Latest / Suite Spot: A Hotel Marketing Podcast / 162 – AI & Guest Review Response: Latest Developments with Jason Lee
Transcript
- Ryan Embree: Welcome to Suite Spot, where hoteliers check in, and we check out what's trending in hotel marketing. I'm your host, Ryan Embree. Welcome to another episode of The Suite Spot. As you can see, if you're watching us on our YouTube channel, we are here in our Suite Spot podcast studio in Maitland, Florida, here with CTO, Jason Lee at Travel Media Group. Jason, thank you so much for being back on the Suite Spot. Jason Lee: Yeah, thanks for having me. In the seated position. Ryan Embree: Yes. I feel like we should be celebrating maybe like a 10th or 15th milestone of you coming on the Suite Spot or something like that. It's been a lot, but we're here to talk about a really important topic, something that's been bubbling up in conferences, conversations that we're having something that you're not a stranger to. You've been doing some research on this and, and have some updates for us. But AI, I mean, it seems to be the, obviously the, the hot topic, like I said, at these conferences or just in the industry, outside of the industry in general, and then pairing it with guest review response, something that Travel Media group obviously known for. We're coming up on the 2 million guest review mark. Which Is a fantastic milestone, really exciting milestone, but I figured we could start the episode by just kind of talking about how Travel Media Group got into guest review response. A lot of the stuff and solutions that we've talked about on this podcast have been through hotel your feedback. Talk to us a little bit about this journey in review response, and then we'll get to maybe where the state of the industry is now, and AI and guest review response. Jason Lee: Well, you know, working with hotels the way we have for a lot of years, almost 10 years ago, we were getting a lot of requests to do review response. And so, we're like, yeah, I think we can do it. And then as we got into it, we realized like, wow, this is complicated because every hotel's different. The vibe of the hotel is different, but also the amenities and policies, the quantities, the types of sites that are, have more text or less text ratings and like, I think even getting deeper into, you know, how a hotel wants to use language. And so like, it's even beyond voice, you know? So it's like it's literally grammar and also types of words. And so I think, so we started in this very simple place, very innocent, like, we can do this. This is gonna be awesome. And I think the first offering we had was 20 reviews. It was an SLA of 20 reviews a month. It was TripAdvisor only and <laugh>. So we, like, we, we jumped to that and quickly, almost like instantly it was like, well, we need, we definitely need Google, and well, we definitely need Expedia. We definitely need booking. You know, so we had to adapt. And it went, the adoption was not just in the sites that we've brought on, it was also in the complexity of the response. It was also in then systematically creating a flow. So the review comes in, we have a writer that writes a response. That response goes to the property. So like, that stuff wasn't initially there. It goes to the property, the property can edit it, and then, and then it goes live back on the site. So that whole process was also sort of formulated during those first, that first year, and then has become very perfected over time where, where we've taken that and we have all sorts of mechanisms to make sure we capture hotel voice, make sure we're doing the right things, critical review pieces that we're doing. So all of these things to really sort of make it as authentic as possible. So the guest, when the guest gets that response, it's like, this is from this hotel. Ryan Embree: Yeah. I mean, a lot of people will ask with the review response, how can you talk so much about a subject? I think we've done over 10 webinars just talking about the nuances that go into a review response, because it is a feeling, right? Which is something we're gonna talk about today. Because when people talk about AI, they talk typically about efficiencies and talking about streamlining processes. But really when it comes to guest review response, the end outcome, and you've talked about this before, is you want that guest to feel heard, and you want them to feel resolved If they have an issue, you want to show and show gratitude if they're giving you a positive review. Any hotelier that's listening to this that has responded to a review, knows the effort that it takes to respond to a review. You talked about in a video on your LinkedIn about the response sandwich and how, talk to us a little bit about that. I don't wanna give that away too much. Jason Lee: The, so the, the, the response sandwiches that was this sort of traditional way that we have of creating a response, and it has to do with the intro of a response, the kind of middle of the sandwich being all of the issues of the review, and then the, the bottom layer of bread being the sort of outro, you know, how do we, how do we sign off of this? And, you know, I think kind of going to what you're saying, you know, inside of that, we get into a much more complex version of that when we start to see guest emotion. So we start to like really understand where the guest is coming from. And I think if you look at like, why I do guest, why I would respond to reviews at all, why I would even engage in the activity, I engage in the activity, I feel like, number one, as a guest retention strategy, whether that's positive or negative review, I feel like it's a very strong guest retention strategy to, to give a great response back to, to the guest because the guest doesn't experience it publicly. The guest is experiencing it through an email that they get back from the site that they wrote the review from. So they, they're experiencing that like directly from you. So I think you start there, it resonates with, with, with potential guests as well. So if you think of the review responses, like these two major pillars of being guest retention, guest acquisition, right? In that order, you're gonna have a successful review response strategy. Ryan Embree: So talk to us a little bit about when you first started kind of thinking about AI and where that might fit in with review response. Maybe, maybe just that story I think is a, is a really good use case for, 'cause I'm sure there's a lot of hotels listening to this right now, being like exploring chat GPT and going, well, maybe this can help me respond to reviews or do some of these daily tasks that, that I, I'm told to do. So talk to us a little bit about your journey. Well, Jason Lee: What's interesting is that we started, we started, we had a project in, in the works for a long time. This was early days this, so this, we, this is maybe going back four years. And so we had something called predictive response that was inside of, inside of our development strategy. And the deeper we got into predictive response, and this was using, machine learning and also using like traditional AI to create responses, but also sort of to like help our writers, create like initial responses. And what we found immediately, right, so like as we started that is that we would have to create a model, a individual AI model for each hotel because of the complexity of a locations, amenities, policies, you know, voice. So that stuff immediately was like, we, hey, we're gonna have to do this. Then we, we then kind of pushed to the next level, which, which as generative AI started really coming out, we're like, whoa, you know, this is, this is pretty crazy. You know, this is really interesting output for minimal input. And then as we started kind of pushing in like the early days of, of ChatGPT, and I think the first version that we did might've been, might've been two Oh, turbo or something. Anyway, it was like a very early version of it. And, and we were getting, you know, 60% like accuracy <laugh>, you know, at the, from that one, but it still wasn't even close, and it still didn't satisfy these other pieces and the main pieces to that being that sort of differentiation between locations and voice. Ryan Embree: Yeah. So talk to us a little bit. I, I think those, those learning lessons from the beginning are important because I think it's, it's really stuff that hoteliers should be looking out for. And we're gonna kind of do a questions you should ask and, you know, maybe reading some of these responses. But, you know, what were some of those early lessons that you saw maybe where AI wasn't hitting the mark? Jason Lee: I think in generative ai, the, the thing to really think about is that pretty much all output is based on input. So, so that, that is this really important flow. So input being this sort of like formulated, review. And a lot of people write reviews that way. So if you look at reviews, let's say, you know, a big chunk of them are going to have a very similar kind of cadence to them. And so that's why we have things like the review sandwich when we started off is that you could see a lot of similarity between how, how people wrote they had the good and bad of their stay or the all bad of their stay or the all good of their stay. But the input is really important in this case because especially in sites like Google, actually, I shouldn't say that. All sites have a wide open input. So a guest can literally write anything about anything. You know, they could, they could write a whole, like, we've gotten TripAdvisor reviews about a per, like, it's like a life story almost, you know, leading up to their stay. And so, generative AI, even today, even the most complex version of it today has real trouble with that. It has trouble knowing what to say to it. And so it, it starts to empathize with the guest and talk about their life back to them, right? <laugh>, so, so you can't have that then. And then you have other things that are like short sentences, or sarcasm where it also sort of has a little bit of trouble. And so, so if you think about it in terms of messaging, it becomes really important. I if you take this casual approach to, to review response, which we don't, by the way, <laugh>, but if you take this casual approach to it, you're just like, yeah, you know, as long as it's done, it's done right. It's that binary thing of it's done, it not done. We don't look at it like that. We look at it like, Hey, this is communication with this guest. This is, we think of it as we're standing in front of them talking to them. So that has to be authentic. And the second that it's not authentic, I think the, the reason to do it at all gets really fit. Ryan Embree: And what you're, what we, what we talked about it's easier. It's gonna be easier to spot. And, you know, we were talking about off camera, how those first AI, social media images that you would see on your feed, there was a different sentiment there of like, oh wow, this is cool. How did they do this kind of thinking about, and now you can spot it from a mile away, at least I can, as far as the AI images go, that have been generated from a specific platform. When does that shift of sentiment start to turn about saying like, that this, this is cool and innovative to being like, well, they just threw this into a, a chat GBT, or they threw it into another AI image generator and then threw it on their feed. Maybe the, the care and effort weren't, weren't there. And in a industry where care and service are a priority, that is never the, that is never essentially the sentiment that you wanna show to your customer. You know, and, and talking about kind of in that same vein with, with AI and efficiency, what would be some questions you ask, because we're gonna talk about it in a second. TMG has actually created a smart response and, and that input output is so important because there's gonna be a lot of people right now that see this as a gap that responses aren't being done. And come to your hotel or come to your portfolio hotels and say, well, hey, we can, we can do this for you. It's a low price, it'll be done, don't worry about it. What would you, if you're a hotel, you're on the other side, what are some questions you're asking to make sure that this is the right fit for you? Jason Lee: Well, you know, I think number one, okay, first of all, I think there's a lot of, there's a lot of like shapes and sizes makes and models of AI response, right? I think there's a version of AI response where it's a do it yourself, you know, where you go in, you see the response, you can edit it, and then you have to publish it, right? And then there's another version that I think is the full pass through. So it's the response all the way through to the published. So, I think in both cases, obviously you wanna make sure that all of your hotel details, and I'm talking all, all of your amenities, all of your policies, and anything that's happening currently, renovations, things that are closed, that that, that those things can be added into the mo the response model. Ryan Embree: Quick point on this is, or example rather. So have a property, they have a parrot in their lobby, right Name Simba, right? Okay. If that detail is not in there, that this is a parrot in the lobby, a AI response could royally mess up, what, this is it, it could start addressing it as a person, it could start addressing it. It could see the, the name Simba and think of something else. Jason Lee: It's the Lion King. Right? Ryan Embree: Exactly. So these are the details that Jason's talking about. And I think those are the little nuances that we know that would, would be silly of being like, well, why would I, why would I say that this is a parrot, which, you know, people know that this is a parrot, it's, it doesn't translate. Jason Lee: And, and so I think that those things are critical. And then the next part of that would be, is there a human being looking at at least your critical reviews, right? I would say at the bare minimum, your, your negative reviews, your one and two star, maybe a, maybe 1, 2, 3, but for sure that there's somebody seeing it before they publish live. And, and the only reason why I say that is because there is just way too much variance in terms of the input. So the, you're just, it's, and especially in negative reviews, and probably even more the, the, the one that is the most variant is TripAdvisor. Ryan Embree: And so many things can happen at a property that we've seen. I mean that we've, we're talking about slip and false, we're talking about discrimination claims from employees. I mean, these are things that would have a massive input. And, and I don't want to scare any hoteliers and, and think about just the negative, let's think about the positive, right? Someone had a really meaningful, whether it's an anniversary, whether that they were visiting a loved one in a hospital and had a really great experience because an employee went out of their way to make their day, and then they're greeted with a AI response that doesn't understand the nuance of that sentiment and feeling there. I mean, that is a missed opportunity that can do just as much damage as a negative re review could do. So again, when you have non hospitality people or you know, even, you know, entrepreneurs or vendors that understand the AI space, but they don't understand the hospitality, that's where you can get into some, some really tricky waters here. All that to say, I mean, again, at Travel Media Group, you, we've embraced, we, we've got the guest experience snapshot that we've had an episode about here, but also TMG Smart Response. Talk to us a little bit about that and kind of that story behind that development. Jason Lee: Yeah, so, so we developed TMG Smart response, you know, really thinking about maybe the more budget conscious hotel you're looking at, looking at this other kind of tier of response. So our premium review response solution really hitting all of the marks. And with a live person writing it and a live person curating it and editing it, making sure before it goes live, that all of those pieces are right. So you have this sort of two human touch review response process, which is like one of the best in the industry. And, and so then when we got to generative, AI or the AI pieces, we started looking at ways of incorporating all maybe that same kind of flow, but, but using generative AI as the writer, which we had not done previously, despite what people think, because a lot of meetings we have, they're just like, well, this is done through AI, right? And we're like, no, it's a human. So when we started that when we started this process, we were really, what we were looking at is all the things that I just talked about is like, like how much can we put in from an instruction standpoint? Like what is the inst, how does the instruction set, mimic the hotel voice, mimic the hotel policy, mimic the hotel, hotel amenities and their surroundings, and then also have that other like little variant, which would be when I'm renovating or when things are happening, that that, that the ge that a guest 100% is going to address. If it's something major, like if your elevators are shut down for two weeks, right? And everybody has to take the stairs, or even for a weekend, you're gonna hear about it. So, so not being able to respond, not not being able to respond accurately to that is a, is a crazy miss. And so what's what's interesting about this is that we think of it like the same, it's the same kind of thing. So these same notes that we, that we have in our system for live writers, it's that same concept is going in an instruction set to the generative ai. And the reason why that's so important is because it, it creates accurate, you know, thoughtful responses, but there's also a bunch of other things that we've done, to our version of, of AI response that also makes it very special. Because one of the things that anybody who's listening to this, who's ever used chatt, PT knows is that ChatGPT or copilot or any of the generative AI is that it's verbose. It's verbose. And, and, and depending on the input, it's very accurate to the input. So if, if I, let's say I, let's say I mention four things that were dirty in my room, right? <laugh>, it will, you know, in a, in a, in an untrained generative AI response, it will mention all four of those things. It'll, it'll mimic those back and have a, you know... Ryan Embree: Which is not great for, by the way, for search. Jason Lee: No, it's not. And it's weird. It's weird to the guest because it's like, why are you listing all these things out? Why wouldn't you just say we we're sorry, we're sorry that things were not clean in your room. We're sorry that you had these cleanliness issues. We're working hard to, you know, blah, blah, blah. So I think that's, that's another big piece to what we, what we looked at is, 'cause we, like you said, on the outset, we have, we've had 2 million reviews, right? We, we have, we have tons and tons of data and, and the responses that we have have created are from professional writers. So it's not like the input that we've created inside of this mo in inside of this model is from the internet, right? Where you're, where you're, where you're subject to everything, you're subject to a short response, you're subject to a template response. So, so that's where I think, you know, what we have developed is as close as you can get. And then at the end of that process, there's still a live person. So, so the hotel can still edit that response, they can still interact with it, but there's still a live person before that goes live to make sure that it goes through, that there's no anomaly in it. Ryan Embree: And that's a piece of mind that is, again, you know, we talked about some of those worst case scenarios, those PR disasters that might happen. Some, some even having to do with, again, slip and fall. You do not want that in the hands of, without a real human touch on that. So it's kind of more of a hybrid solution than anything. Alright. So for those hoteliers that might be kind of considering this and going the AI route, kind of what would you caution them as far as using either a hybrid model like TMG Smart response or even another AI review response vendor? Jason Lee: Yeah, I have a, I probably have a unique view into this because of the number of reviews that we've responded to, 'cause we're so close to, and we have so many different types of hotels that need response. And so what I, what I think is fascinating to me, still to this day, you know, all the way we're in 2025, and this is definitely a year for AI stuff, but I still feel like, what is your motivation? What is your goal? What is your why? So going back to review response, why do I do it? Am I doing it because it's a brand mandate? Am I doing it because, you know, there's QA scores or there's guest experience scores on Expedia. What is my motivation to do this thing, this, this activity? And if your motivation to do this activity is for those things that I talked about earlier, guest retention as a guest retention strategy, as a guest acquisition strategy, meaning that at the end of that funnel, when they're looking, they're gonna read my responses and see what my brand and my version of hospitality is. I think it's critical at that, at those point. I think that's critical information inside there. That's my opinion. So that's where a, that's where, where the premium version of it, the live, live version is, is where it's at. Now if you're in a, in in that situation, you're like, man, I have a ton of reviews, you know, I can't really afford to do all of this, you know, this way, but I need to get these done and I need to get them done in the best possible way. I think that then smart response or an AI response becomes, becomes viable. It's a great actual alternative. But again, like we were talking about, not all of these are, they're not all the same. This is the, there's, there's a big difference between what we're doing and what somebody who's like, Hey, there's a need here. I pasted a review into this, into this chat box and, and it spit back a passable response. Let's just go <laugh>. Ryan Embree: I think, you know, my, my suggestion would be a huge red flag is if you're talking to a AI review response vendor and there is no, the only thing they're giving you is a price. There's no questions about your property, what you're trying to accomplish, anything like that. Quantity of reviews is another great question. If there is no questions asked, that should be a huge red flag for you. I think, again, with the respondent resolve, we could be in a space, Jason, where we're talking in a year, maybe a couple years, where almost every review is going to have a response because of this new technology, which is great. And this right now where we're seeing a lot of hotels that have 50% or in the low thirties or something or something like that. Some, some hotels management companies are doing a great job. They're in the nineties, but they're all templates, right? Ryan Embree: So what type of portfolio or or or review response strategy are you looking for? Because again, I think a great way to say is those hotel management companies that have 90, 95% response rate, but they're all templates, is that better than one out of every two reviews having a true, authentic response to it? Right? And that's, that's I guess the question because what's gonna happen is that the tide is gonna rise, and if everyone's doing it, then what's gonna be, what's gonna be the differentiator for you? And if it is, if it's live people, hospitality professionals responding to your reviews and getting that authenticity and trust through review response, that is a channel and a portal to do that and to get through to guests. So really exciting stuff. Thank you for, for joining us on the suite and spot and talking about this. Again, this is a, a very hot topic right now and, and it's changing almost every single day. I think that's the other thing is to try to stay up to date as much as you can, because these models, like you were talking about, they change so frequently, but you know, as, as I was reminded by someone is we're gonna get to a space again where service is gonna be, AI is gonna put service, real service to a premium again. And I'm excited to be in that space and, and kind of see what that, that looks like. So any final thoughts before we wrap up, Jason? No, Jason Lee: Just, I think, again, I, I really wanna stress that part. I know I've been like hammering that pretty hard, but I think you gotta really assess your motivation for doing this and find the right motivation, because I think it really, it, I shouldn't say, I think we've seen that it actually matters to the guest. It matters in a retention strategy. And I, and I do feel like the better the response, you know, the better the output, but it also has, there's a lot of complexity inside of, you know, something feeling authentic there. And, and I <laugh> it's crazy because we're seeing these companies come out of the woodwork with just straight an AI response, just, you know, no touch and there's Ryan Embree: A, there's an inherent risk to that. Yeah, just like as if you were to put a, if you were to put a microphone attach to chat GPT and, and had every single guest come through the lobby and talked, talked to them without any input from, from you or your staff, you know, and we've both been behind the front desk before and know what an angry can spot an angry guest about 20 yards away from us, the way that they're walking. If you're willing to take that risk of, of getting that feedback without any input from you or, or we're hospitality professionals that have years of experience of dealing with this versus what we were talking about, some of the pitfalls of, of right now where generative AI is. Jason Lee: Yeah. And it's amazing. I mean, the technology's amazing and it's evolving, like you were saying. I I do, I feel like it's going in really incredible places. Like I said, that the, that first time we were looking at these models four years ago to today, it's an incredible advancement. And I think AI automation all of these things and I, I think we're gonna see a lot of it in reservations. I think we're gonna see a lot of it in on the, on the supply side. Ryan Embree: And that's a great point if you're gonna become more efficient, because I think that's where the FOMO is for hotels is, is we're not efficient and, and AI can create us being more efficient. Do you want to do it in a space that has such a direct connection to service and customer service for, for your guests? Jason Lee: Absolutely. Ryan Embree: Well thank you so much Jason for joining me on the Suite Spot. We'll definitely keep a close eye and ear, maybe have an update for, you know, when some of these technologies because of how fast changing it is. Thank you for listening to The Suite Spot and we'll talk to you next time. To join our loyalty program, be sure to subscribe and give us a five star rating on iTunes. Suite Spot is produced by Travel Media Group. Our editor is Brandon Bell with Cover Art by Bary Gordon. I'm your host Ryan Embree, and we hope you enjoyed your stay.