SHA532 Transcripts Transcript: Forecasting Accuracy Forecasting is probably the most important thing that goes into a revenue management system in particular, an accurate forecast. Just think what happens if you over- forecast, as in you say too many people are going to be coming to the hotel. Well, if you think too many people are coming to the hotel, you might go ahead and restrict your rates and put very strict length- of- stay controls on and end up turning people away. Or the other thing is if you are underestimating how many people will show up, you could end up having to walk people, and be in a really bad overbooking situation. An accurate forecast is good not only for revenue management it's also good for labor scheduling. Think about what it's like at the hotel when the forecast has been too high. There are too many people working at the front desk and too many housekeepers. It's obviously boring for the employees, but it's also rather expensive for the hotel. And then of course the other side too is, if you under- forecast, you end up with people running around like their heads have been cut off, and there really aren't enough people there to help out. It's also very useful for purchasing, in terms of figuring out, for example, how much food to buy, how many sheets to have washed, all those different sorts of things. A forecast is good not just for revenue management but also other things in the hotel. Transcript: Meet The Booking Curve Booking curves are very, very useful for revenue management, because they show the rate at which reservations come in. Let's think about it. If you are at a resort hotel, when do your reservations come in? Usually months and months ahead of time, while, if you are at a business hotel, they might come in two or three days ahead of time, and if you're at an airport hotel, maybe they just come in the same day. If we have information about when reservations are made, it makes it a lot easier for us to make a decision about which people to say yes to and which ones to say no to, and it also helps us with our forecasting. 1
When we look at booking curves, there are two terms that we look at. One is ROH, which stands for reservations on hand, so, basically, rooms on the books. The other one is DBA, or days before arrival, so how far ahead of time the reservation has come in. In some situations, for example, with a resort hotel, you might even be looking at weeks before arrival, or WBA, or even months before arrival, while even at an airport hotel you might be looking at hours before arrival or HBA. But let's just get a real simple example of this. Let's say you're at a business hotel and you find that on average in the last week you pick up 50 reservations. Let's say that you're a week ahead of time and right now you happen to have 100 reservations on the book. The forecast that we'd be able to build from this booking curve is right now we have 100 reservations on hand plus an expected pickup of 50, so the forecast would be 150. We'll talk about this a little bit more in a little bit. The other thing with this that you have to think about is that the booking pace for every day is a little bit different. So if I look at this Tuesday, the booking pace for this Tuesday might be different than last Tuesday and maybe the Tuesday before. So usually what we'll do with this is we'll come up with an average booking curve, maybe an average of the last six or the last eight Tuesdays or last eight Thursdays, because you always want to be breaking these up by days of week. If you want to, you can also break these up by rate category. You can break them up by market segment it really doesn't matter. What matters is that you're able to put these into a form that's going to help you make better decisions. Transcript: Groups, Channels, And Segments We've talked a lot about booking curves and pickup forecasts, but, as I've mentioned before, you could use these for forecasting all sorts of different things in your hotel. So let's say for example that you have a market segment that you're particularly interested in let's say it happens to be corporate meetings. You could use this booking curve to show how corporate- meeting groups happen to come in, how their reservations pick up. You could use this information then to come up with a forecast for them for the future. Or, similarly, with a distribution channel. Let's say you are trying to figure out how many rooms that you might sell, for example, through Expedia. You can go through and build a booking curve for this and see how many reservations do I normally pick up during the last three days before arrival, or the last two weeks before arrival? Once you have that information you can 2
take that, add it to the reservations you currently have on hand, and come up with your forecast. Again, reservations on hand plus expected pickup is going to equal our forecast. We could also use pickup forecasting for forecasting for groups or for tour operators. Let's say you have a very large tour operator from northern Europe who brings you a lot of business. You're trying to come up with a better forecast for that tour operator. What you can do is you can look to see again on average how many reservations does this tour operator provide me during the last five weeks, during the last eight weeks. Basically you are coming in with the expected pickup. You could use that information in conjunction with how many reservations the tour operator has already given you add that together. So basically: reservations on hand plus expected pickup and that's going to give us the forecast. We can do the same thing for groups. Let's say that we happen to have large national associations coming in. They've been to our hotel before, or we have some history on them. We can look again to see how many reservations do they normally pick up during the last two months, during the last one month? We can use that information to help develop better forecasts. And also decide perhaps if we might want to go ahead and either give them more rooms or restrict the number of rooms that they might be wanting to use. Transcript: Errors In Forecasting When we talk about forecast error, you can either be too high or too low. We talked about this a little bit earlier, but let's go back to it. If your forecast is too high, that means that you're expecting to have more people to come to your hotel than you are normally expecting. And so what this typically will mean is you would close off your lower- rate categories. You might have very, very strict length- of- stay controls. And essentially you end up turning people away. Now, if your forecast was too high, this means you might end up with an empty hotel when you had people who wanted to come to it. Obviously not a good situation. The other thing can happen too is when your forecast is too low. When your forecast is too low usually you open everything up. Open all your rates, no length- of- stay controls in an attempt to try to build demand. And all of a sudden you have all these extra people coming in, plus the people who you were expecting anyway, and you might end up in an oversold situation. You also might end up diluting your rate. And so, clearly what we want to do is try to be as accurate as possible. 3
Sometimes people will say, "Well, my forecast is absolutely perfect. I'm always within 1% of accurate." When someone says that to you it's probably a sign that they might not be measuring their forecast accuracy properly. What you really want to be doing is looking at how far off are you on average whether you are off positively or negatively. Basically, you want to take the absolute value, which means changing all the negative numbers into positive numbers. Let me give you an example of how that could work in the wrong way. Let's say that you have a hotel that sells 100 rooms every day. The first day the forecast had been 90 rooms, the second day 110 rooms, the third day 90, the fourth day 110, and if we go through and calculate the error, well, some days they're off by minus 10 and some days they're off by plus 10. If we add all those up it comes out to be zero. If we average that, it's still zero. And it means that their forecast is absolutely accurate. Obviously this isn't true. They were off on average by 10 that was their absolute error. What the zero error does tell us though is that they didn't have any bias in their forecast. It means that they were off too high as often as they were off too low. Not really good, but still, they didn't have a whole lot of bias. But we also always want to make sure that we do use the absolute value. When we measure forecast error, we usually have two different methods that we can look at. One that measures how far off you are, on average, and that one is referred to as the mean absolute deviation. Mean is the same as average, absolute means taking that absolute value again, and deviation is how far off. So this will show you: am I off by two rooms? am I off by ten rooms? That can be a very useful measure for a hotel. The other thing we'll look at is how far off you are on a percentage basis. That one is referred to as the mean absolute percentage error, or the MAPE. So, am I off by 2% or 10% or 20%? Now, if you have several hotels that you have responsibility for, the MAPE is probably better, because that allows you to be able to compare the hotels. So let's say one of the hotels that you manage has 100 rooms; the other one has 1000 rooms. Obviously, if you have an error of 10 rooms for the first one, that's a lot worse error than if you have an error of 10 rooms in the 1000- room hotel. The percentage error can help correct for that. The other thing that you want to be looking at is when are you measuring this error? If I am looking at my forecast error three months ahead of time, it's probably going to be a lot higher than what my forecast is on the day before or let's hope it's a whole lot higher. Typically, the closer you get into the day of arrival, the more accurate your forecast is going to be, because you have additional information. But again, the key point with this, whether you choose to use the MAD (mean absolute deviation) or the MAPE (the mean absolute percentage error) is that 4
you always use the absolute value so you're not canceling out your negative numbers with your positive numbers. Transcript: Demand Control Charts The demand- control chart provides a map of the hot, warm, and cold zones of the reservations- on- hand or ROH landscape. In the hot/warm/cold approach, hot is when we have a really high RevPAR or a high revenue per available time- based inventory unit. Warm is when we have a medium RevPAR or medium revenue per available time- unit, or cold is where we have a low RevPAR or low RevPATI. If we take a look at the demand- control chart, we can definitely see how it looks like a booking curve. On the x- axis, or on the bottom, we have days before arrival, and on the y- axis, or on the side, we have reservations on hand. We can see that "days before arrival" goes from day minus one and you probably remember that from the booking curve, that's the day after what actually happened and it goes all the way out to seventy days before arrival. And then the y- axis goes up to 500 rooms. Sometimes you'll see a demand- control chart referred to as a threshold curve, and basically what that is is that it's an average booking curve that we know that sometimes we're a little bit above average, some days we're a little bit below average. Hence, the threshold. When we cross the threshold, we're going to be opening or closing different rate categories. If we take a look at this, let's assume that we're at maybe fourteen days before arrival and we happen to have 200 reservations on hand. If we take a look at this, this is well above what we normally would expect and would be considered to be hot. A hot period means that we're expecting to be very busy. There probably aren't going to be any discounts available on that day. Let's take a look at another day that maybe is four days into the future. So "days before arrival" is equal to four. And we also have 200 reservations on hand for that day. Hmm. That's kind of low. That's normally...normally at that time we might expect to have three hundred or so. If we only have two hundred, that day is going to be considered to be cold. We're not expecting a really high demand on that day when it actually happens. And so, in that case, we're going to have some discounts available. And in the middle...so let's say I happen to be warm. So let's say that I was at about four days before arrival, and I had about three hundred reservations on hand. That's kind of average. It's not super busy, so it's not hot, and it's not that bad it's not cold. And so, what this is saying is that I might have one or two discount rates available besides my rack rate, just so I can get some customers in. But I am expecting to have kind of an average day. 5
So, why would I bother creating one of these? A demand- control chart can help us figure out when we're going to change our rate based on our demand. When our forecast is above a certain level, we're going to close room rates. So, for example, when we're saying a day is going to be hot, we're going to close all the discounted rates and probably just have rack rate available. But when the forecast is below a certain level, we're going to open room rates up. Now, when we look at revenue management, when we look at demand- control charts, demand- control charts are probably one of the easier things that you can do. And for that reason, that's great. I mean, one of the main advantages of the demand- control chart is it's very easy to use and understand. And when it's easy to use and understand, it's also pretty easy to implement. But of course with all good things, there are some bad things. One thing that doesn't happen with a demand- control chart is it doesn't consider length of stay. But this isn't so much of an issue right now because many hotels are going with best- available- rate strategies and might actually charge different rates based on what night people are staying. One of the other disadvantages is that it lumps all demand together. It's putting all the different market segments together and really there might be different demand patterns based on the different market segments. But still this is a technique that has been very successful for a number of hotels and has led to revenue increases of two to five percent. So, it's got good points; it has some weaknesses as well. But overall, it's a good strategy. One of the other nice things about demand- control charts is you can also apply these to other parts of the hotel. First thing you have to do is identify the hot and cold periods. Then, once you've done that, come up with your strategies. During the hot periods, you're going to be looking at increasing the value at maximum price, which means trying to get as many customers through and not having any discounts available. And during the cold periods, you're going to be trying to get more customers through, so whether through discounts or promotions, whatever this might be. You could very easily do this for a golf course, for a restaurant, for a spa. The whole thing here again is trying to identify what your hot and cold periods are and come up with your strategies accordingly. Transcript: Rate Recommendations Let's say we've got a hotel that has 250 rooms. They've calculated their forecast and they're trying to figure out what their rates should be and obviously when they're busy they don't 6
want to have too many discounts open. But when they're slow, they need the business, so they might want to open up some discounts. But the key is to figure out when we want to have these discounts available, what rates to be offering. In this case, they've classified everything over 100% as hot (notice the red), and they've said that everything under 80% is cold (notice the blue). Everything else is warm. So we take a look at this, for example on the 8th of June they're forecasting that they'll sell 214 rooms, or an occupancy of 85.6%. That would be classified as warm. But on the 11th of June, they're forecasting not so busy a day 170 rooms or 68% occupancy. So they're forecasting that they're going to have a cold day. But then on the 15th of June, or the 21st of June, they're forecasting (and note this is a forecast, not how many rooms they've actually sold), they're forecasting that they're going to be well over 100%. They're forecasting a hot day. And on these days you want to be sure that you're not offering discounts because you know that you're going to be able to fill the hotel. Now once we take a look at this, the next thing is, well, how can we use this kind of information to come up with different rates that should be offered? So we're going to keep with the same idea of cold, warm, and hot. But now we're going to say that this hotel, just to keep it simple, has three different rate categories available: $250 a night, $325, and $425. And so now what we have to figure out is when we should charge those different rates. Well, one of the concepts that we talk about with demand- control charts is something called a trigger point. Basically the trigger point signals the opening or closing of a rate class. And in a way we've already used a trigger point when we said if it's over 100%, we're going to classify that as hot, or if it's under 80%, we're going to classify that as cold. All we're going to do now is attach some dollar figures with those hot and cold definitions. So basically, if demand is above a trigger point, we're going to close appropriate rate categories. In this case, we're going to use 100% and 80%. If demand is below a trigger point (or 80%), we're going to open appropriate rate categories. And we can have multiple trigger points. Right now, we're just looking at 80% and 100%. But we could easily have six, ten, whatever you might want to have. So in this case, let's say that our trigger points are 0%, 80%, and 100%. Basically, what we're saying is that if our forecast is under 80% it's going to be cold (right?) and we're going to say the minimum rate is $250. That's not to say that we wouldn't sell rooms at a higher rate (at $325 or $425), but what it is saying is that we have a minimum rate available of $250. If it's warm, as in it's between 80% and 100%, we're going to be saying that the minimum rate available is $325. Again, we're quite willing to sell our rooms at rack rate, at $425, but if someone meets the qualifications, they might be qualified for this discounted rate of $325. 7
And finally, when it's over 100% or classified as hot, our minimum rate is going to be $425. "Do you have any discounts available?" "I'm sorry no, we don't have any discounts available." So it's kind of a three- step process. First thing we're going to do is come up with a forecast. Then we're going to come up with the trigger points, and then we're going to come up with the minimum rates to quote. Now when we came up with our rate recommendations, we made things pretty simple so far. We've said that the hotels just have three rates, but obviously hotels have a lot more rates than that. If you look at your hotel you probably have fifty, one hundred, two hundred different rates. So when you come up with rate recommendations, what you want to be doing with this is kind of putting rates in rate categories or rate buckets. And so, for example, we might look at all rates that are between $100 and $119, or all rates that are between $120 and $139. Generally it's a good idea to not have any more than about four to six rate buckets or categories. But you also want to make sure that you have more than one rate per bucket. So for example, if you're saying the rate bucket is everything between $100 and $119, you want to have more than just one rate in that category. You might want to have $102, $105, $107, $109.50, a variety of different rates within there so there is some activity within that bucket. Another way you can look at the way you set up your rate buckets is to split it into different levels, based on the amount of discount. So, for example, level one might be your rack rate, where you have no discount. Level two might be your 10 to 20 percent discounts. Level three, 20 to 35. Level four, 35 to 50 percent. And level five, 50 percent or more. Then when you did your demand- control chart, where they're coming up with, for example, the $150 rate is the lowest rate, or the $120 rate, you could come out with the levels. The lowest level available is level one, the lowest level available is level three. That allows you to use the complexity of the rate structure that you probably have in place. And very simple to understand, but again, you probably don't want to have more than four to six of these rate categories, because otherwise it gets a little too confusing for the reservation agents. Transcript: Controlling Length Of Stay One of the things that you want to do with revenue management is to try to control the length of stay. So for example, some people might want to stay for one night, some people might want to stay for four nights it's all a matter of trying to pick out which of those lengths of stay are going to be the most valuable to you as you try to maximize the hotel's revenue. 8
Let's talk about the different sorts of length- of- stay controls. One thing that you might be using during a busy period is a minimum length of stay. Let's say that you have four busy nights and that you're going to have some slow periods. You're trying to decide which reservations to accept at the beginning of those four busy nights. If you have people that are willing to stay four nights, you're going to be a lot more open to having them at your hotel than maybe someone who's only willing to stay one or even two nights. And you might decide that you want to put in a minimum- length- of- four- night or minimum- length- of- three- night availability control to try to pick the right people. Now, where you've got to be very careful with this is: what if you don't have anybody who wants to stay for four nights or who wants to stay for three nights? You can't exactly lock them in their room. And so this goes back to what we were talking about earlier, with the arrivals- based forecasts. In order to be able to use these length- of- stay controls, you've got to make sure you have sufficient demand to be able to use them. Otherwise, people are going to end up just leaving early. Now there's some very busy periods so for example at a resort hotel over a winter holiday or at a major national festival where you might be able to get away with a length- of- stay control because everybody wants to stay at your hotel. The key thing here is whenever you use a length- of- stay control, you're going to be using it during a high- demand period. Some of the other things that you might be looking at might be a maximum length of stay, and these might typically be put on a lower- rate category. So again, let's say that we've got some very busy days coming up and we're expecting to be sold out at rack rate. And let's say that we've had someone staying at the hotel at a lower rate, a discounted rate. We might say, "We'll give you that discounted rate for a maximum of two nights, but we would like you to leave at the end of those two nights because we have other people coming in who are paying a higher rate. By the way, if you would like to stay longer, you're welcome to. But the rate is going to change." Again, you've got to be very, very careful with this because what do you do if someone decides to stay longer? In most situations, by law, you can't force them out of their room. So you've got to figure out some way of either setting up where the rate has changed...you are able to get them another accommodation...something like that. Because they might choose not to leave. Another length- of- stay control is closed to arrival. You have to be very, very careful with this because "closed to arrival" means that you're shutting off all reservations coming in for that day. That's not only going to affect today, but it might affect tomorrow, the next day, the day after. Basically, you're allowing people to stay through from previous nights. Just nobody is coming in on a particular day. 9
And with any of these length- of- stay controls, you've got to be very, very careful with them because you don't want to end up turning away demand when it finally ends up you ended up with empty rooms. You want to be only using these when you're in extremely high demand. Transcript: Thank You and Farewell Hi. This is Sherri Kimes again. I hope you've learned all sorts of things about forecasting. You know all about unconstrained demand and demand- control charts and arrivals forecasting and length- of- stay controls, and I'm looking forward to hearing how you're using these at your hotel to help make more money. Thank you, and I hope to talk with you in another one of the courses. 10