the cap and the review loop for a hotel overbooking strategy that is a decision rather than a hope. How the desk handles the guest in front of it is a separate skill, covered in how to walk a guest when your hotel is overbooked.

What is the difference between a deliberate oversell and a double booking?#

A double booking is a failure of inventory control; a deliberate oversell is a priced bet that a confirmed guest will not arrive. They look identical at the desk and need opposite fixes.

A double booking happens when two systems each believe a room is free: a rate left open on a channel after the room was sold, a manual reservation entered twice, a sync that ran late. Nobody intended to sell the room twice. If this is your problem, the fix belongs to distribution and channel sync, and it is covered in hotel double bookings across channels: causes and prevention.

A deliberate oversell happens when the hotel confirms more rooms than it holds on a date where its own history says a known share of confirmations will fall away before arrival. Someone chose the number. The fix, if the number is wrong, belongs to measurement and sign-off, not to sync.

The test is simple: can you name the person who decided, the number they set and the reason? If any of the three is missing, you are not overbooking, you are hoping.

Both problems end in the same place: a guest with a confirmation and no room. That is why the walk procedure has to exist before the first deliberate oversell, not after it.

How do you measure late cancellations and no-shows from your own data?#

Count, for each segment and channel, how many bookings were still confirmed when a set window opened and then cancelled inside that window or never arrived. The window must match the moment you actually decide to oversell.

Fix the window first. For example, cancellations inside the last 7 days before arrival plus no-shows. If you decide the oversell level 3 days out, a 30-day window measures the wrong behaviour.

Use the right denominator: bookings still confirmed when the window opened, not every booking ever made. Early cancellations that were never going to reach the window would otherwise dilute the rate and make the oversell look safer than it is.

Split by segment and channel. Free-cancellation OTA bookings, non-refundable direct bookings, agency bookings, corporate accounts and groups behave differently, and an average across them hides the only signal you have. Split by date type as well: ordinary weekdays, weekends, bayram nights, event dates, and the lead time at which the booking was made.

Use at least a full season and recompute monthly. A rate from two years ago is not evidence about next Saturday. Keep the raw booking list behind every percentage: you need to see which bookings fell away before you trust the number enough to bet rooms on it.

Treat group blocks and allotments separately. A block released by contract on its release date is not a no-show, and counting it as one inflates the rate. How much of a block has actually been used is its own report, covered in hotel pickup report: how to read booking pace day by day.

What the measurement table needs#

What does walking a guest actually cost?#

More than a room. A walk costs an alternative room of equal or better standard, transport, staff time, goodwill and sometimes the channel relationship, and the total is usually several times the room rate you gained.

The alternative room is bought at the worst moment: the night the city is full, which is exactly the night rates are highest and availability is thinnest. Transport both ways, the calls to find the room, and the manager on the phone at midnight are real costs, not overhead. Goodwill has a price too: a gesture, a free night, a refund of the difference, or a future stay you give away.

The channel consequence is slower but real. The agency, OTA or corporate client that made the booking will hear about the walk.

The legal layer matters when the guest came on a package. Under Article 51 of Law No. 6502, the package tour organizer is responsible to the consumer for any loss caused by the contract not being performed, or not performed properly, and the consumer may claim suitable compensation for wasted holiday time, according to the consolidated text of Law No. 6502 on mevzuat.gov.tr (opens in a new tab) (2026). The walk does not end at your desk. Confirm how this applies to your own contracts with a legal adviser.

Compare the two numbers honestly: one extra room night of revenue against a walk that may cost several room nights. The bet only pays at a low probability of walking, which is why the level is capped by the cost, not by the average.

How many rooms should you oversell on a given date?#

Start from the rooms still confirmed when your window opens, multiply by your measured late-cancellation and no-show rate for that mix, then cut the result back to the number of walks you could actually absorb on that date.

The expected fall-away is a multiplication, not a judgment: confirmed rooms times the measured rate for that segment mix and date type. But the expected value is not the decision. The downside is a walk, so choose a level you can survive on a bad night, not the level that is right on average.

Cap it twice: never more rooms than you could realistically re-accommodate at acceptable cost and standard in your area that night, and never more than the desk can handle if two walks land at once. Set the level per date, not per month, and re-check it daily as the mix changes. A level fixed a month out is a guess about a different booking base.

Record the decision: date, level, rate used, and the name of the person who signed it. Without that record the monthly review has nothing to compare.

Overbooking is a hotel decision. Software can compute the rate and display the level, but the level itself should not be set automatically, because the cost of a walk is a judgment the system does not hold.

A worked example#

For example, a 120-room hotel on one Saturday:

The numbers are illustrative only. Note that step 4, not step 3, sets the answer: the hotel expects to lose 6 rooms but oversells only 2, because 2 is what a walk would cost it.

When should you not overbook at all?#

Do not overbook dates where a fall-away is unlikely, where a walk would be very expensive or impossible, or where the room is not yours to release.

Contracted group blocks and allotments are a commitment with a release date. A group that arrives complete is not a no-show, and the rooms were never free to resell. Peak bayram nights and city-wide event dates are the worst nights to walk a guest: the whole city is full, so there is nowhere to send them, and the cost of the walk is at its highest.

Room types you cannot substitute belong on the list too: accessible rooms, connecting rooms, suites promised in a package, and any room a guest booked for a specific reason. Long stays and repeat guests multiply the cost, because the walk is not one night and the relationship is part of the price.

Do not overbook dates where your own history is thin: a new hotel, a new segment, a date type you have never sold, or a channel that changed its cancellation policy this season. And do not oversell a date when the person who would sign off is not available. An unsigned oversell is an accident with a delay. How to handle the guest who does not arrive at all is a different question again, covered in hotel no-show handling: when to charge, release or reinstate.

How do you review overbooking outcomes every month?#

Once a month, put the rooms you oversold next to the walks you caused and the revenue those extra rooms earned, and change the rate or the cap, not the principle.

Count five things: nights oversold, rooms oversold, walks, the recorded cost of each walk, and the revenue from the oversold rooms. If the walks cost more than the extra rooms earned, the level was too high, even if the month ended with no complaint.

Check the measurement itself: did the rate hold across segments, or did one channel change its cancellation behaviour and quietly move the number? Look for near misses as well. Nights where a walk was avoided by a phone call are evidence about the cap, not about the rate.

Write the change down and keep the previous level, so next month's comparison is possible. Overbooking that cannot be reviewed cannot be improved. The record is the asset, not the number.

What to do next#

Pull the last twelve months of confirmed bookings, compute your late-cancellation and no-show rate by segment and channel for the window you actually decide in, and set the oversell level for one date with a name and a rate written next to it. Measurement, cost, cap, sign-off and monthly review are one loop: the rate tells you what to expect, the cost of a walk tells you how much of it you dare to take, and the record tells you next month whether you were right.