Not all wasting assets in inventory are products in the traditional sense of an object sitting on a shelf. Two important examples are airline tickets and hotel rooms. Here is why these can be thought of as a completely wasting asset. An airline has a regularly scheduled flight, let's say from Raleigh Durham Airport to San Francisco. The minimal variable cost to the airline if it fills that seat is just the extra fuel to lift perhaps an additional 100 kilograms of a person and their baggage and maybe a bottle of water and some peanuts. Almost all the cost associated with that seat are fixed, or sunk costs. The airline can't spend less on its airplane lease or on maintenance, or pay the pilot and crew less just because there was an empty seat on that flight. The money spent to create the seat is completely wasted if the seat is not sold in time. Similarly, on any given night a hotel room that goes unrented represents a lost opportunity for the hotel owner. The variable cost of renting a room is very low the supply, the cost of supplying clean linens, and maybe a tiny bottle of shampoo. Almost all the cost of making that hotel room available is unrecoverable fixed cost, it is sunk cost. Visiting Cocera for a workshop this year I stayed in The Grand Hotel in Sunnyvale from a Saturday through a Friday night. I was quoted a different room price for the same exact room for each night of the week. This is variable pricing. How does a hotel come up with it's variable pricing? Business analytics of course. We will look at a simplified model of applying business analytics to the Occupancy rate metric. Assume that a hotel has been charging a single fixed rate of $150 for all its basic rooms, every night, for the past two years. Looking at the data from the last two years, the hotel has a slightly above average occupancy rate of 66.4%. I'm making these numbers up. The average hotel occupancy rate in the US was 62.3% and 64.4% in 2013 and 14. So far, nothing useful. However, if we recalculate average occupancy rates sorted by day of the week, We see the following pattern. There are clearly many more people traveling to the hotel during the week and Wednesday is a peak day, with far fewer travelers on Saturday and Sunday. Business travelers want to be home for weekends and holidays. This tells us this is mostly a business hotel. We can infer from these weekly data, that we might be able to increase overall occupancy rates, if we charged a somewhat lower rate for Mondays, Tuesdays, and Thursdays, and a much lower rate for rooms on Fridays, Saturdays, and Sundays. Wednesday is a special case. If we dig a little deeper into the occupancy data for Wednesday, it is clear that there are seasonal patterns to occupancy as well as weekly patterns. So the Wednesday before the Thanksgiving holiday, let's say it's 35% occupancy. Wednesday during the week after Christmas, only 40%. The Wednesdays in August, 75%. Our average Wednesday we'll say is 95%. And our Wednesdays in September, October March and April were 100%, meaning the hotel was completely sold out on Wednesdays in those months. So we are turning away an unknown number of guests. We have more demand than supply in September, October, March, and April on Wednesdays. It's not clear how much demand exceeds supply, but it does appear that we are charging less than we could for those nights. So our strategy for Wednesday would be to charge maybe three different prices depending on time of year. Less than $150 for holidays and August, the same $150 for most Wednesdays and more than $150 for Wednesdays in the four peak business travel months. By how much should we raise or lower prices? Here, we get into the arena of complex models that weigh price against demand. The best basis for these models is empirical. Try different prices and see whether the rooms sell too slowly or too quickly. One way to do this is to list lots of room on online travel agency sites, or OTA sites. OTAs are not ideal for hotels because they charge commissions of 10 to 25% of the room rate, so hotels would rather sell direct to consumers. However, a typical strategy would be to assign a block of rooms on a particular night to an OTA at a pre agreed price for that block. If the OTA fails to sell all of them the unsold rooms are released back to the hotel to try to sell directly. OTA's provide a mechanism for hotels to test different prices on different dates to see whether or not that block of rooms sells out at that price. A challenge hotels face when experimenting with their published or rack price, tat's the undiscounted room rate, is that it becomes a new maximum for even the price insensitive, last minute business traveller. Ideally, our hotel would like to be able to offer some rooms at a lower rate, even all the way down to their theoretical break even price, what's called the floor rate, to reach vacation travelers while at the same time maintaining higher prices for our business travelers. There is one way that this is possible. The very lowest prices for a given room are typically found in these so called opaque, inventory market. Over opaque inventory pricing, OTA's, and at various times this has been available through Expedia, Hotwire, Priceline, Qunar, and Travelocity, they offer rooms in a given city by price only, maybe also providing some basic rating and category information such as the number of stars for the hotel. But they do not disclose the hotel name or any other information that would allow you to identify what hotel they're talking about. Buyers in an opaque inventory market only learn what hotel they'll be staying at after their purchase is complete and all purchases are non changeable and non refundable. This type of uncertainty and unchangeability does not appeal to business travelers. So this market makes it possible for hotels to maintain at least two separate prices for the same room on the same night. Although exact statistics on the size of the opaque inventory market are scarce, I suspect that this market is about 2 to 4% of hotel room sales in the United States and Canada. And it may not be growing. Fancy hotels do not want to publicize that they are renting some rooms at 30% to 60% below their usual rates for identical rooms on the same night. Also, opaque pricing deals tend to be barely profitable if at all for hotels. Prices tend to be near what's called the floor rate, which is an accounting calculation of the hotel's break even per room, based on allocating fixed costs per room plus variable costs to one room. Instead of opaque pricing, hotels can and do use various models to sell rooms at prices that fall between the rack rate and the floor rate. For example, when a hotel realizes that some rooms may go unsold they, can offer discounts off of the rack rate to a mailing list or a mobile text list of customers who have shown some loyalty to that hotel and that brand in the past. People who have some brand connection or brand loyalty will be willing to pay more than the opaque inventory floor price. Question. When should a hotel offer a room at 80% of its rack price to a loyal customer? Answer, where there is a less than 80% chance that the room will be rented at the list price before it expires. In conclusion, a hotel room typically has at least three potential prices for each night. A rack, or listed rate, the floor rate and an intermediate promotional rate the hotel can offer on a last minute basis to loyal customers. Only the floor rate is constant. The listed and promotional rates vary by day of the week and by season of the year. So in theory, a hotel could rent the same level of room in one year for over 1,000 different prices.