Hyperpersonalization: Tailored Experiences in the Age of AI

All right. Uh, so I appreciate everyone holding off as we transition from the first, uh, presentation today to the next. Uh, I'm Brett Kjuski. Uh, I am the VP of research and growth here at Accelerant Research. Uh, you know, thanks for joining us today 17th for our ARC conference. Um, you know, just going over those ground rules again, uh, that Bill probably started with, which was, uh, obviously we're at the mercy of tech. Um so we're going to be sharing there might be video clips or might be uh different interactions uh throughout these uh different presentations today. So again we are at the mercy of technology and and you know uh please be respectful of that. Uh the other part is being respectful to others. Um obviously you know these presenters are you know volunteering their time to speak about things that they're passionate about but also those within the comments. We hope that it does stay engaging that you do ask questions, comment, um, you know, follow up with with things that resonate with you. Uh, and then finally, use this as networking. Um, you know, the point of putting this on is to bring our community and research, market research, all the different disciplines together. Uh, and please use this as an opportunity to network, opportunity to engage, interact. Um you know today we have the pleasure of having Gavin uh Gavin Johnston who's the director uh of UX uh research at Hyatt Hotels. Uh Gavin's talk today is going to be about hyperpersonalization uh the tailored experiences within the age of AI. So uh something that's relevant and top of mind to everyone here um in terms of AI and and where that's going and how we utilize the personalization and customization of that. But uh without further ado, Gavin, I will throw it over to you. I'll be here for questions and being able to uh you know respond to the chat as as people bring that up for you. But um outside of that, it's it's your show and I'll stop >> and throw it over to you. >> All right. Thank you. Um as you all notice, my beard has grayed since that last picture. I need to update my LinkedIn profile, obviously. Um just a heads up, I had a new computer arrived the other day, so hopefully everything will go as planned. Uh, Mac isn't what it used to be. So, uh, I'm going to share my screen here and fire up a presentation. I'll have to excuse the Carvajagio piece up here for just a second. All right. So, uh, there we go. So, as was mentioned, I'm going to talk a little bit about hyperpersonalization. And this has become a hugely important element for us at Hyatt and everywhere I'm pretty sure. And a lot of the questions after investing a lot of money over the last couple of years that are finally starting to emerge that didn't at first really revolve around uh just because we can do it, should we do it? Right? And I'm sure that that's a big point for everyone. I think we're kind of at a point um where six months ago, a year ago, you know, if you'd said, "Hey, you want some toast?" No. Do you want some toast with AI? Yes. Give it to me. I don't care how much it costs. So, we're running into those issues. And I think for our standpoint on the research side within Hyatt, certainly with UX and CX, the question becomes how to balance all of that, right? So, just real quick, a little about me. Some of this will seem strange, but it becomes relevant later. I've been doing this for about 20 years. I'm not a designer. I'll be the first person to say that. Um my background was in linguistic anthropology about a billion years ago. Um couldn't make a living at that in the professor world. So I ended up moving into um design research instead. I've worked in everything from startups to Fortune 100 companies. So me alone by myself doing work to managing teams of 30 people plus on any given day. Um, as I mentioned, UX research, spent a lot of my career on the digital advertising side and strategy side. Uh, and I'm actually a former chef, which is largely irrelevant until we get later into these things. I, uh, did that before I went to grad school. Um, finally, yes, hospitality is where I work now, but um, I've worked in everything from doing field work and things like gyms to working with live fire exercises um, to working in surgeries. So done a whole lot of strange things and I mentioned that mainly because a lot of what you'll hear me say today is the word context. Um coming out of anthropology, coming out of more of an ethnographic background originally, um obviously that focuses on understanding why people are doing it and all of those subtle nuances. So as we get into this um what do we mean when we say hyperpersonalization? I think there's a lot of different definitions uh obviously from different different uh any organization you go to you're going to get a different perspective on this but things I think that come up constantly are it's proactive meaning we are reaching out it's obviously a push rather than a poll it's predictive we know that people are going to do x y and z or assume they're going to do x y and z so we can model around that it's clearly individualized it's highly targeted um and Ideally, it's supposed to be contextualized. Now, I can get into that a little bit further as I mentioned. Um, we are using obviously AI and machine learning. We have been for a while. Uh, everyone is. There's nothing unique about that. And the goal is obviously to create those tailored experiences. So, for better or worse, we're getting that's becoming more frequent. We have a lot of work going on in these areas right now. But I think the key point is at the end of the day, what this all really boils down to is you want to make something very relevant that will drive a sale faster uh with less clicks or less steps. Okay. Uh there we go. If I can get it to move forward. There we are. So hyperpersonalization um it offers obviously significant potential but I think there's also a lot of challenges. The ones we've all heard data privacy concerns it's obviously an issue. It's very resource intensive. Um we're discovering that writing a prompt frankly often takes as much time as just simply doing the work itself. Right. Um there is an element where we can get overpersonalized. We'll get into that in a lot more detail uh later on. Um with some of that too there's a process of constant maintenance, constant change. What chat could do a week ago, right? It's doing differently today. Uh 6 months ago people were interested in Claude. Today it's the darling of the industry. So from our perspective, certainly at Hyatt, um it's important to put hyperpersonalization into the perspective of why people are looking for a hotel. What we do is radically different than say shopping for groceries. So again, we'll get into that as well, but I think the key thing is we want to align business behaviors with customer behaviors and actually be thinking about behavior change through time. what can we do immediately but how will that change behavior down the road and have that as a constant model for us. So the goal today for me is and I'll be using a lot of examples as we get into this because again I think you're going to know some of this. Uh my goal is really to walk through what I think are some of the most effective ways to optimize the experience, how to minimize those risks, um and figuring out how to avoid the creep, right? how to make it less annoying um and really optimize it because uh at the end of the day we need to align what's going on with the quantum qual. We need to recognize everyone's familiar with the human in the loop idea. What can AI do well? What can't it do well? What shouldn't it do well? Um how do we avoid those pitfalls, right? And how do we avoid that creepiness? So with that, I'll go ahead and kind of get into the details. Everyone's familiar with this statement. I think quant's really good for what and the how. Um, and I do think large sample data is the backbone of a lot of business decisions, right? But by itself, as we all know, it's pretty problematic. Uh, it's incremental. So longtail behaviors, I contend, can't really be well established. That's partly by that's partly the nature of the work but it's also largely the way business works right we don't think in terms of 6 month 8 month 9 month behavioral observation we tend to do AB tests that run for you know 3 weeks 3 months at most things like that um it also can be skewed fairly easily that data right at least and that depends mainly on the industry but as an example uh within hotels if you on uh a test let's say in April that's not a very good predictor of what's to come because that's the time of year when we see a decline um in reservations similarly with the situation going on with Iran um we have no way of predicting what the data was you know how it was relevant a year ago doesn't matter today so there's issues within that and I think it's also a matter of data focuses is largely when we're talking about personalization on individuals rather than cohorts, meaning relationships. So, as an example, we can track my data, right? When I'm booking a trip, but I don't actually book that trip alone. While my wife may never actually touch the computer, I'm still going back and forth with her and having conversations, which changes my behavior. And obviously, the data can't account for that, right? So how do we balance that? We break it into three areas uh at Hyatt and try to obviously again there's a million ways to do this but for us we try to base it around three key principles. First of all psychology taste temperament experience, all of these things we know about an individual through time through booking behaviors and things like that. On top of that we take a close look at semantics and semiodics. um the word choices that we have for people to respond to become extremely important. And admittedly, like I said, my background is in linguistic anthropology, so I'll nerd out on language all day. But the point is, what we serve up to people, how we create an offer needs to be tailored to their region, their accent, subtle cues within it. If you're going to be trying to sell something to somebody, say in Alabama, uh you would be wise to make a reference to SEC football at the beginning, right? There's little things like that that you start to build in. And the finally, the last one is culture, shared behavior. We like as a society to talk about individuals. Makes sense, but we are all products of a cultural lineage. And so, we do have things that are subconscious we're responding to that we don't even realize, right? All of those things for us have to be factored in so we can mitigate the risk as we go along. AI uh has a lot of promise and I think we're all discovering this. I also think you know that I mean last week as an example I had eight vendors calling at different times saying they had the perfect AI solution to all of our problems and the world was going to be great. And you know bluntly I think we've all heard this before. Uh that's why I threw this image up here from 2016. Big data was the thing that was going to to solve our problems. Two years later, people were strapping EKG machines onto people when they went shopping in the grocery. 2000, ethnography was the thing was gonna be great for everything. So with that in mind, the things we've discovered and we drive home with our business partners to say AI is really good for removing clutter, right? And it's helpful in identifying the noise, veined causality, right? It's very good at that sort of thing, making sense of the data. Um, it's very good, I think, at connecting potential dots. We use it quite a bit. We I love outliers. There's something that frequently those are people who are very innovative. They're doing something unusual. So, it allows us to use that to generate new hypotheses, new design ideas, concept testing. Um, and with that, it's great for experimentation, right? We can do more rapid test and learn through AI. um it allows us to actually build things very quickly and let's say Figma make we're then able to test out with individuals or groups across the board and gather that data. So I love it. I'm not nearly as frightened about it as I was say a year ago when everybody was panicking about the future in AI. Um but I do think we need to be very cognizant of its limitations in relation to the industry we work in. Okay. Um that said where it falls flat is it doesn't really understand context and this is I think for researchers where we will remain relevant going forward and ne be a necessary component in tailoring experiences to individuals. It doesn't do well with things like understanding cultural nuances, shared interactions, positive outliers, social right norms, all of that sort of thing. And that's I think where we come in. uh give you an example of this uh from past work um using AI and I ran this experiment a little while back to see if it still hold true. Um grocery shopping is a fairly simple seemingly simple straightforward thing, right? Uh it's about making a very constructed list. It's about certain behaviors that we can quantify and predict and consequently we'll lay out a number of things to say, "Oh, give this person a coupon for Kool-Aid or whatever it might be." The catch is shopping for groceries is a lot more complicated than that. Why? Uh because the list itself can be a shared thing. Some folks have a formal one. Some folks have a piece of scrap paper that they write on and it gets passed around the house. Shopping for groceries is a place where we teach our children norms and behaviors. It's an expression of cultural aspects of who we are. Um, frequently people will shop together and it becomes a point of play, right? Women still tend to lead the shopping experience in terms of like technical procurement. But what you often see is the husbands will be with them and there becomes a game in a lot of cases where you'll slip something in. you know, you're slipping something into that cart and getting caught. That's the point is to be caught. So, it becomes relationship building. It becomes a tool for expressing love and affection, that sort of thing. So, the point is AI can't get to that sort of thing. And in our business, certainly in hospitality or very emotion-driven industries, um, understanding this is extremely important and allows us then to tailor those messages to individuals based on a number of factors AI can't necessarily get to. So we employ both of those. Oh, so how do we get around that? We no longer really use journey maps. We have moved primarily to servos blueprints. Um again I think this is going to depend on the industry. So depend right it's this is what works I think for hospitality hotels specifically because people are putting a lot of effort and thought into this. Um, we also combine not just the the the individual data, but we look at the persona across a number of factors to see, okay, what are some of the shared behaviors we might be able to capitalize on, not just what that individual needs, but things we can then tap into and kind of tweak to to really make that more not just personalized on an individual level, but again on that deeper subconscious cultural level. Um and with that it also allows us to define what we can actually influence right what's the proper place what's the proper time is it a digital experience is it training our uh front desk people to speak differently to do certain activities that are a little different right uh by laying out the blueprint that helps us stay on top of this when we do that and again I'm sure everybody's familiar with seeing these um we lay out the stages of that journey, all of the pieces along the way. Uh within that, we look at all those hum the user interactions. We catalog those all of the um the lines of interactions, the decision points that people have to make and it gives us a much richer map of areas that then we can affect and tailor different kinds of messages and offers at different times in the journey. So this is one example and these are admittedly look to a lot of people I think scary and we've had to do a lot of work with our product teams especially saying no no no don't be freaked out we'll we'll explain to you how this works but this is an interaction an example of where we ended up broken out into sections um of a call center right these are the various situations that people have to work through the call workflow email workflow because they're frequently talking to someone as they're looking at an email or a chat reservation workflow There's all of these elements that come into play when you're trying to do anything at a call center. We take it for granted that's a relatively simple thing. It's actually extremely complex and every interaction we have is like that. If you take something like uh you know picking up cold medicine, am I picking it up for somebody else on my way home from work or am I sick? Uh which means I am less likely to easily process cognitive right information. um I'm making decisions differently, right? There's a whole series of things that we would work into that. Once we do that, our team takes it and puts it into um a plan, a fairly simplified action plan so we can start doing experiential design, allows us to spot those areas for constant adjustment. And within that, we start to put opportunity maps. What are they doing today? What can we start to act on that will change behavior so that in a year we can be doing this instead? Right? I'll show you an example. I use myself as a guinea pig in this. I'm multiple personas on multiple people as they would break me out because I work for uh the hotel company. I'm a luxury traveler, right? We I get a discount. So, of course, I'm going to go for these hotels. Uh, depending on what I'm doing, I might be considered an adventure traveler, right? I'm going for a deeper experience. Going to Santiago, Chile is not the same as going to Chicago. I'm also now an empty neester. That changes what I look for and how I look for a hotel. I'm also a former chef, which means uh when it comes to things like food, I'm insufferable. My wife tells me that, I know that. We all know that. But it's something that I will look for when I go to a hotel. Yeah. I care about what their whiskey selection is. I pay close attention to that. Um, I'm also an event traveler, right? We all have to travel for different things. So, I mentioned this because as we start to go through it, we'll start to calculate my locations. Where might I be? Where am I going? And what can we capitalize on there? If I'm going for work, can we get an extra couple of days out of somebody like me? How do we go about doing that now that I'm an empty neester to say, "Hey, let's bring my wife along." Maybe it's the spa. Maybe it's an earthther experience that we want to build in so that after I do a couple of days of work, right, we can now build that into the booking funnel. Um, and think about all of these things like seasonality, right? Profiling, targeting, um, secondary profile, targeting, meaning my wife, right? Um, and thinking about cultural norms. So all of these things then come into play for us when we're trying to build a tailored experience and aligning the needs of the guest who may not even know their needs. Ideally that's actually a great area to start and then we can tailor those offers for them and time them um tie them more closely to various events throughout the day. Right? So, as an example of this outside my industry, um there's a reason that people like into it target people to do tax work at about 11:00 a.m. It's because people do their own taxes over lunch at work or they do them when they get home. So, they're able to tailor those things to people and look at when you're actually going in and doing your taxes and giving you a reminder an hour or so before to do X, Y, and Z. Right? So, pretty cool. It also gets to a point where it can be a little bit problematic. Um, we've certainly had our issues of course like everyone else in any given industry. Pitfalls and dangers, things that we have seen that really step out um, and we have to pay close attention to. And some of these I'm sure folks are fairly familiar with. filter bubbles cause problems for us because you that that excessive personalization uh means that people end up getting trapped in their bubbles, right? And they don't see content that maybe we want to get them to be looking at new things. They start to fall back on old habits. Once people start to fall back on old habits, at least in hospitality in the hotel industry, you actually start to see a decline, right? They it becomes pad. it becomes boring once we fit them into the filter bubbles. It's really difficult to get people out. So, we have to constantly be thinking about, okay, where are other areas we can target and directions we can drive people to so that we keep it fresh without it becoming irrelevant. Um overpersonalization also can lead it's similar to the the bubbles. It can lead to uh limited exposure to new ideas and options that hinders engagement. That means they start to ignore the the um offers and and messages that we start to push through specifically for them. Um satisfaction becomes like they say here hedonistic and it diminishes those experiential highs that we get from finding new things to do. Right? We have hotels all over the world, some of which are in strange places. We have a a resort in Bulgaria, absolutely gorgeous, mindblowing resort. Uh very definitely geared towards golfers. I'm willing to say 90% of the population or more never says, "Hey, I want to go to Bulgaria." Right? Um so if we don't consider those sorts of things as well, um that overpersonalization means they never even consider those sorts of trips. So for us, it's very important to break people out of that process. Uh you only get one chance. Trust erosion for us is a huge issue. Um especially since when people increasingly you're seeing people book um through a chatbot through an agent whether that be in the search bar or an actual chat and they're very good but they can also cause a lot of problems if you mess it up once um and you don't have some sort of human intervention to correct. Um that's it right. We are a high trust industry. This is probably less so if you're buying potato chips, but certainly the more you're committing dollars to something um the more experiential that thing is, the the the likelihood of being forgiven goes down pretty radically. So again, we have to pay very close attention to this to get it absolutely right. Um the other one simply and I think again we all know this it's oversaturation. Too many offers becomes annoying. Uh it becomes unbelievable. That breaks down trust. It breaks down relevancy. We can target right and and we can target things and get them out there. But you know it's one of those those things like well as an example you look at any of the Microsoft products we're using today. um AI is on the verge of becoming the next clippy, right? It becomes intrusive. The same can be said when we're talking about uh personalization and hyperargeting for people. Keeping it dialed back can actually build excitement and anticipation. So oversaturation is something that we obviously look out for um pretty closely. Um the other thing is we've discovered certainly that um just because the system can identify correlation doesn't mean that we should act on that. I think everybody's familiar with the old story about you know that convenience stores they they discovered that uh men bought beer and diapers roughly at the same time. They were usually shopping late at night. That's great. Um, but if we start to make connections between data points, um, and AI will frequently give us things that are frankly a little spurious, um, we need to be very careful about how we react to those things so that we aren't, um, well, frankly, so we're not doing just stupid ideas and stupid offers to people, right? So, we need to know our customer very well. With that, I'll talk a little bit about avoiding the creepiness. So, the pitfalls are there. They're fairly obvious. Um, for us, we've had to spend a lot of time working through this over the last year. Um, because we we are driven by loyalty in the industry. We have a lot of competition around cost because of the OTAAS, the the Airbnbs and the Hotels.com of the world. We make our money through direct booking, which means all of these these moments to target people very very closely individually. Um, becomes extremely important to us. But it's also, as we've heard from people early on, hey man, it's getting a little weird, right? Stop it. So, how do we do some of that? We focus, as I mentioned at the outset, very heavily on context and trying to find areas of influence um that are driven by various things like the environment, right? Um what what kind of a place are we going to? If you're booking a hotel in Cleveland in January, it's radically different than if you're booking the Bahamas in January right? um looking then through those conditions that might be pushing people into a certain direction or another, understanding those needs. We gather a lot of information about our guests at uh the point of check-in for instance so that we can catalog that and keep that so that we know that you know in well like in my case um if I show up in warm weather uh you know yeah I'm going to probably drink a glass of bourbon at the bar but I'm going to switch to scotch and cold weather seems like a trivial thing but it's actually extremely important for building those relationships. Um similarly my wife is somewhat disabled. So remembering those things and tying connections to why that might be relevant various things we might need that's extremely important for us as well. And finally looking at what is the actual desire. So point here is that context of use really refers to these specific elements that we try to pinpoint um and understanding when, where, how, why, all of that something's being used and how we might be able to make not only benefit ourselves but obviously benefit that guest. So we do still start to build on loyalty. But again, you want to make sure it doesn't get too personal. give you an example. We'll go through a few of these. Um, a product I worked on many years ago was something called Vivitrol. It's a uh opiate suppression drug. It's once a month. It's a shot for 30 days roughly. You basically you're not it's you're literally unable to get high. Obvious, oddly enough, it actually works for alcohol as well. Um, there's a lot of problems though with this, right? Um, for the people taking it, it has an awful lot of weird side effects that people are embarrassed to talk about, but they do need help with, like constipation, things like that. Everybody's too embarrassed to talk about this, right? The other part of this is the target audience is not actually the guy on the left necessarily. During the opioid crisis, it was more frequently happening happening amongst uh men and women and slightly skewing more towards women uh in the suburbs up, you know, middle class and upper middle class folks um from about 40 to 60 years old. There's a very different need that's going on there and there's a different kinds of embarrassment now that come with it. People feel ashamed about this. Um, on top of that, it's the person you're targeting actually is not the addict. Most people have somewhere between three and six relapses before they actually get well. The caregiver, for lack of a better term, changes at each stage of those, right? So, you see people coming in and coming out. So, you're actually targeting people at different times with different needs, different expectations. So, this is really about those interactions. As you can imagine, that becomes very very difficult to navigate without um without it being intrusive or insensitive, right? So, you don't want it to become too salesy. So, the more we know about who it is, at what stage they are in the in the um recovery journey, we can change the language, we can change that. And you can start to tease those sorts of things out because right people will come in, you can ask them questions and out of that not only are you satisfying the need in the moment, but you can then start to target them down the road and actually make um adjacent industries like I mentioned constipation, things like that. You can now start to tailor um you can start to tailor messaging around those things as well. Uh another example we discovered early on actually um what AI could do which was a little bit dangerous. So we discovered that AI was able to pull certain they was able to make certain connections that we would have not necessarily thought of. Right? And this one was specifically around Lent. Um, and we were able to start to push certain offers to people, Catholic folks, um, during Lent to on Fridays to ride to do fish. Okay, that's great that we're able to nail this down. We're able to do things like target based on probability of going to a traditional Latin mass, probability of frequency of um, church attendance, yada yada yada. The problem is that gets really invasive really fast, right? So the language then becomes very well first of all the language becomes difficult. How do you extract information to see if people are even open to this idea? Um but also it starts to get really creepy because you get into that privacy issue. You know too much about me without me volunteering that up to you. Um at which point you know again it starts to make assumptions. So for our position we did have someone was pushing this idea. Hey, let's target people with certain things. Only to say, well, hold on. This probably doesn't work well for individualized targeting, right, with messaging. It may work okay for certain areas focused more on a broad marketing campaign or something to references, but we're actually going to build this to tailor to somebody, we're going to get a lot of push back, right? Um AI made a great suggestion made right on the on the surface. Um, turned out it was a really bad decision though. had had we gone with it if you start to think about right all of those contextual pieces around the individual, the culture, all of that. Similar kind of thing. We have obviously a lot of um gyms in our hotels. We have a lot of fitnessoriented things and those wellness is the big push in the industry for the last few years. Fitness centers are part of that. uh Bahamar, one of our properties, it's a huge uh resort in the Bahamas, has an amazing gym, has trainers there, all sorts of things. So, as you go through that, some of the things you notice are um outside of that of our particular industry, but across the board, if you're looking at things like fitness, we see a spike in membership in January, right? Um I used to do work for Planet Fitness, and this is one of those areas where it came up. See that spike in January? We all know that's going to happen January 1st. We do our resolutions. Uh and then you start to see people cancelling in February, March, April. Um because they go in with unrealistic expectations, right? So we can and should to a certain degree target people for things like um classes because it's a social thing. It helps keep people motivated. They engage. But even then you start to see a drop off where you for us where you started to see a real benefit was to say hold on we know it's a social thing. We know that so and so only shows up on the weekends or and they stay 2 hours that tells us right that they aren't just working out. If they showed up every day and worked out two hours we know they're a powerlifter right or a bodybuilder or something like that. We've got people showing up a couple days a week probably talking a lot. Um, we know some of the behavior moving around the gym based on field work that we've done and the data we're collecting. Uh, so how do we keep them here? Well, Tuesdays are the slowest day. Um, let's stop worrying about telling people how to get jacked, how to, you know, build muscle. Um, let's make it a social event. Give them pizza. Come work out. You get free pizza. You get a couple of slices. Everybody's happy. And we can start to say based on the fact that you are only there a couple of days a week, not five, you're probably a likely candidate to target with this messaging to get you in and treat it as a social event. That worked out very well. From that we started to build out monthly parties, right? There's all sorts of things from that that you can do. Um that actually though we also use as a way of doing behavior change. This is an opportunity to get people in. That's a point since it's a social event, we could start to have them talk to our instructors, right? So, you start to build and in fact you do get them to start coming to the gym more frequently. So, right, starting out with this simple activity builds into other activities and then brings them in more often. Okay. Um, I think the other one for us obviously specifically to hotels, this is where it goes back to some of what I was talking about at the outset. Um, American culture in particular is notoriously bad about vacation time, right? I'm sure we're all guilty of this. I had to look the other day. It's June. I have 130 hours of PTO I need to blow out before the end of the year. Ain't going to happen, right? And a lot of us are in a similar sort of situation. Uh, we work ourselves to death, right? And consequently, when we take trips, and there's some changes here based on age, obviously, and a number of factors, but for the most part, um, we feel a need to really justify why we're doing this, right? Yes, it's my vacation, but then we throw in when we're talking to colleagues with, "Boy, I don't know if I can get away for that long, and you know, I'll I'll try to not check email, but I probably will." Haha, we do that. We don't take little moments and we're not going to change that behavior. We know that's just a cultural thing we can't change, right? What we can do though is start to say, okay, how do we give little incremental kicks to people to allow them to do more vacation time and to spend those those days? So, let's tie it to business trips. We know certain things about business trips. We know the amount of time they they last typically. We know that um the types of hotels of in our case where people are staying for those um we can start to then align it with probability based on age and title which is typically in the reservation as well. Do they have kids? Are they an executive? Right? All of these sorts of things. Taking all of that, we can then start to do things a discount to say, "Hey, you know what? uh without being too pushy. Instead of saying something like bring your bring your spouse along, we can do that. But instead, it's makes more sense frequently for us to target something to say, hey, you're in Chicago and it's, you know, spring, whatever. Uh if you stay an extra day, you can get reservations for two at half off at such and such a restaurant, be that in the hotel or another one. So, you can start to actually tailor this around the partnerships as well. It feels less creepy at that point um because we're not pushing it so hard. We're tying it subtly to your business trip, but being able to push new ideas to you. Um based on that, we can also say, right, we know the type of hotel you're staying at, depending on the time of year. um we can target you then for the 6 months out or a year out to say, "Hey, um we see you like to be right in a place that's that has a spa or whatever it is when you come visit us in spring. Um maybe you ought to be looking at this, maybe you ought to look at that." Um point being, there's a number of areas in that we can start to target, but it's really about subtlety. AI will get us a great deal of information about the data that we know that then allows us the idea being to build hypotheses around um again our industry is a little bit unique in that we do gather a lot of information about people that's volunteered um the catch again for us and I think this holds true for everybody across the board is how do you do this without getting weird uh or offensive. So with that um few points I guess to summarize all of this. First of all as we all know data is not enough unless you tend to think about data as in the traditional original sense of information rather than just quantifiable elements. Right? Data points things are not enough. It's about finding out that system in which they work and designing your process and your how you're tailoring all of those those uh messages to different things throughout the year for people with that too. Uh AI, you know, as I mentioned at the beginning, look, it's it's it's here to stay. I have no problem with it. The lites out there, I think you're missing the point. It's great. It streamlines our process. It allows us to do more rapid experimental work as we go along. Um, but it's not going to replace any of us anytime soon. I will say then, you know, we debate's been going on. Is it going to mean trying to um get in entry- level people that we could train into more strategic thinkers? Is that going to be impacted? Yeah, it will. Uh, and I'm not sure, frankly, how we get around that. But I think in the long run the key point is AI is actually very helpful provided we stay in control of understanding the subtleties and the cues um that are really tied to little elements that'll that'll make people happy or make them angry. Um aligning user context as I said before we're all hide about context context context. aligning that with the business needs is probably more important than it's ever been. In our case, again, we have a ton of competition even to the point of, you know, my my god, you can go into chat GBT and say, "Build me an itinerary for a two week vacation in the Alps and it'll do a great job of it, right?" So from our perspective, not only do we have to align those business deeds um with context, but we need to think about how we're going to change people's behavior through time and how every small personal interaction can actually nudge them in a direction that we hope that they'll they'll go again to remain loyal to Hyatt above all else. Um, like I just mentioned, it's really for us about seeking that behavior change through time, not just reactionary impulses. I think that's a frequently what happens with this with a hyperpersonalization, it becomes very much of a knee-jerk reaction. That's great, I guess, to a point. If you're selling maybe cold medicine, sure. But for more complex interactions, uh, more complex businesses, I don't think that that's necessarily helpful. So for us, it really is about directing people and reacting to them in a positive way. Uh and again, you know, at the end of the day, it's sort of that apply the golden rule to do unto others. Before you act, step back and think, okay, how would I react uh if I were trying to, you know, think about where I want to go for vacation and I'm bombarded with 25 messages that know, you know, my middle name and the high school I went to and all of that sort of thing. if it would upset you, it's going to upset them as well. So, always step back and put yourself back into the loop of things. Okay? And with that, I know we got a bunch of questions, comments. Uh, so I'll just open that up. Um, and I'll try to go through some of these. Let's see. Um, I'm glad it was interesting for the person who said that. I appreciate that. You're free to tell me I'm wrong, by the way, everybody. I'm good with that. Um, the pizza post-workout was a great idea, actually. Um, got a lot of energy and a lot of carbs that come out of that. So, one thing I didn't mention, but I think also is important, is the speed at which the technology is changing. I our team well they didn't necessarily like this but I insisted that as we were starting to use more AI tools one group was working with claude one's using chat GPT and the other one's using copilot we're discovering they're all good for different things but it's not all the same no one's going to be a master of all of those god knows I'm not um my suggestion to everyone then on the team is you're going to become an expert in that one area. Learn how to write good prompts and always remember that's that that the AI is going to lie to you to make you happy, which we see a lot. There's a lot of a lot of hallucinations. So, uh just taking a look at some of those look like most folks had comments. Hopefully, uh you know, we've got a little extra time. Didn't know if there were any questions, but it would open that up. Don't know if Brett or Bill you had anything you wanted to throw in. >> Yeah, I think the what stuck out to me is the framework in which you're using. I think AI is still the wild west in a lot of different uh 500 companies and agencies. So it seems like at high you guys have actually developed a framework of if this then this but also don't overutilize AI, don't overutilize personalization, customization. You said it's creepy, it's big brother watching, it's uh all those aspects. So it was interesting where you guys are leaning in versus not to customize the experience and personalize it but not overindulge and over uh uh overutilize the data that you have. Right. >> Yeah. And I think you know we and I'll be honest we we did this we found this out the hard way. Um, and I think a lot of companies are finding it out the hard way, right? That we had the idea initially within the search function, right? That the people might come if we if we built an AI and allowed people, for instance, to go in there and say, "Hey, uh, I want to do I want to learn how to surf, but I don't want waves that are too high, and I like water warm, blah, blah, blah." Well, guess what? Nobody does that. No one goes to a hotel site to plan all of that out. Um, you know, back ages ago, I remember working for Kelloggs and they wanted their website to be the go-to for all things health food related. When you make cereal, man, you're not right. You come on, wake up. And I think it was a similar thing. We discovered very quickly that, oh, hold on. We're making some assumptions here. We need to dial it back. So we need to have a very clear plan moving forward of how we're going to use it in what capacity. Um and even to the point of little things like you know big debates going on about synthetic users I for my team at least for the time being no it's an absolute zero we will not do that it's not the same you end up with a very generic response consequently right um but I think similarly yeah it's I would you know again I do you're right it's the wild west right you go on to LinkedIn and everybody's an AI expert which means no one's an expert Um, and we're all I still think trying to figure it out. The most important thing is putting a plan in place. And for this especially, really thinking through okay how how intrusive can we be? Should we be? How personal should we be? Right? I don't I don't necessarily want my hotel company knowing, right, my favorite color, so to speak. So I think it's a for anybody moving into this era area with AI as the backbone which is becoming the foundational for it. Um before you act on anything get a really clear plan in place. When is the human in the loop? How frequently can we do this? How frequently should we do this? Um what's going to offend? What's not going to offend? Uh and and again like I said know whether or not you truly trust your uh the the data that your AI is providing you. >> It it will lie to you. >> I I think what also resonated was the aspect of you know we have all this technology and that's great. Um but you still have your full-blown your your technical and and research methodology and your methods that you fall back on, right? And there is no silver bullet. As you went through the kind of history of what research is like ethnographies or eyetracking or all these different things that have been developed really stuck out. I loved um the aspect of you know personas right and personas were such journey mapping personas were huge. jobs to be done were huge versus >> now site blueprinting, right? And during, you know, love the example of of grocery list was, you know, I had a research for Walmart for a good time and uh we develop missions versus journeys or personas, right? Because >> right mission is today, Gavin, is completely different when you're shopping for your wife's birthday or your wedding anniversary or your, you know, your weekly food uh grocery list, right? And those different missions create a different >> avatar persona or archetype uh that you fit within that that time frame. And it seems like you guys have scaled that um in in a very mature process at Hyatt of where you fit in. So love that graphic and and the infographic that you attached to it, right? >> Yeah. You know, it's and I will say it's funny too is I always remind people of this. We're a global company and that throws a whole big wrench into this as well right? um what and you know even in the United States the folks in Maine aren't the same as the folks in in Alabama right um there's different expectations there's different ways of communicating there's all of these things that really add up and matter and then now from that move out to South America to southern Europe northern Europe China broadly speaking versus you know Macau it's like yeah man you need to be very cognizant of all the details. Um, you know, AI is probably not going to be doing a very good job of understanding, as an example, that folks in China, like in our industry, um, for a wedding, if they've got the money, we'll book it in Hong Kong or Macau and it's 25, 30 people and it's a week and a half. >> Wow. That right. That's like how do we tailor that experience for you knowing you are about to drop a ton of money for this wedding? Um and you got a million choices. We're not right the only set of luxury or upscale hotels out there. So it gets it there are there are cultural nuances we have to be very very careful with as we start to promote things to people and really tailor it for them. >> Yeah. It's less translation, more localization, right? and and that >> involves the cultural norms and all that but >> fascinating in that aspect and I'm sure we can spend a lot more time there, but I guess it's more you guys always concrete yourselves in the user problem or meeting the user where they're at versus a generic cookie cutter, hey, this is what we're pushing out to the international uh uh space and and you know, we need to try to almost uh you know, white glove this where it's it fits everywhere, right? >> Yeah. I think two the other things I had mentioned this is not going to be true for all products certainly or services but like in our case like I said right somebody somebody may be going into the website two three times before they actually book um but it's usually not one person involved if it's a business trip it's you me and the computer right that's it when we're talking about things like vacation planning um typically somebody's respons responsible for the actual booking. They're typically looking on a big screen, right? Desktop, laptop, whatever. But you've also got people who they're saying, "Hey, go look at this thing on your phone. Go do this." And there's a lot of communication then that goes back and forth, right? Uh we also know that the further the distance from your home, the longer it takes for you to get through the booking funnel. I'm more inclined to book quickly relatively going well I'm based out of Kansas City so going from here to London I'm gonna book more easily than if I'm going from here to Saudi Arabia for instance right we know that that's a process and more people are involved in that so becomes the question am I actually the right person to target just because I'm clicking this doesn't mean I'm the one who needs the nudge right if if you design for instance the app to say yes I'm the primary account holder but here are other people in it well maybe it makes more sense to send that message after right be triggered by you know I'm in there looking at it now do you send that personalized message to my wife yeah that probably actually makes more sense than sending it to me >> yeah relevant topic and and there's a question um you know it sounds like in terms of you know hearing that but sounds like you're using feedback loops to know uh you've hit the creep factor. Uh what are the feedback loops or social listening avenues that you're using to get to that feedback? Uh what are the feedback loops? Uh yeah, what are the feedback loops or social listening avenues you're using to get to that feedback? >> So we have some limitations as you can imagine. Um we do actually though a lot of outreach to people as well um throughout the booking journey and especially during the stay and the followup. We have a number of tools that we've built internally. They just don't exist out there. So we that's why we have a team doing a lot of enterprise work to pull that out. Um but we are also looking at things like we can align a person say if they're out chatting in Reddit with what they have to say bring that back to their their loyalty number right and we can start to to learn from that and then tailor accordingly. Um, we with that because we do a lot of outreach, we also incentivize people through points. Points are cheap, right? Your loyalty points, contrary to what people might tell you, it's super cheap for all of the industry, airlines as well, doesn't cost as much. So, having people reaching out to them during or after the stay, and we know when to target them, like a couple of days after, not a week after, not right after. um we can start to ask those questions as well and do satisfaction surveys. Um but we're discovering more relevant is to talk to people just prior to the state and just after the state directly via chatbot. So we can program it to ask those questions of what worked, what didn't. And then we're not that we're not that blatant. We're not that crude about it. Um but we try to set it up such that people can give us continuous feedback. um and then we can respond to it and kind of build that into the model for that individual going forward. >> Got it. >> Hopefully that answered the question. >> Yeah, absolutely. Um looks like that's all we got in the chat. Uh I do see that we have our next presenter on. Um so Gavin, appreciate your time. Uh I think a lot of people found a lot of value in and thoughtprovoking uh relevant content as you can see in the chat that uh you know this presentation today gave us. So, appreciate the time. Appreciate you joining and um great topic. >> Happy to do it. Thanks, guys. >> Awesome. Uh we should be right back. Uh I believe Bill is is going to introduce the next uh talk, but you know, I appreciate the time, everyone. I think we're uh set to start at 1 pm. So, take the next couple minutes and um you know, do what you need to do, but we'll we'll start up again.