SHA532 Transcripts Transcript: Course Welcome Hello from Ithaca, New York. This is Sherry Kimes. And in this course, we're going to be talking about forecasting. Forecasting is the building block of revenue management. And we have kind of a saying or initials in the US, it's called G-I-G-O. It stands for Garbage In is Garbage Out, and so if you have a forecast that is garbage, or bad, you're going to end up with very poor revenue management decisions. So it's very important that we develop an accurate forecast, that we understand our demand, and we're able to forecast accurately into the future. Because otherwise our revenue management system isn't going to meet its potential. So let's go ahead and get started. Transcript: Why Forecast? And you might ask the question, well why forecast? Well, you kind of have to do this. I mean how many customers are you going to have at your hotel two weeks from tomorrow, or three weeks from tomorrow, or two months? And we try to answer these questions and a lot of others through forecasting because a forecast not only works for revenue management, but also for a number of other things at the hotel like purchasing and labor scheduling, but we will get into that in a moment. So when we talk about forecasting we are using historical booking data and other information to try to make an informed estimate on how many people are going to be arriving at our hotel on a given date. Notice that I use the term 'historical booking data.' One of the things that you need to realize with forecasting is we're assuming that the past is a good predictor of the future. If things change drastically, so whether it's maybe it's a natural disaster or something economic that's really changing demand, your forecast is probably not going to be all so accurate. You're going to have to look more at what's been happening in the near past. But we'll talk a little bit more about that later on. But we are assuming the past is a good predictor of the future. And they're important for a lot of reasons. So we've got setting up room rates, implementing a length of stay control. So like minimum length of stay, or maximum length of stay, scheduling our staff, purchasing. So let's just kind of look at what happens when there's no forecast or if the forecast isn't so accurate. And so with revenue management, okay, if we over forecast, we might end up turning people away because we have very strict controls, and we don't have any discounts available and everything. And if we underestimate how many people are 1
going show up, we could end up having to walk people, as we would be in a really bad over booking situation. Neither of these are such a positive environment to be in. With labor scheduling, and I'm sure you've all been in situations where it's been off, forecast is too high, there's too many people working at the front desk, too many housekeepers, and it's really kind of boring. Plus, if you think about it from a cost perspective, it's pretty expensive. And if you under forecast, there aren't enough people there to check out, to help out, and so we're at the front desk, and there's a big long line of people wanting to check in, and we're the only person working there, or there aren't enough housekeepers, and all these sorts of things cause poor customer service but also, it's really stressful for the employees. And then with purchasing, we kind of need to know what the forecast is going to be so we can figure out how much food to buy, how many sheets have to be washed, you know, all of these different, different sorts of things. And so forecast is good, not just for revenue management, but also for managing the hotel. So, in this course we're going to be looking at how an accurate forecast will enable us to use revenue management effectively, but realize that you could use this forecast not only for revenue management, but also for other parts of the hotel. And in the next few videos, we'll look closely at demand, what it consists of and how you can estimate it. Transcript: Measuring Demand We're going to be talking a little bit about how do you actually go about measuring your demand? And there's a difference between what your demand was and what actually happened. So there's a historical performance, but you might have had actually higher demand for the hotel than what you actually see. We'll talk about that in a minute. So we're going to be looking at what's happened in the past to estimate demand in the future. And again remember that with forecasting we're assuming that the past is a good predictor of the future. So let's just take a look at some of the issues involved with documenting and estimating our demand. So if we have a lot of bookings it indicates the demand was high. Having a few bookings says, well, demand was kind of low. But these are just kind of rough estimates because we just have what happened. It doesn't tell the whole story. Because there might have been some people who tried to book our hotel who couldn't get in. We might have had availability controls, all these different sorts of things and so we want to kind of be looking at that as well because it can give us an estimate of what our true level of demand was. And so you'll think about this. Have you ever tried to go to a restaurant, and you've seen a big long line outside, and you've said, 'You know what, there's no way I wanna go'? Or have you called to make a reservation and they've said, 'Oh, we don't have any availability'? Similar sorts of things 2
happen for hotels. So, sometimes people try to book a hotel and maybe there aren't any rooms available, or maybe the rates that they want aren't available. So we might just be looking at, when we go through and look at the number of sales we have, we might actually be looking at only a fraction of the demand that we have. And so we got some potential customers as well. So if we look between the customers who actually booked, who bought a room. And those who were denied that's our unconstrained demand. Let me give you an example of this here from Cornell. In the wintertime in Ithaca, New York it is not a pretty place. Weather is bad, you know we get on average a couple of meters of snow. And we have a hotel right here on campus, the Statler Hotel. And demand is so low during the winter time, also the university is closed and so forth, that they actually close for two weeks around late December early January. Now lets move fast forward to the end of May when its graduation. Our hotel has 150 rooms, it's the only hotel on campus. And we have about 4,000 graduates, and all of their parents come, extended family, and we have a lot of people. There's probably 30,000 people in our tiny little town, and they all want to stay at the Statler Hotel. And so in the wintertime, they could shrink their hotel down to nothing because there's hardly any demand. But during the graduation weekend, if they could expand their hotel, if they could inflate it, they could probably fit easily 20,000 people. And so this unconstrained demand is how big the hotel would have to be to accommodate the potential demand. Now, of course, and the question is, how do you actually measure this sort of thing? Well, you can start looking at the people who were denied, people who tried to book and so forth. But we want to be looking at this very, very carefully, because we want to look at the denials, customers who we turned down, not customers who turned us down. So maybe they were shopping for a particular rate or something. But this is important to know, because if we have unconstrained demand, it means we have the potential to be having more rate controls, to have higher rate, but then at the same time, it might, if we have a lot of, we always have excess demand, it also might be a sign that there's a possibility for expansion. But just kind of keep this in the back of your head, because just when you look at historical information, that's what happened. It's possible that there could have been additional customers. And if we look at the combination, people who booked and those who wanted to book. That's our unconstrained demand. Transcript: Ask the Expert: Neal Fegan on Demand Forecasting How do you approach forecasting at Fairmont? The way that we forecast at FRHI is we look at it a few different ways. There's different types of forecasts. The one that we focus on within revenue management is the demand forecast. That's the one that's going to help drive incremental revenue, so when you're taking a look at where those peak night's going to fall, how do I optimize that? How do I figure out how to boost up the shoulder-night demand around those 3
peaks? We use revenue management software systems in order to help us understand what's going to happen into the future and segmenting that out so that we can drive the most optimal mix of business through both the peaks and the off-peak periods. Transcript: Meet the Booking Curve Then we're going to talk about the booking curve. And the booking curve is talking about, when are customers actually making their reservation? And so, it's going to vary depending on whether it's an airport hotel or a downtown hotel, and so there's a couple ways we can display this information. One is we can have it in a table, and so you can see the different number of reservations on hand as you get closer to the day of arrival. Or you can have a booking curve, where you graph it. So we've got on the bottom, on the x-axis, the number of days before arrival, and then on the y-axis, we have the number of days the reservations on hand. So whichever way you do it, but it's going to vary, the shape of this is going to vary depending on the type of hotel. So think about your hotel. When do customers usually make their reservation? Is it a couple of days ahead of time? Is it a week ahead of time? If you're an airport hotel it's probably the same day, and there's some terminology that we'll be using, I will be talking about reservations on hand, so we'll abbreviate that as ROH, reservations on hand. And we're also going to be talking about the time before arrival. So usually we'll talk about DBA, so 'days before arrival,' but if you're at a resort hotel where people make reservations very far ahead of time, maybe you look at weeks before arrival, or at an airport hotel you might be looking at hours before arrival, so HBA. And when we start looking at the booking curve, take a look at it, DBA = -1. What in the world is that? Well day before arrival zero means that's today. You came to work this morning and you looked to see how many reservations were on the books. Now when you come in tomorrow, and to see how did I actually do that day, well it's going to be a little bit different right? You're going to probably have some walk-ins, you'll have some no shows, so the DBA minus one, this is the day after the arrival date, so it's representing the final number of reservations, or rooms that we sold for that particular day. So day zero is today. Day minus one is the day after. And the booking curves, again, they show when reservations come in. So, at a resort hotel, you're going to see the bookings come in very, very early and probably not very much happening in the last little bit. At a business hotel, you'll usually see not too much until maybe about a week before arrival, and then it starts going up. And an airport hotel, flat, flat, flat, flat, flat, and all of a sudden, on the day of, it starts to pick up. And you have to understand that booking pace because, let me give you an example, one of my former students was a front office manager at a business hotel, and so most of her bookings came in about a week before arrival. And she got transferred to an airport hotel, and there it was, the morning of arrival, and she hardly had any reservations on the books. She said, I was totally, really scared, what was going to 4
happen? And after a week or two, say wait a minute, I've got a different booking pace here. I can't rely on my past experience. I've got to change it to this hotel, because it's a little bit different here. And then when we start looking at this again, the booking pace for every day is going to be a little bit different. And so, first of all, I'm going to have my booking pace for Mondays, for Tuesdays, for Wednesdays. So I want to break it down by day of week or at least by weekday, weekend. But let's say I've got Tuesday this week, booking pace might be slightly different than the Tuesday the week before. So what we'll do is, we'll do an average of the Tuesdays or an average of the Wednesdays, and so forth. Because we don't want it to be just for one specific day. And then we could also break these up by rate category, or by market segment, but it gives you an idea of how the data is coming in. One of the ways I like to think about it, it's kind of like a glass of water. How quickly is that water getting poured into it? With an airport hotel the glass is empty for a long time and the water gets poured in very, very quickly at the end whereas with the resort hotel it fills up early. And then it might evaporate a little bit with a few no-shows. But the booking curve, again, is the pace at which your reservations come in. And it gives you a very good estimate that you're going to be able to use to help develop your accurate forecasts. Transcript: Introduction to Pickup Forecasting Let's talk a little bit about forecasting. How do you go about doing your forecast. There's a lot of different things that you look at. You look at what the competition is doing, you look at the market, you also look at how your hotel has worked in the past, how you've done historically. And so the method that we're going to be talking about is the pickup forecasting method. And one of my co-authors and I have done a lot of research on this and we found that the pickup forecasting method, it's one of the most accurate methods out there, and it's really simple. Basically what you look at is you say okay, how many reservations do I have on hand, and then you look to see how many do I normally pick up until sometime before the day of arrival. And then I add those two together and that's my forecast. That's all there is to it. So let's say that I have 20 reservations on hand and it happens to be 21 days before arrival. Normally in the last three weeks I pick up 50 reservations, so my forecast would be 20 plus 50 is 70. Or let's say that I'm three days before arrival and I happen to have a hundred reservations on hand. Normally, in the last three days, I pick up 30 reservations so my forecast would 100 plus 30 is 130. So all you do here again is you look at how many reservations do you have on hand? And you look, how many reservations do I expect to pick up by the day of arrival? You add those two together, and that's your forecast. And in a later video, we'll talk about how to do this in Excel. 5
Transcript: Creating A Pickup Forecast in Excel We've been talking about the pickup method where we take the number of reservations on hand, then we add the expected pickup, and that gives us our forecast. It's a really easy and effective method, but it's even easier when you try it in Excel. So let's give that a shot. If you take a look at how I've got this set up. I've got data for five different Fridays, so the 16th of May, the 23rd of May. And notice that this is just for one day of the week and it's important that when you're doing your forecasting, that you split your data by day of the week. And across the top I've got the different reading days or the number of different days before arrival. So day minus one is the day after, that's when you go into work the next day to say how many rooms you actually sold that night. 7 is 7 days before arrival, 14 days is 14 days before arrival, and then 21 and 28 days before arrival. And I didn't include all the different number of days before arrival just to kind of keep our example pretty simple and so we didn't use up the whole spreadsheet with a lot of numbers. If you look here, okay the 16th of May, I ended up selling 100 rooms, but a week before arrival, I had 60 reservations on hand, at 14 days before arrival, I had 40, and then 30, and then 20 reservations on hand at 28 days before arrival. Those five numbers there, are the booking curve for the 16th of May. So you can see that they started off as kind of slow. And then they started to pick up quite a bit in the last week. But then, take a look at another day. Let's take a look at the 13th of June. The 13th of June, they ended up selling 120 rooms, as opposed to, on the 16th, it was only 100. And a week before arrival, they had 80 reservations on hand, as opposed to 60 on the 16th of May. And then two weeks before arrival they had 50, and then 30 and they only had 15 reservations on hand at 28 days before arrival. And again those five numbers are the booking curve for the 13th of June. And if you look here, each of these Fridays has slightly different booking patterns. And so what we want to do, is we want to be coming up with the average of these when we come up with our pick up forecasting. And so first thing we're going to do is just average them. And so we average these five and it looks like it's about 110. Okay, 110, if I drag these across, you can see on average we had 70 reservations on hand at 7 days before arrival, 42 at 14 days before arrival, then 28, then 16. These five numbers, those are the booking curve for Friday. So how are we going to use this to come up with our pickup? Well take a look at this. They ended up selling 110 rooms and at 7 days before arrival, they had 70. So this looks like they picked up 40 reservations during the last week. So 110 minus 70. Well how many reservations did they pick up from two weeks out? Well, let's see 42 is how many they had on hand, and they ended up selling 110, so it looks like they picked up 6
68. Well we could do this in our head, but let's do it on Excel, it's faster and probably more accurate. And so what I'm going to do is I'm going to be subtracting each of these numbers, the 7 days before arrival reservations on hand, the 14 days before arrival, 21, and 28, I'm going to be subtracting that from how many rooms I actually sold. So I'm going to be subtracting that from 110. So I'm going to say = 110. Now this is cell B8. I'm in Excel, and Excel only does what you tell it to do, and I always want it to go back to cell B8, and so we've got to make that an absolute reference, and so if you're on a Mac, you say Command-T, if you're on a Windows-based machine, it's F4. And what that does is it always makes sure that Excel goes back to a particular cell, it's called an absolute reference. So for this one, I'm going to say B8 minus 70 And so that is 40, you could see 110 minus 70. If I just drag this across now, there's 68. What that is, is that's 110 minus 42, and the 82 is 110 minus 28 and so forth. And this is on average how many reservations do we pick up during the last week, the last two weeks, and so forth. And so let's just take some examples, let's say that I'm looking at a day that's 7 days from now, its a Friday. And I happen to have 55 reservations on hand. Well, its 7 days before arrival. How much do I normally pick up? Well it looks like I normally pick up 40. To come up with my forecast, I'm just going to add those two numbers together. 55 + 40, it's going to be 95. Or let's say that I've got a day that happens to be 21 days before arrival and right now I happen to have 25 reservation on hand. Well how much am I expecting to pick up? Well it's 21 days before arrival, so I would normally pick up 82 I'm just going to add those two numbers together, so its going to be equal to 25 + 82, and then let's just do one more example. Let's say it's 28 days before arrival, and I happened to have 25 reservations on hand, I normally pick up 94, you can see over here, ok? And to do my forecast, I'm just going to be adding my reservations on hand plus my pick up, so it's going to be 119. So again what we have here, if you look at this, I've got my different number of days before arrival, I've got booking data for my different days, and so usually the research we've done on this is if you have maybe six to eight different days, that works the best, I'm going to average those and then to come up with a pickup, I'm just going to be subtracting the average reservations on hand from the average number sold, and we did that with an absolute reference. And then to do the forecast, all we're going to be doing is well how many reservations do we have on hand for a particular day What was our expected pickup, we got that from this table and then we just added those together to come up with the forecast. And as we've talked about the pickup forecasting method is a very effective and accurate method, and one that's going to work very well for you and your hotel. And in a later video we'll talk about how you can actually measure the accuracy of your forecasts. 7
Transcript: Ask the Expert: Alfonso Delgado on Types of Forecasting How does forecasting vary by industry? We've said earlier that, when it comes to the inventory level, it's very important to understand, what is it that we're trying to sell? What is it that you need of inventory that we're trying to sell to the customer? It's also very important to determine what is the unit of forecasting? Are we forecasting passengers, are we forecasting rooms, are we forecasting occupancy? So first identifying ways that we are trying to forecast. What is it that you need to forecast that eventually is going to become the input into that revenue management algorithm and optimization process. Once we've done that, we need to understand how much history are we going to rely on to derive that forecasting. There are industries, and there are environments where the more history the better, because it's a very stable environment with very predictable seasonality. While there are other industries where the forecasting method has to be much more reactive to the changing environment, where the competition is doing etc. Once we've considered those two elements, then we have to determine how we're going to forecast total demand volume, or are you going to forecast distribution of demand, so for instance, do I need to know that I expect to receive x number of bookings by a certain date, in a certain hotel, or do I want to understand that by a certain date, prior to departure I will receive x percent of my total demand. So in one case we're trying to understand number of passengers, in the other case we're trying to understand what percent of the total demand I'm going to receive has already arrived into that hotel, that flight, that train. That is going to be critical because eventually that is going to determine whether we're trying to maximize the price in an environment where I'm trying to reach 100% utilization. So for instance, if I'm talking about a train, or I'm talking about a hotel or a cruise line, the incremental cost of servicing those customers is virtually zero. And therefore my objective is going to be maximize utilization. I'm going to be shooting for 100% load factor, 100% occupancy, etc. In those cases, what I'm trying to understand is what percent of the demand that I'm likely to receive has already arrived and how much is left? And based on how much demand is left and how much inventory do I have left, I'm going to modify my pricing to try to capture as much of that residual demand that I have available. In other cases where 100% utilization is not likely or is not possible, then I'm really going to look at what volumes of bookings do I have left between now and departure. And based on that I'm going to determine what price do I want to establish in order to maximize our revenue. So whether the endpoint is 100% of occupancy or utilization or the endpoint is a number of bookings or of passengers that is below that capacity, is one of the critical elements to determine what type of demand forecasting you want to use. Once you've determined 8
that, then you typically have a short list of demand forecasting methodologies that you're going to evaluate. So whether it's clustering, whether it's exponential smoothing, as simple as a moving average, hot winters, linear programming, the type of environment is going to give you a short list of forecasting methodologies that you can apply. And then based on the error rates that you observe, for each of those methodologies, eventually you settle in the methodology that actually use the lowest error. So it's a very step-by-step approach that first, understand what you're trying to forecast, that gives you a short list of methods that are most suitable for that environment and eventually you try a few of those and you settle for the one that gives you the most accurate result. Transcript: Forecasting Groups, Channels, and Segments We could also forecast for groups and segments and different distribution channels by using booking curves. And so you could break it up, so you could have maybe, let's say we're looking at market segment for our groups, so maybe a booking curve for corporate meetings. or maybe we're looking at a booking curve for weddings, or for other social events. Also, what about your different distribution channels? Like Expedia, or Booking.com or Agoda, what's the pace that those reservations are coming in? What about with groups? If I book a large group, when are their reservations coming in, and also, what about my tour operators? and this is for those of you who are at resort hotels, I'm sure you understand what I'm talking about that with that, you have a lot of different tour operators. You need to know when their reservations are coming in so you can make better decisions on the other rooms that you're going to sell. But basically these booking curves help us develop better forecasts so that we can make better decisions. Transcript: Errors in Forecasting Why is forecast accuracy so important? Well, in the airline industry, during a high demand period they found that a 1% increase in forecast accuracy, so going from 10 to 9%, leads to about a 2, 2 and a half percent increase in revenue. So it's pretty substantial. We've already talked about if your forecast is too high essentially you end up turning people away, and if your forecast is too low you're going to have all these extra people coming in, plus the guests you were expecting anyway. And sometimes you run across people who say my forecasting's so good I'm absolutely perfect, or I'm within maybe 1% of accurate. But I could pretty much guarantee you they're probably not measuring their forecast accuracy properly. So let's take an example of a hotel that sells 100 rooms every day. The first day they sell 90 and the second day, I mean the forecast is 90, second day it's 110, third day it's 90, fourth day is 110. So some days they're off by minus 10, some days they're off by 9
plus 10 and if we add all these up it comes up to zero. We average them that's still zero. Does that mean that the forecast is absolutely accurate? Well, no. I mean we can see that they're off by, on average, 10 rooms. And so what, the way we get around this is we start looking at the deviation as an absolute value. For example the absolute value of minus ten is ten. The absolute value of ten is also ten. So basically, absolute value changes negative numbers into positive numbers. And so when you look at this, we want to know what that forecast error was, and so in this situation we were off by ten rooms. Now, is that a good error, or is that a bad error? Well, it depends on the size of the hotel, and how far ahead of time we were looking at, and all that sort of thing. But it gives us an absolute, it gives us a better estimate of how accurate our forecast is. And we've established that an accurate forecast is critical. Now, the next question is, how do we actually go about measuring our forecast error. We'll be looking at that next. Transcript: Calculating Error Using MAD and MAPE We've talked about forecast error and how having an accurate forecast is critically important. And in fact an accurate forecast, especially during a high demand situation, can lead to revenue increases. So let's talk about how you would actually go about measuring your forecast error. Obviously you could only do your, calculate your error for in the past, and we also talked about how we want to make sure that rather than using pluses and minuses, so positive and negative numbers, we actually want to be using absolute value, making all of our numbers positive so we can get a real idea of what our error is. So let's take a look at this in Excel, and if you look at this, I've got data for 7 different days, so the 3rd of June through the 9th of June, and these are all days from the past. You can see, here's how many I actually sold. So, I sold 104 on the 3rd of June and 110 on the 4th of June and here's what my forecast had been. So, I had forecast 110 for the 3rd of June but I only sold 104. So, I had over forecast by 6 rooms. In contrast, on the 4th of June, I had forecast 106 but I actually sold 110. So I was about four too low. And let's do this in two different ways. One is we're going to go ahead and just calculate the error. So this is going to show how high or low we were. So it's going to be 104 minus the 110, and what this shows us is that minus six. So you might think minus means that I under-forecast, but actually what it means is I over-forecast. And if I go ahead and bring these on down, you could see on the 4th of June I was under forecasting by four, and so forth. And so I've got all these pluses and minuses. If I went through and averaged these, let's see how that comes out. That says, on average, zero. Well, you know that's not true, because some days I'm way too high, some days I'm way too low. And so, what the zero means is that on average, you are not biased, you're overforecasting as often as you are under-forecasting. But we don't want to be looking at mixing together our pluses and minuses, that's why we want to use the absolute error. And the way you do this in Excel, it's really easy, which is =abs, that stands for absolute. It might vary depending on what language 10
you're in. And I'm going to be taking the absolute value of cell D3, okay. And that's going to turn it into a positive number. So see, it goes from -6 to +6. And if I bring these on down Notice that it doesn't change any of the positive numbers, like 4 still stays 4, but if a number was negative as in the case in the 6th of June, it changes that minus four to a four. Now what I want to do is I want to come up with, on average, what's my error? So I'm going to do average, and just take those seven numbers. My average is 4.57, so on average, I'm off by about four and a half rooms. Well, what's that number? What that actually is is that's your MAD. That's the mean absolute deviation. So we say, well, is that good or bad? Four and a half rooms on average, and it looks like I sell maybe 80 or 90 rooms on average. Depending on how far before arrival that is, it could be pretty good. But we really want to be looking at this in relationship to how many that you sold on average. So let's see if we can come up with this. I'm going to average these numbers here. So it's going to be equal to the average number that I sold. And so on average I sold 84 rooms. So ok, 4 and a half is my error and on average I sold 84. So that looks like I'm off by about maybe 5%. Let's see. So it's 4.57. I'm going to take that cell, E10, and I'm going to divide that by this 84, and it comes out to 5.46%. I'm going to change that into a percentage. And what that number is, that's my MAPE. That's my mean absolute percentage error. And so on average I am off by 4 and a half rooms, that means that my average percentage error is 5.5%. And both of these error methods are equally valid, but what I find with the MAPE, the mean absolute percentage error, it allows you to compare the error among properties of different sizes. So, let's say you have one hotel that's got 100 rooms and one hotel that has 20 rooms a four and a half room error makes a big difference in that 20 room hotel not as much of a difference in the hundred room hotel. That's why this MAPE percentage I could compare those and again all I do to come up with these numbers is I subtract my forecast from the number sold and I take the absolute value of that so that's ABS, I do that for all the days, I average those, the average of those numbers in this case was 4.57, that's my MAD. And then to come up with my MAPE, my mean absolute percentage error, I divide that by the average of the number that I actually sold, which was 84 in this case, and that number comes out to be 5.46%. And you could only do, again, when we're calculating our forecast error, we can only do this for days in the past. It's very important for you to be tracking your forecast error, and also be tracking it at different number of days before arrival. As you would expect, you're going to be more accurate when you're closer to the day of arrival than you might be, say, 30 days before arrival. But this is a very important indicator to see how well your forecasts are doing and if you find that your error is consistently off, you're consistently too high, go back and take a look. What am I doing wrong? Is there something that I'm missing here when I'm coming up with my forecast to make sure that you're trying to be as accurate as possible. 11
Transcript: Setting Trigger Points We've looked at how to create forecasts so now let's look at how to use that information and other sources to set rates based on demand. And so let's say we've got a forecast that kind of looks like this we've got the different days of arrival. We've got our reservations on hand, so that's the ROH. We've got the days before arrival so DBA. We've got the pickup. The forecast is just adding together our reservations on hand plus our pickup and then we've got out arrivals forecast and we've also got the occupancy percentage, and we got that for each day. So okay, so that's nice. But then we're starting to look at this. We're trying to figure out what should we be doing for charging different rates? One of the ways I've found that is really helpful to be able to come up with these different kind of trigger points at what rates should be available is to split things into is it busy? Is not so busy? Or is it really slow? And you could think about this as when it's really busy, we're going to call that hot, when it's okay but not great, warm, and when it's really low demand we're going to call it cold. So hot, warm, and cold. And we're going to come up with rules and we'll call those rules the trigger points that are going to trigger when a rate opens or closes. So usually, I'll say hot is what I'm projecting, 100% occupancy or higher. Now, you might say, "Well, how can I have higher than 100% occupancy?" This is the forecast, remember. The unconstrained demand. This isn't that I'm actually overselling the hotel, It's that I've got a lot of demand. And so I want to put restrictions on my lower rates. When its warm, maybe you say, warm is anything between 80 and 100%. And so I'm going have some lower rate open, but I'm probably not going to have my cheapest rates available. And when its cold, and here I'm saying maybe that's between zero and 80%. Or maybe you might say between zero and 60% I need everything. I'm not going to have anything close. I need as much many customers coming in as possible. And I use the example here of hot, warm, and cold. So three different rates, but you can use as many different as you want, and you could also define what these trigger points are but one thing I should warn you about is that hot should always pretty much be 100% or more. I've seen some hotels where they'll say, well our busiest periods are when we're at over 40%. That might be their busiest period but that does not mean, it's really not that busy, given that 60% of the rooms are empty. And so these trigger points, and we're going to be kind of using these in conjunction with what we'll call a demand control chart to be able to determine the minimum rates to quote for each day. Transcript: Demand-Control Charts So let's see how we can use these trigger points and the forecast to assign rates, and then the hotel in our example has got three different rates. So $80, $100, and $120. So the way we're going to kind of do this is that if we're forecasting over 100%, the minimum rate's going to be $120. If we're forecasting between 80% and 100% The minimum rate 12
will be $100, but the $120 rate will still be open. And if we're forecasting under 80%, the minimum rate's going to be $80. Everything's going to be available. And so if we take a look at this, we've got this table now, and we can see we've got our different forecasts, and our different forecasted occupancies. So for forecasting under 80% so like on Tuesday the first Tuesday, we're forecasting 56% the minimum rate is going to be $80. That does not mean that we're not offering the $100 and the $120 rate, those are open too, but we have our deepest discounted rate available, why? Because our forecast is for 56%, we don't want to be turning anyone away. Conversely, if we take a look at the Friday our forecast is 101%, so we've got unconstrained demand there. Our minimum rate is $120. If someone calls and wants the $100 rate, we don't have that available, but we can suggest that they come on other days when we do have those available. So there's a couple ways we can kind of take a look at this. So, kind of the rule again is the $80 rate is available to projected occupancies in the cold zone, and so it's, you know, our demand is very very low, we're not forecasting anything over 80%, and so all of the rates are going to be available. The $100 rate is when we're in the warm zone, so when we're between 80-100% in this situation, but the $120 rate is also going to be available, and if we are forecasting over a 100%, so we're hot, the lower rates are not going to be available. Those are going to be closed. And so from aside from our rates, we're going to have a lot of different rates. I mean I'm talking about three rates in a hotel. I mean take a look at your hotel, you probably have 50, 100, a couple hundred different rates out there, and so some of these are negotiated, some are public, and so forth. And so what you want to do is you want to put your rates into rate categories or rate buckets, and usually probably five to seven, so maybe you would be looking at rates between say $60 and $79, between $80 and $99 dollars, that sort of thing, depending on the sorts of rates that you have. And then when you have these different rate buckets you could also then start assigning hot, warm, cold, or whatever degrees you want to have. You could also, another thing you can do with these rate buckets, the way you put them together, is on the discount off of rack rate. But you're not going to be going through and looking at this for each different rate that you have cause there's just too many of them. You're going to be putting them into some sort of rate categories or rate buckets. And we can display this demand control information in a couple of different ways. One way is we can display it as a chart. And if you look at this chart, it kind of looks like a booking curve doesn't it? Yeah, it will, it is in a way. So along the bottom, we've got the days before arrival and along the side on the y axis, we got the reservations on hand, and you could see there's a red line, so above the red line is hot, in-between the red and the blue line, that's warm, and then below that is cold. If we graphed our booking curve into this it would go in between the hot and cold lines. And so, this is kind of a way of just looking at it graphically, trying to figure out, okay, what, what should my, what should my minimum rate start to be? So you can see, you know, if I'm forecasting something, that's going to be in the hot zone, minimum rate is $120 and so forth. The other thing you can do with this is, you can look at it as a table a kind of like, a map, and so you can start to see when is it when I'm hot, when is it when I'm cold, when is it when I'm warm? Because that can help give me an idea of what rates that I should be giving to my staff so they know the minimum rates to be charging. And this goes back to 13
when I start doing something like this, I have to train my reservation agents so they know how to quote what the rates are going to be, I also have to be able to communicate this with my website and with my online travel agents so that they also know what are the minimum rates that are available at my hotel. So we talked about the demand control chart, but what are the pluses and minuses of the demand control chart? Well the benefits is that it's really easy. And hotels that have used this, they've ended up getting revenue increases of two to five percent. So, easy is always good, especially when it works, and it helps us figure out what rates we're going to charge, and there's some logic behind it. But then the drawbacks are, well it's not looking at length of stay. It's kind of lumping all our demand together. So it's certainly not perfect, but whenever I work with a hotel who's just starting on revenue management, this is the strategy that I suggest that they start off with. You can start to get a little more sophisticated afterwards. And it not only works for hotels, I've used this for restaurants, I've used this for golf courses, this just kind of gives you a map of when you're busy and when you're slow. So overall, I mean, it's a very good strategy to use and it's one that's also very, very simple and easy to understand, and when things are simple and easy to understand, you've got a better chance of getting it implemented. Transcript: Controlling Length of Stay Let's talk some about length of stay, because customers come, some customers come and stay for one night, for two nights, for three nights, and we have to understand that because that's going to help us determine who to say yes to, and who to say no to, because we want to try to make sure that we fill as many as those rooms as possible, In a way that we're going to be maximizing our revenue. And so there's three basic length of stay controls. There's a minimum length of stay. So minimum may be, minimum you've got to stay for at least two nights or three nights. A maximum length of stay, which often we might put on a discounted rate. You could have this rate, but you could only stay here for one night. And then close to arrival is, we're forecasting such a high demand that we're pretty sure that if we take any more reservations, we're going to be oversold. But you want to be careful with all of these, these are only really to be used during a very high-demand situation. And a minimum length of stay, let me give you the best example I've ever heard of application of minimum length of stay. I was teaching a course at Cornell, a summer course with a bunch of managers from all over the world, probably just people like you, and there was a guy there from Hong Kong, and he was always smiling, and we finally asked him why he was such a happy guy And this was a long time ago, it was maybe about 20 years ago, 1995, and he said well, we've got the hand over coming up in Hong Kong and so this was a couple of years before the hand over in 1997. So he said my hotel's been sold out for six years, rack rate, minimum length of stay is seven days prepay. It was really high demand, it worked really very, very well but if he had tried to do that the following year, for 1998 or 1999, it would not have worked very well because demand had started to go down a little bit, and so what I want to do with this is I'm trying 14
to manage those really peak periods very well, because I'm going to have people wanting to stay for longer periods, and so forth. But I've got to be careful with this. What if I say minimum length of stay of four nights, and nobody wants to stay for that long? I mean, what are you going to do, lock them in their room? And so you've got to be careful, this is back to the forecast. Ideally I want to be forecasting by a length of stay, how many people want to stay for one nights, for two nights, for three nights, for four nights? That will allow you to be able to determine what that length-of-stay restriction should be. Also, what do you do if your customers decide to leave early? And usually if they have to leave early, there's some sort of an emergency Whether a family emergency or a business emergency. So they're probably not in that good of a mood anyway, and when some hotels will charge an early departure fee, you got to be very careful about that because that could really get your customers upset. There might be other ways that you can go about doing this, like having them initial when they leave, when they get ready to check in and so forth. You can also forecast how many people are going to leave early, and kind of work around that. But you've got to make sure you've got sufficient demand, because if you put this restriction on, you're restricting the number of people coming to your hotel, so you could end up, if you're really too strict, you could end up with a lot of empty rooms. You've got to be careful with these. With a maximum length of stay you're expecting to be able to sell out all of your rooms at the higher rates, and so you want to be, you can be pretty restrictive on some of the discounted rates saying, okay, you can get this rate but, I mean, you're going to have to only stay for two nights or one night or something like that. But what are you going to do if someone decides to stay and they lock themselves in their room and they don't want to leave? Well that's kind of an issue. I mean in most countries in the world, you can't necessarily throw them out. But again, this comes in to looking at when they check in, you know initialing the departure date, and so forth. But only use this when you have really high demand, because again, if you put restrictions on, when you've got lower demand, you're going to end up with a lot of empty rooms. In the close to arrival, as I mentioned, you want to be very, very careful with this, because you would only use this when you're so, so busy that we're afraid that if we accept any more reservations, we're going to be oversold. Because basically what you're saying is I'm not going to let anybody else come into my hotel. I'm not going to accept any more reservations. And be careful because that is certainly appropriate sometimes, but you don't want to be using this on a regular kind of basis. And to sum it up, the length-of-stay controls; you got minimum length of stay. So usually these occur over very high-demand periods. So at Cornell, I talked about graduation, graduation weekend I think we have a two or three night minimum length of stay. Can we get away with it? Yes. Because there's super, super high demand. Maximum length of stay is usually used on lower rates because you're pretty sure you can sell everything else out later on at your higher rates. And close to arrival is when you're saying I'm not accepting anymore reservations. But again, be very careful with these; they can be useful, but you only want to be using them when you're in very high-demand situations. 15
Transcript: Thank You and Farewell Hi, this is Sherri Kimes again. I hope you've enjoyed this course on forecasting and you've learned how to apply these tools to your hotel. I mean we've talked about the booking curve and how you can use that to develop your forecast. We talked about demand control charts, so how you can understand your cold, your warm, and your hot periods. And we also talked about length-of-stay controls and how you can use those to help control your capacity. But again, remember with forecasting, this is the building block of revenue management. So it's important to spend a lot of time looking at your forecast, developing it, and measuring its accuracy. 16
17