Evan: Hello and welcome to the Mining Your Business podcast. I’m your host, Evan Wimpey, and I’m excited to introduce you today. Jay Lanterman is the Senior Manager, of Operation Analytics and Insights for the Americas region at IHG Hotels and Resorts. He previously worked there on the Competitive Intelligence Team, where he analyzed and forecasted revenue performance across the global hotel industry. Jay’s background is in mathematics, where he holds a Ph.D. from the University of Georgia. Jay, welcome to the show. How are you?
Jay: Thanks. I’m excited to be here. I’m stoked.
Evan: Okay, fantastic. Let’s get right into it. I gave a very brief introduction. Maybe you can tell me a little bit about your background and then the current role that you have here at IHG.
Jay: Yeah. So I went to grad school for mathematics and everybody, you have the right impression when someone tells you that, you think, oh, you’re going to do research and be at a university forever. So that was the goal. And I quickly got into the more applied side of things and decided I wanted to go out in the industry and model some real stuff. And so I ended up at IHG. No, no one in the hotel industry judges me for this, but I am a little skeptical, right? Like the job interview went. Well, I was excited to kind of like meet the team, but the subject, I was like hotels or they said anything with meat on it. They’re like, hotels seem basic, but no, the hotel industry is wild. There are so many factors going on, and so many components to think about. So that’s been a great problem set for me since then. And working on the competitive intelligence team I think was the right way for me to kind of get my chops analytically in the hotel industry because there are so many brands. You’ve got so many factors to consider. You know, global placement, you’ve got chain scale is one of the things, you know, kind of the brand level like luxury versus economy. So it’s been a really good journey and now in Ops, really working on it, kind of a focus on the guest experience, you know, like that’s a lot of where the operations team gets their foothold is making sure that hotels are delivering what we consider the product, which is, you know, the clean room and the good service in the hotels.
Evan: So, yes, there’s a lot there. And I think you hit on it when you hear a Ph.D. in mathematics, it’s your typical career path that I think you plan out. It’s not going to the hotel.
Jay: Definitely not.
Evan: So, yeah, you mentioned that you know, you’re hoping that there’s some meat on the problem. There’s someplace you can presumably use your mathematical chops there. Maybe you can just give us a little flavor of the types of projects that you work on and the types of data that you have there?
Jay: Definitely. So one of the best projects that I worked on actually was during the pandemic. And we can talk about some of the obvious effects of the pandemic in a minute, too. But the project itself was really, how do we re-baseline? What are we going to do? How do we plan anymore? If you kind of rewind your brain to what feels like a hundred years ago but somehow only two people were wondering when people were going to go outside again, much less sleep in another bed. Right. So we had to worry quite a lot about, okay, we know what we’re going to do maybe, but we don’t know how to be ready to do it when to be ready to do it. And there’s a ton of planning and contingencies and scenarios that had to go into that. And so what we’ve managed to do was build a couple of scenarios that were able to make sure we were ready at the end of 2020 and then coming into the beginning of 2021, especially in the US, that demand did pop back what to expect. We kind of had some scenarios ready in the forecasting. That said, it looks like we’re in. I’m going to make this a scenario to be like how things are happening. We need to be ready to do A, B, C, and D by the end of this quarter because it looks like in Q1 we’re going to have this happen. And so we were pretty right on it with that and I was proud of that to be able to kind of in a very uncertain time manage to make sure we had enough contingencies there forecasting-wise. So that’s kind of one way that you could think about operations is a lot of the financial reporting and analysis. You think about brand companies that way a lot, right? How do we make sure that we’re offering the right amenities? How do we make sure that we have the right brand image? How do we make sure we’re marketing to the right customers? Operations is a really interesting lens if we’re worried about what happens in not the hotels, but in that hotel. Are they keeping the rooms clean or is this, you know, what are they doing right? Maybe we can take it back to other individual hotels and tell that individual front desk manager, this is what we’ve seen working and this is going to help your hotel a lot. So it takes a lot of that macro smooth messaging to a lot of the micro. Very. Noisy and super hard to model out on low-level messaging. So that’s been a really good experience too. And like you said, it’s been a ton of meat to work with. Right. There are so many more problems than I thought it was going to be.
Evan: Yeah, That’s exciting. Yeah. Even. Even just a cursory thought about it. There’s a lot that goes on at an individual hotel. I’ve never worked with hotels, but I’ve been a guest many times and I can think of so much of the data that’s collected for each guest experience there. And on that data collection, you know, you’re focused at the very micro level, at the hotel and the guest experience level. A lot of your hotels you can correct me here, maybe this isn’t right, but I think a lot of your hotels are franchise operators. So does that data all sit in a nice clean data lake somewhere for you to go? I take it from your life that it doesn’t.
Jay: So again, if anyone’s watching, is it IHG data architects? I love you. Thank you for all you do. But that doesn’t eliminate that. There are huge challenges. You’re exactly right. The overwhelming majority of estate and the big three U.S. are America’s hotel companies, Hilton, and Marriott night sheet. The overwhelming majority are franchised. And so there’s the obvious implication there. Someone else owns and runs this hotel and they just have an agreement with us for the brand. And so there’s a lot of symbioses there. But there are challenges where, you know, a lot of times you get buy-in and we’ll recommend you should use maybe this PMS system and we’ve got great connectors that’ll help us speed data out of there. Occasionally you get maybe an older hotel or hotels, converting brands, right? You’ll have a hotel that used to be a Wyndham and now it’s a Hilton. Right? They might have existing infrastructure and they may say, I don’t want to invest X thousand dollars to do that. And so now you’ve got to be able to get any kind of data out of that hotel, to know how many rooms they’re selling, how many rooms they’re listing, etc… You’ve got to be able to support that property management system, too. And so there’s a ton of even before you get it into a data lake, there’s a great amount of nonuniformity and then there’s a ton of cleaning to get that on like the same granularity and the same basis. And then those come in maybe on different levels and you’ve got to merge three data sets and I could go on there. It’s an enormously complex problem and I’m very fortunate that at least it’s cleaned up some by the time it gets to us on the analytics teams. But it’s certainly not perfect and requires a ton of kind of getting in the zone of how this is structured. This data sets on a day-to-day basis. This data sets on an individual booking basis and has this other information and all kinds of crazy nuances like that that make it tough to keep clean.
Evan: Yeah, well, that’s good. Nobody wouldn’t want the listeners to get too jealous that you’ve got to know, I think I think you’ve got it’s not unique to hotels problem for sure. But it’s interesting too to think about how many franchises there are and how many independent data generation tools there are. I’m curious if you did. I appreciate the shout-out to the data architects. Can you mention maybe you’re aware if there is a dedicated data architecture like the data engineering team that tried to serve some of this data or does that come from the same analytics team that you’re on?
Jay: So no, definitely not. It is there is a dedicated kind of architecture team that handles a lot of the ETL and makes sure that we get all those stages we just described right, there’s some degree of what we call a daily check out in the hotels that generates some kind of file that has to be aligned so that it can be ingested over here so that it can be put out based on a lot of conditions in this format to this dataset. All those steps are handled by a dedicated data architecture and governance team. The follow-up, though, with analytics is a lot less central, so we have a really good analytics community. Make sure that we all stay in touch. But here’s what I would call large pockets of embedded analytics. So it’s almost a hybrid approach. So for example, the marketing organization has a large data team and so it’s very central to all of marketing’s functions, yet it’s embedded in the marketing function. And so a similar kind of thing for finance and then operations each in operations just because of the complexity of being in a hotel and having to operate in lots of different countries. We have one per region. So there’s like I said, it’s kind of in between it’s certainly nothing close to all the way central, but it’s not super fractured where like every team’s got a data scientist or something.
Evan: Sure. Okay. Got it. And are there a lot of crossovers? Do you work in Operations and Americas? Are there a lot of crossovers? Do you work with your counterparts in other regions or do you work often with presumably you’re dealing with a lot of the same data and probably a lot of tangential questions, at least with marketing or finance?
Jay: Definitely. Yet, as I said, we have a really good community that tries to keep everyone involved. And so you could imagine maybe a team that’s focused on a certain brand notices that one particular metric, it’s just really skyrocketing. We’re doing great here. And they know they’re doing a good job, but they want to know what we point this to? We’re doing three or four different initiatives. Is there one that’s doing this or is this just a compounding effect that we’re doing four good things at once, or conversely, to imagine it, but maybe something’s falling off? What’s going on? You know, the sky is falling a metric down, right? And so they’ll reach out to the appropriate region. And then, as you mentioned, we’re working with a ton of the same kind of data for seeing slightly different trends from the other geographic regions. And so we might reach out and say we are seeing something bizarre. Are you guys seeing that, too? Is this something that’s maybe a global consumer trend? Is this something that is regional pockets? Can we pinpoint this whenever you have phone problems, right? When you get to something a little forensic or you’re not, you know, you’ve detected something that you’re not 100% sure what it is yet. There’s a ton of crossover with the other regions there to see, you know, are we alone in that? Is there something we can point this to, etc.? So it ends up feeling like one very large team, even though it is structured like a lot of medium-sized teams.
Evan: Yeah, that’s great. I’m sure there is a ton of shared learning there. Yeah. So that’s super exciting. And I, it feels like with operations probably more so than with maybe some other business units where you care about the micro and how a single hotel is performing it. When you find somewhere, you find some of that signal or something interesting or a new best practice or a new risk is that your team is at the analytics team that’s trying to action that. Are you going to hotel operators or hotel managers or is there a different function within operations that tries to sort of translate your work or put your work or insights into action?
Jay: Yeah. So there is a second layer, right? So there’s the way I think of it as almost like a reverse pyramid, right? You’ve got at the very bottom as far as granularity individuals like working at a single hotel, maybe the GM of this hotel, they’ve got some brand reps that come to them on different components. Right midday you might have someone who’s helping them in diagnosing. How are you pricing appropriately? Do you have a good revenue management strategy? If somebody else is coming in and saying, hey, we want to make sure that you’re using the latest clean policies or so on, right? So they’ve got support managers that are kind of in a literal sense physically going into hotels every day, but they don’t work at that hotel. So it’s kind of the in-between level. And then one level above that is me. I like to remind people I’m in my ivory tower. I don’t know how hotels work. I just analyze them. Right. And so I will go to them and say, hey, we’re noticing, you know, Brand X has this weird phenomenon going on. And we notice that these 12 hotels seem to be leading them. They’re skewing the trend here on whatever we’re measuring. Right. These managers are responsible for interfacing with them. Can you guys go? Do you already have insight? Maybe so. It’s kind of bi-directional. They’ll let us know. Oh, we already knew that this is something we didn’t expect would affect the numbers, but it’s got to be this. Or on the other hand, can you go investigate? Maybe that hotel’s got something interesting going on. You know, I’m sure everybody’s seen it in the news. Hotels and restaurants, just hospitality in general, have got a crazy labor market right now. It’s been a real challenge to get people recruited. And then when you get people recruited, it’s kind of a cycle of what we’re understaffed, but you just started and now you have to take more and people burn out. So there’s a kind of revolving door in several different properties, right? So you sometimes get explanations like that where it’s maybe this one management position not staffed and maybe that’s why this one feature of the hotels falling off. So we get a ton of really deep insights that way, kind of with an iterative process that we’ve done some analytics we can identify. There’s an anomaly here and it’s led by these hotels. And then let’s get kind of the field insight of where we’ve kind of put this in a magnifying glass and focused on the right point. We can figure out what’s going on and scale that backup to solve a macro problem.
Evan: Yeah. That’s great. I love the visualization there, the reverse pyramid, and self-deprecation. And you’re in your ivory tower. Of course. And of course, you’re up there with your Ph.D. in mathematics as well. And so I think you sort of just hit on it. But is there anything that you tried to focus on specifically when you’ve got to articulate something down to that intermediate or even that individual hotel level you’ve found? So maybe it’s something easy like, hey, here’s an anomaly. You know, this metric is lower than we projected it to be or lower than it should be. Can you investigate? But presumably, some things in data and analytics are that are challenging to convey without using extremely technical language. So is there any rule of thumb or any way that you sort of think about trying to convey that?
Jay: When I was in grad school, one of the things that I did not enjoy so much was trying to teach undergrads. I was an undergrad. Again, no offense to undergrads, but especially, you know, you’re a grad student and some of the jokes about the grad students are true. You don’t get the fun classes. And my wife thinks it’s funny that I think there are any fun math classes, but you don’t get to teach like, you know, the stuff that you’re interested in, that you’re excited about, that you remember being a fun class, you’re teaching like freak out. And I’m sure everybody remembers and no one wants to be there. You’ve got one student who thinks this is cool and the other 49 think that you’re an idiot, right? So you’ve got to convince them now, right? You’ve got to get a relationship with them and figure out what makes them go. And that has translated well for me. So thinking about that, like as an analyst, you know, you’re interested in stuff and you know that your stakeholders are interested in the things that you’re modeling, but trying to kind of build up a relationship where they know they can trust you, you know, you’ve delivered some results and you’ve got that good relationship and then boiling down to truly like the essence of what they care about. So if I go to some of these stakeholders at this point, you know, I’ve been in IHG almost a little over four years now, and I tried to build some individual relationships with a lot of stakeholders so that when I tell them, Hey, something’s weird, they don’t think I’ve modeled it wrong. You’re calling me out. There’s an ulterior motive. They think Jay’s looked at something and it’s weird and it might end up being nothing. But they already kind of are primed. I hope that it’s worth a cursory check. It’s worth making sure. And I’m just like I said, boil it down to as simple as something’s going on with this score, it looks too low. Do you think it might be A, B, or C? The second piece that I’ve seen is boiling it down to a hypothesis you might have and taking a little bit of ego out of that, you know, like be okay if your hypothesis is wrong. I’ve encountered a few people I think everybody has to get clung to. I’m just sure it’s this, you know, sometimes it’s not. And so helping them with a little bit of lead, that’s what the analytics is for, right, is to guide the business and help them focus rather than tackling everything all at once. Give them a little bit of lead and just boil it down to this is what they care about. And they know if you’ve got a good relationship with them, that you’ve done the work on the back end, you’ve done the good analytics, you’ve built a good model, you’ve made a good predictor. And if this is off base and you think it’s significant, the hope is that they’ll trust you here and they’ll be able to act on it for you.
Evan: Yeah. I think those are some great lessons that I think resonate with what we’ve heard from a lot of guests on this show, a relatively young show thus far. Yeah, I think you’re right in expressing a little bit of humility that it’s not. I have the answer. I’m smart. I’m super smart. I can run this super smart computer and I’ve put out this answer. So you just operate a hotel and you listen to these answers. That’s, I think, the bad perception that analytics teams get. And I think it’s probably easier when you’ve got the embedded analytics teams to help try to build those relationships and garner some of that trust. Not that it’s impossible, but it’s probably a bigger challenge with the more federated, centralized data science team that is not able to build closer relationships.
Jay: Right. That’s been my experience, too. I think that for really huge macro projects and especially in an environment that might be very project-based, I’ve seen a ton of the results from those types of teams because they can be focused, laser-focused on this is my skill set and we’ve got a huge, great workflow structure where we can make sure that when it gets to my skill set phase, I can focus on my skill set, some of these broader problems. I mean, you’ve kind of heard me discuss it here. We might be worried about how many towels there are. Are our guests satisfied or are we charging the right rates? Are our guests? Is there a weird obstacle? We’ve had things before where there are obstacles to parking and guests will leave and cancel reservations. Like, How are you going to find that in analytics? That’s a bizarre, crazy thing, right? So you’ve got to have it’s kind of this weird what I described with like the magnifying glass, right? It’s like a zoom-in, zoom-out kind of approach of let’s get tactical and forensic, and then let’s zoom out and see if there is a bigger problem and then zoom back in over here. And I think that without having the relationships because we don’t have experience running hotels, a lot of us analytics professionals don’t have experience running whatever business we’re supporting. And so having that intermediate layer of letting me go talk to you. You’re in here and you know what’s going on in this building day to day. I see this and I think maybe it’s X, Y or Z. Let’s talk about what it is. And being able to take that back has been hugely critical and I don’t think you can get that as well without that relationship-building component.
Evan: Yeah. That’s a great takeaway and expresses humility here. Analytics professionals. Maybe this isn’t relevant. I’m kind of hopeful that it is. But our previous guests, we’ve talked to folks in fast food, in retail, and it’s pretty straightforward if they want a little bit of visceral. Low level. What are things like? They can visit a chain or they can visit one of their stores. Do you ever see operations analytics in operations across the Americas? Do you ever do some data collection trips where you just go stay at some properties to see how things look?
Jay: On my previous team, definitely on the competitive intelligence team, you can imagine, I mean, to maybe abstract myself if I was in fast food, right, I might want to go eat at another restaurant. Right. And see what they offer to do? I think it’s good. So that was hugely valuable for us in the competitive intelligence team, especially thinking about brand positioning, and making sure that there are always good marketing materials. But in my opinion, in a hotel, until you get there, you don’t see how much each of those might be, say, weighted, right? They’re saying we’re excited about our social space. Well, you might be excited about it, but nobody’s sitting. And I’m not so worried about that competitively. Or on the other hand, maybe it was just a throwaway, but that’s where you got the most pop-in lobbyists out, right? So that’s that it’s hugely valuable there in operations. It’s often kind of experimental, so it’s a little less frequent. It’ll be something like, I’d like to get this data point, but recognizing especially again, this labor environment, we don’t want to create a ton of work streams. We need to figure out what’s a sustainable way to get more of a data point. It might be something very technical like let me see how you’re doing action and let me see where opportunities for us are as close to seamlessly as we can capture information about that without disrupting operations themselves. And so a lot of times this ends up being. Like I said, highly experimental, and a little bit of hope in the wind. You know, it often ends up being that we proxy it with something else, but we learn that that’s something else. The proxy, which by kind of knowing what their process is and saying, oh, these two things are very related, they happen almost always at the same time. So if B falls off, we know A has probably followed us. Right.
Evan: Yeah, that’s great. Yeah. You get so much more from the physical, real-world experience than you do from bits on the computer screen. So I do want to ask you quickly, you mentioned earlier about some of the covered work that you had done, and I’m curious if you maybe can expand on it a little bit. You mentioned mapping out potential scenarios. So I guess sort of a two-part question is how do you do it? It is the breadth of potential scenarios just really wide there in mid-2020 when seemingly anything could happen. And then are we still back into something more predictable, something more normal now, or do we still have scenarios? What happens when there’s a new variant or when people get scared to travel or when new policies are put in place?
Jay: So I think the more interesting way I can answer this is probably on the guest experience and guest satisfaction side than the demand side. That’s, if you think about it, a hard thing to predict. There’s not just a demand level factor of what guests and their guests are satisfied with and what they are worried about. There’s additionally that micro level of this brand. Operate as well under those conditions, even if I have a good baseline of how satisfied our guests are with this brand or that hotel. This is a new set of conditions. How much can I rely on that baseline? And so a lot of scenarios were based on how. Is this hotel or segment performing versus those well-established benchmarks that we had pre-pandemic? How is that deviation perhaps moving? And then the scenario piece comes in on how much we can trust that trend. And in particular, this is where a really interesting crossover is. Where does demand play in? So you could imagine maybe I’m going to make this up intentionally wrong, but maybe leisure guests are being served a lot better now than they were pre-pandemic. And anybody who’s done any guest satisfaction in the hotel and she’s laughing at this example already. But maybe they are. And maybe we expect that leisure. Like how many guests are leisure guests going to drop off shortly? We’ll like to think about modeling that out. That’s the obvious implication as well. That might have a lesser positive impact on how our scores are going to look. But the demand is what’s been hard until recently to predict which is you know wins business going to come back to one of the corporations but it’s going out when a corporate policy is going to allow when our meeting is going to come back. A lot of that is starting to look a lot clearer, especially in the Americas, which is convenient for me. I have nice data and a nice outlook. The model globally is a little less clear, but still much clearer than a year ago. So in general, we are having pretty what we’re thinking is pretty concrete outlooks on both guest satisfaction and the demand outlook. But during COVID, in particular, you hit the nail on the head. It was, you know, when are people coming? Where will they go? How long will they stay? Even that was that we used to have really good benchmarks on. Sure. At this time of year in this market, people stayed this much one night, this much three nights, etc. We had no idea truly going into it. It was, you know, opening Pandora’s box. What are all the things we think are feasible and just see what rolls out of the models? Right.
Evan: Yeah. It’s wild. Well, gee, I hope. I hope things don’t get too predictable. I want to keep it. Keep you on your toes, and give you plenty of fun, analytic stuff to do. You’ve been, I think, just over four years. You said you’ve been at IHG. I’m going to give you that. The magic wand here. You’ve seen a little bit on the insights team. You’re in operations now you get to pick yourself and your team. It’s the Jalen Truman Show and you get to decide what your team gets to focus on. Things are in relatively stable shape post-pandemic and you get to solve a fun problem or an interesting problem or a big impact, whatever you want your team to go on. And everybody’s aligned from hotel operators up to up to the executives at IHG. What’s the thing that you like to work on?
Jay: So, Evan, I cheated and I watched the old episodes of this show, and I knew this question, and I’m going to cheat twice. I’m going to have to answer once the boring but practical answer and once the finance the practical answer is exactly what we discussed at the beginning of this, which is the franchise data, right? It is. There are lots of difficulties. And if I had a magic wand and I could convince everyone, look how great it would be if we all used the same system that reported the same way and we could get the cleanest clean data and everyone operated it and set it up perfectly. That would be wonderful. It’s impossible. So that’s the boring answer because it’s kind of clean data, right? Everybody is saying.
Evan: It’s a shame that’s the boring answer because it would be the honest answer for most folks.
Jay: The fun answer, though, is I think that. Especially the last few years, even right before the pandemic, I was thinking there is room for cool machine learning problems for aggregate hotel modeling. And in particular, I think that what’s missing is a fancy clustering algorithm. I think. And so take my idea, people. I’m interested. AUDIENCE Please take it away. Come back to me with a cool product. But I think that there’s a lot of room to instead of, you know, we think about that. We have, what, almost 5000 hotels in the Americas, right? Well, a lot of people think about 5000 data points or they think way out. You know, let’s just model one brain at a time. And now we’ve got like 15 data points. Neither one of those, I think, is right. I think that the right move is can we build a nice machine that will know these hotels have to depend on what metrics are looking at the same performance profile and let’s pull those together and maybe there’s I’m making this up 20 types of hotel and they perform a certain way that’s really what and so now we can start building 20 models that we can aggregate up and say here’s how the industry is going to move without it being as you know, there’s a ton of factors a lot of folks maybe don’t think about until they get in the thick of it. Right. Like, what kind of market are they in? Are you an urban or rural hotel? Are you on the outside of the interstate? Are you at the beach? Even all those might be the same brand. Right. And so that’s a complicated problem. But I, I just feel in my bones, man. There’s an answer to it.
Evan: Yeah. That’s certainly exciting to the listeners in the hotel space or with access to any hotel data. Get back to J. Let him know what those nice, robust, and repeatable clusters are for the hotels there. Jay That’s all the time we have on the show. You’ve been fantastic. Thanks so much for coming on and talking to us today.
Jay: Yeah, I appreciate the invite. I’ve had a blast on here and thanks a ton.
Evan: Okay. All right, everyone stays at IHG hotels. If this is interesting content to you, make sure you like to subscribe. Multiply divide. Come back next time and we will see you on the Mining Your Business podcast.