Capacity Management - Why Hotels and Airlines Deliberately Overbook

With my recent investment blog post about Air Canada current troubles, I began to look at the service industry and capacity management (a staple concern and metric of success).

Any seasoned manager in the hospitality industry is familiar with capacity management. In the services industry, maximizing capacity (as services can't be time shifted to other periods, they are either consumed or not) is a great way to increase overall profitability.


Most people would assume that hotels book until they are at 100% capacity and fill all the rooms. They are usually surprised when they find out that the hotel is over booked. How can that be? It's not operationally difficult to track the number of rooms and the number of guests! Let's explore why:


Profitability for hotels, the top two most important metrics for revenue are ARR and OCC (Average room rate and occupancy rate respectively).

Revenue = ARR x OCC

Now, many services will try to boost their ARR through marketing, improved services etc, but for this example, let's assumed it's fixed. It's quite obvious that the best way to boost revenue is to increase occupancy (have more people visit).

Now as a hotel you've booked 100% of your rooms. Great right? Well not entirely. Unfortunately, for a variety of reasons, people cancel, have no shows, etc. So your OCC drops from 100% to it's real life value (usually 70-80%). One solution to recover costs is to instigate a cancellation or no-show fee to recover some of the 20-30% of OCC lost (in the form of a 10-50% fee depending on when the cancellation is made).

Revenue = (ARR x OCC) + (Cancellation fee x number of cancellations)

Now some clever analysts determined that by trying to predict how many people would cancel, it was possible to deliberately overbook the hotel in hopes of boosting revenue. For instance, if 10% of people booking canceled, it is possible to over book the hotel proportionally so that the final result (expected visitors) would be 100% (and you could also charge cancellations the regular fee).

OCC = # of bookings x (1 - Cancellation rate)
[Target = 100%]

So if you expected 10% of people to cancel (90% show up), and you had 100 rooms, you would book about ~110 guests, expecting only 99 - 100 to show up.

However, this begins to creep into the case most of us are familiar with: more people show up than expected. In this case, the hotel has a problem. For simplicity sake, let's just say that in order to resolve this issue (upgrades, alternate accommodations, loss of good will etc) there is an "all encompassing" penalty cost of Y dollars.

The full revenue cost formula becomes:

Revenue = (ARR x OCC) + (Cancellation fee x number of cancellations)

- (# of over bookings x Y)

Now in order to maximize profit, the derivative of the revenue function is taken with respect to occupancy, to find the sweet spot where revenue is maximized. Essentially, you begin to counterbalance the benefits of maximizing occupancy with the cons of over booking.

Of course there are some major assumptions which (if refined) can provide a better picture of the profitability and costs:

• Average cancellation rate can be refined to more by accurately segregating guest demographics to improve accuracy in predictions (is cancellation affected by age, geography, time of year, other consumer qualities)
• "Cost" of overbooking (especially if it's pervasive) can be detrimental to brand quality and could adversely affect ARR (guests feel like the quality of services drop)
• Hospitality is seasonal by nature so assumptions and rules governing overbooking need to change accordingly
• Over booking assumptions for finding alternate accomodations might be prohibative if there are simply no alternates during a "hot" season