Josh Graham is the Head of Market Development for North America at Cloudbeds, a leading hospitality technology company that provides unified, cloud-based solutions for hotels and other lodging businesses. In his role, he drives market awareness and helps hoteliers adopt unified, data-driven solutions to modernize their tech stacks and improve performance. Before Cloudbeds, Josh spent over a decade in hotel operations and held senior roles at TravelClick, Amadeus, Revenue Analytics, Salesforce Travel, and FLYR for Hospitality. His combination of operational experience and tech expertise allows him to bridge hotelier needs and the adoption of innovative technology.
Real hotel AI and expensive guesswork can look identical in a demo. The difference shows up in production, and it comes down to data: AI built on fragmented systems can only guess, while AI connected to one source of truth can actually make decisions. On this episode, Kin Sio sits down with Josh Graham, Head of Market Development for North America at Cloudbeds, on The Lights On Podcast to explain how independent hotels turn unified data into a real AI advantage.
Josh's core argument flips the last 20 years of hotel tech on its head. Every major shift rewarded scale: the internet handed bookings to the OTAs, mobile rewarded the smoothest app, the cloud favored the big brands. AI breaks that pattern. It rewards context, and context lives at the property, not with Marriott, Hilton, or Booking.com. The independent hotel knows what happens after a guest walks through the door. The brands and the OTAs don't.
That advantage only pays off if the data lives in one place. A hotel in 2026 is managing distribution across 30-plus channels and competing against OTAs that spend $20 billion a year on marketing, yet most properties still run ten separate tools, ten logins, and ten versions of the truth about what's happening in the building. That fragmentation was annoying ten years ago and expensive five years ago; this year it's the thing holding AI back. Josh walks through how to audit the stack, the five questions that reveal whether a vendor's AI can act or only suggest, and why the PMS has to grow beyond the 1990s records database it was built to be.
This episode is sponsored by Lights On.
Lights On helps hotels grow revenue more consistently by managing pricing, distribution, and digital marketing together.
We help hotels identify new revenue opportunities, so they don't leave money on the table. We also manage the full revenue and marketing operation, enabling the on-the-ground team to focus on the guest experience.
If your hotel needs stronger revenue growth, visit lightson.co to learn more.
Powered by Rise25 Podcast Production Company
Kin Sio: Welcome back to the Lights On podcast. I'm Kin Sio, CEO of Lights On and your host today. On this podcast, we share stories across hospitality about building and growing hotel businesses.
Before introducing our guest Josh today, a quick sponsored message by Lights On. Lights On is the revenue and marketing team for independent hotels. We work with over 40 hotels across the US, managing more than $85 million in annual revenues for our portfolio. Lights On fixes the hidden problems that cost you revenue, such as room rates that don't work with demand, OTAs that outrank your own website, or past guests that you never hear from again. We find these opportunities and help you put together a plan to boost your revenue. Recently, we just helped a 34-room Florida resort find $400,000 in just six weeks doing all these fixes. We also helped a 13-unit Montana property. In five months, their revenue just jumped 40% ahead of last year. If you want results like this, talk to us and learn more at lightson.co.
Josh Graham, our guest today, Head of Market Development for North America at Cloudbeds. Josh started his hospitality career on the operational side from front desk to GM to now senior leadership across a few of the hospitality tech companies like TravelClick, Amadeus, and now Cloudbeds. Today, we're going to talk about what actually changes in the tech landscape for hospitality and what stubbornly stays the same for independent hoteliers when you've seen the other side of the industry from the tech provider perspective. Josh, welcome to the show.
Josh Graham: Thanks for having me on, Kin. Pleasure.
Kin Sio: Yeah. So I've been talking to lots of tech providers. They all came from the tech background, myself included. So I don't really talk to that many people who had an operational background first, then moved into the tech background. So I would love to just go back in history—how you got started in hospitality and what led you to move to the tech segment of hospitality?
Josh Graham: Yeah, it's sort of a series of odd coincidences and happenstances. Very little of it was necessarily planned.
So I was a political science major at a university here in Washington, D.C. Ended up having to take a year off, not for great reasons, but some bill reasons, let's put it that way. And ended up in Lake Tahoe and randomly got a job at Caesars. It's not there anymore, but a Caesars Palace at the front desk. I wanted to deal cards to be clear, but I was only 20 at the time.
And so I wasn't old enough to deal cards. And I loved it. It was the first kind of job that I'd had that I really liked going to work every day. And so got back to school, moved over to business, hotel management, and just kind of kept it in operations. And it was my dream to be like GM of the Waldorf Astoria was honestly where I kind of saw myself wanting to be.
And then was looking at GM roles after finishing up while I was finishing up at the Watergate Hotel, independent property here in D.C. And got told by a couple of people that they really wanted a GM with sales experience. And I've done everything in hotels almost except for sales. And a friend of mine from my Hilton days was my rep from a company called TravelClick. And she said, well, we're looking for salespeople and why not give it a shot? And so I did. And turns out I loved that even more.
Plus I didn't have to, this being the early 2000s, I didn't have to show up at a hotel six days a week, 10 hours a day wearing a suit. So that alone was enough for me to say, all right, I'll give it a shot. But the plan was always—the plan in the beginning was I was going to go back. This was going to be a detour and I was going to go back to hotels. Well, I fell in love with it so much. I very little decided to go back to operations from that sales role, kind of moved into and expanded into a series of product hybrid type roles within a company called TravelClick that was rapidly expanding at the time, doing things where I was either the bridge between sales and product or leading teams that were subject matter experts from the product team to help sellers who had a very broad portfolio.
So that's kind of how I moved into the tech side of the business. And in doing so, I sort of realized as I worked through all of the different acronyms in the stack, practically that I don't claim to have a unique skill set, but while there are a lot of people that have much deeper domain expertise than I do in one or two areas, I'm very fortunate that I have a very broad set of experience. So what that helps me do other than having an amazing network of amazing humans is, you know, in some ways I'm able to see a slightly bigger picture, you know, when it comes to a lot of hospitality technology. So long answer to a short question, but that's my story.
[5:00]
Kin Sio: Hey, that's great. That's great. So, and the transition, it makes you very unique, right? I think you are understanding yourself in some way—not many outside of the tech people ever spend time operating in a hotel, right? So when you moved over and being on the tech side of hospitality, what are some of the misconceptions that technology people had about running hotels? And after that, as a follow-up question, what are some misconceptions about the operational side that hotel people have as a wrong myth about technology?
Josh Graham: Right. Well, let me answer this. I will say this—in my not too distant history, I had a senior leader come, you know, say, you know, we need to say that our, you know, we have amazing people like you and a lot of other people on our team that have deep background in hotels. And, you know, we just say that our products are built by hoteliers for hoteliers. And I said, red flag, red flag, stop, stop.
Because hoteliers come to this business because you want to take care of people. Honestly, that's, you know, you either love that or you are not a great hotelier, right? And that's a very different skill set than most technology. And so most people that come from hospitality, not all, but most, have amazing ideas, perhaps to solve specific use cases, things they see every day. But because they're so laser focused on those use cases, they don't always see the big picture. So I actually don't think that the idea that, you know, something is built by hoteliers necessarily makes it great tech.
On the flip side, what I think that most people that come from technology don't always understand about hospitality is that, and I think it's this fundamental—and I've had to explain hospitality to folks that come from a wide variety of different industries. At TravelClick especially, we did a lot of recruiting from technology and different financial services groups. And so people came in without that baseline understanding of the industry. Two things.
One, I think people often underestimate the complexity of hospitality. People look at a spreadsheet and they go, oh, travel, that's really big, and everybody travels and I travel, so I know travel. Well, you get down into hotels and you start realizing that there's very specific stakeholder groups and there's owners and managers and brands and everybody's got, you know, slightly different KPIs and what's important to them.
And then secondly, the other thing, and big tech I think often, very often forgets this—and I just read a very interesting piece about Google with their universal commerce protocol, UCP, that they just launched at the Google I/O a couple of days ago. They're basically putting hospitality on the same rails or in the same bucket as retail. So that's going to be very interesting to see how that works.
So much of hospitality is not logical. Look, every hotel, somebody put it this way—it's not an original thought—every hotel does one thing. They rent used mattresses and toilets for the night. That's what a hotel is in its most basic stripped down form, right? So the difference between a $90 dollar a night limited service property and a $900 dollar a night luxury property is not necessarily $810 worth of stuff. It's $810 worth of perceived value.
And that perceived value varies greatly from person to person, right? And I mean, I won't bore you with personal stories that illustrate this, but I think any hotelier will immediately recognize that and think about that. And so tech tries to just put everything into neat, tidy buckets. And hospitality is messy, especially the human side of hospitality, which is the real side of it.
[10:00]
Kin Sio: Yeah. Yeah. And the perceived value, right. To your point, I think tied to that is the emotional value, right. You know, I think hardly anyone will tell us that they go on travel for functional reasons. I just have to go sleep at a place for a while. They all want to go and experience something that is beyond just the mattress for the night that they're renting.
Josh Graham: Yeah. So I will say, I don't want to start off this conversation to disagree with you, but I think there actually are a lot of people that do that. And sometimes I fall into that bucket, right? I mean, I'll be honest. I mean, I have standards. I want more than just a mattress and a toilet. But like, you know, I mean, if I'm traveling and I do a lot of this kind of travel where, you know, I have to speak at an event or something or I'm going to meet with a client. So I'm away for one night, right? And I'm probably going to be—I'm going to be on an airplane for longer than I will be awake in a hotel room.
So my checklist for what I want out of a hotel in a situation like that is a location that's close, a price point that doesn't get me in trouble with my boss, a quiet room where I can get a good night's sleep because, you know, fortunately, as someone who travels as much as me, I'm a light sleeper and, and preferably, you know, like if there's a loyalty program or it's okay, cool. But like that's it. So I actually divide this.
There's hotels that provide accommodations. And there's hotels that provide hospitality and accommodation. And, you know, I know you work with a lot of independent hotels, right? This is where independent hotels can really be special because it's easier when you're an independent hotel to provide that sort of hospitality, not just accommodation. The brands have kind of—I mean, there's a benefit to standardization. Like I know, if I stay at brand X, I get product Y, right? So there's pluses to that, I suppose. But for independent hotels, it's that opportunity to be hospitable versus just provide an accommodation.
Kin Sio: Yeah. Yeah. But I think what we find over the years working with independents is that sometimes people forget that original reason of what makes independents unique, right? Especially when you have all these brand properties, you know, competing around you, you know, people start kind of forgetting the sense of, okay, let's just, you know, really get into the competition, you know, with, you know, the pure price and whatnot, you know, all the mechanicals when the hospitable side, the experience and what makes you unique in the first place, start slipping away, right?
So, you know, which I think there is a human element part of it and also how technology, you know, when used in the correct way could help a hotelier be better at delivering that experience. I love how you talk about, you know, the big tech, right? You know, the way they think about products and solving problems, they have to be very logical, putting something in black and white so that they can actually decide certain things, right? So I think there's really that human element in between of how somebody can take that technology that could be designed scalably for many different use cases or scenarios, use that, apply that, implement that specifically for a unique experience at an independent hotel, which I, you know, just from our observations, this is where many times things go wrong.
[15:00]
So I would definitely love to, like, now pick your brain because you've been sitting on both sides of the table, right? Like, technology is great, but I think the integration side of it, meaning, you know, when a hotel or when a property, you know, let's say, adopts a new technology, switching to a new PMS. And I think, you know, now you're at Cloudbeds, you've probably seen, you know, how many hoteliers are now switching to Cloudbeds. What are the biggest, I wouldn't say, for the lack of a better term, problems, right? When it comes to integrating and implementing new technology for a hotel.
Josh Graham: Yeah. So, I mean, I think there's a couple of things. I mean, the first thing to think of, especially when you're looking at a system like a PMS, is not to underestimate the complexity of that switch, right? It is complicated. And any vendor that says, oh, no problem. Don't be cautious. Just run, right? That salesperson is either telling you what they think they want you to hear.
Like, you need to be real. You're not looking for a seamless transition. You're looking for a partner that can help you manage the inevitable bumps that will come in a transition. That's sort of thing number one, right?
Thing number two is that one of the biggest bumps is often around historical data, right? And we'll talk a little bit more about data, I'm sure, as we get into this and this idea of unified data and why that suddenly has become absolutely critical, right? But when it comes to importing historical data, yes, the idea that data is the new bacon and bacon makes everything better. But if you have all the bacon, you have all the heart attack, right? And so at some point, data—in other words, if you've spent 20 years with garbage segmentation and all of these kinds of just compounding problems and you try to bring all of that dirty data into a new system, you're actually probably worse off than if you start clean fresh, right?
Now, we'll bring in the data that you need and the kind of stuff, but I think that's kind of thing number one, right? The other thing that I think is really critical and this kind of might be a bridge into some other topics is you mentioned integrations, right? And so what I think we are starting to see, and we'll talk more about this, is integrations have been a pain point for the industry for years, right? But what we're starting to see is people looking more and more for single platforms to do more and more.
Now, I'm not saying that all-in-one solution is the right answer for everybody, the majority, like that's still an evolving question. But the idea that every hotelier can manage 15 different tech vendors is antiquated, has been proven wrong, and in an era where unified data is critical, it's just amplified.
Kin Sio: Yeah, well, before—this is the topic I really want to dive in, but before we get to that, let's actually start with some basic stuff too, right? One thing that struck me when one of our previous conversations is you mentioning the role of PMS, the property management system, is different in the past versus right now. It used to be a system of records, right? Now it's becoming more, right?
So for a hotelier who's been running a property management system, probably like for decades, they never changed it. From their perspective, they just say, okay, it is a database of reservation data, people coming in and out and all that, we record, making sure that people can check in and check out. I think I would argue that there's still many hoteliers who think that is the role of PMS, which we know that perception leaves lots of missed opportunities on the table, right?
So why don't we just start with that basis, right? Why would somebody even care about looking at the PMS? It works. Why do we have to care about it? Why do we have to change that? And from there, we can extend into the data problem, which is really the gasoline in the new technology era.
[18:00]
Josh Graham: So look, every hotel in America runs on a PMS pretty much. They're pretty standard.
So I started the industry in the late 90s. So I'm one of these folks that actually worked in a green screen DOS prompt PMS environment, both in the late 90s and in the mid 2000s, believe it or not. But look, so the PMS concept was built then, and it was to solve one problem: get the front desk off of paper, digitize the calendar, stop putting reservations in a ledger, and it did its job.
And for 20 years, that was enough. But running a hotel in 2026 has very little in common with running a hotel 25 years ago, right? You're managing distribution across 30 channels. You're competing against OTAs that are spending $20 billion with a B in marketing, right? You're personalizing guest experience across multiple channels, multiple platforms every time of the day, right?
So what hoteliers did—this is the thing—they bought tools, more tools. And we can talk about the fact that PMSs were capex. And so that was a whole thing in and of itself, right? But hoteliers bought a revenue management tool, a guest messaging tool, a reputation tool, an email tool, a website builder, right? Ten different tabs open before noon, ten vendors, ten invoices, ten support teams, you know, ten passwords, right? But more importantly, ten different versions of the truth about what's happening in your hotel, right?
Now that fragmentation was annoying 10 years ago. It was expensive five years ago. But this year, something changed and now it's fatal. So what we think here is that the PMS is history.
Now, I don't mean the PMS is bad and I don't mean the people who built it weren't smart, right? I mean, the category solved the 1990s problem. But despite what my 16-year-old thinks from the way she dresses, we're not in the 1990s anymore.
Kin Sio: And let that sink in for a moment. And I think that needs to be a wake-up call because there are many hoteliers nowadays, they still run their business as if it's back in the 90s, right? So to your point, you know, over time, there are technology advancements, you know, property owners or hoteliers, you know, they buy more solutions because there was never a time that, you know, PMS probably like, you know, in the entire tech segment that's not moving arguably as fast as many other segments, right? So, you know, lots of the solutions are still kind of stuck in the past being that system of record, right?
So any examples of what hoteliers might be missing? Like, you know, there's opportunities where when they use the PMS the right way, the modern way, that can actually put more money in their pocket.
[21:00]
Josh Graham: Yeah. So I mean, I think what's important before we get into the specific use cases, understand sort of what changed and why it changed, right? And look, let's be really clear. Right now, it's pirate season. And what do I mean by that? Right? I mean that there's gold on the water because of AI, right? And there's a lot of vendors out there that like, you know, everybody's got a boat with a skull and crossbones flag, right? But some people have real ships and other people are just dinghies with a piece of cloth, right? And so hoteliers need to figure out which one is which, right?
Because here's what changed. Here's where the difference is. Every major technology innovation of the last 20 some odd years rewarded scale, right? So let's look at this. The internet, right? You wanted all the rooms. So the OTAs won. You could go to one place, get all your rooms. Mobile, you wanted the smoothest experience. Airbnb won. Everything moved to the cloud, right? So the brands quickly got, you know, got stuff in the cloud and independent hoteliers kind of started to figure things out.
But here's what—AI doesn't reward scale. AI rewards context and context lives with properties. Context lives with operators, not the brands, not the OTAs. They don't know what happens at a property once the guest checks in, right? You could say, oh, Marriott or Hilton, the big brands, they really don't, right? Maybe they do, but the data is in some other system somewhere, right?
The property knows. So what's interesting is that with this technological revolution, right? I believe first of all, there's been no better time to be an independent hotelier than today, right? That's point number one. And point number two is that those independent hoteliers have the power because they have the context. And so I apologize. Look, we made it 15, 20 minutes before we use the term AI, but you knew it was coming, right? So anyway, so I think, so that's, I think the important thing. And what that means is that hoteliers that have their data in one place, more unified data, more context, right? Are really going to be able to leverage this new technology AI in ways that hoteliers that have disparate data across all of these different systems can't.
So the hotelier might think, well, I don't have all of these tools, but I'm not competitive, but it's not the number of tools. It's the amount of data.
Kin Sio: Yeah. Yeah. And for listeners out there who haven't heard the concept of context, right? Just a quick explanation, right? Think about it's a way of how AI can really personalize responses to you, right?
So if I talk to ChatGPT or Claude today, or, hey, I need some restaurant recommendations, right? Because AI knows about me as a person and my history, because I've been talking to these, you know, AI tools for a while. They know, you know, I live in Hawaii. I have my company's here. My family's here. I have a six-year-old daughter, right? You know, the recommendations are so personalized to me with my situation, having a young family. If I want to go out to eat, it's going to recommend things that are more like kid-friendly versus like a speakeasy bar, because I can't bring my six-year-old to a speakeasy bar, right?
So these personalizations are all based on context, which is the data that you, as a hotelier, have. And what Josh is alluding to is that when you have a property management system that is properly managed with data structure, what you understand is like, you know, imagine AI knows the pattern of the kind of guests that are coming through your door, how long they stay, are they staying as couple, family or for business travel? They know about you as a property, your amenities, your features, the city that you're in, the kind of events that run in your city, right?
This is the level of personalization that was very challenging back in the days, right? You know, Josh was talking about when big tech, right? All they built was like the big scalable, right? You know, they need to build this one tech that works in, you know, Hawaii versus Washington DC versus in France. Now with all this data, hoteliers can really use the data that they already own with the advanced technology evolution with AI to really build some really crazy personalization that make your experience pop.
[27:00]
Josh Graham: 100%. And when I think, you know, hoteliers need to understand too, and without turning this into a tech class, but to understand the difference—we'll keep it really simple—between what we call probabilistic AI and deterministic AI, right?
So probabilistic AI, basically think of it as guesses, right? When we think of generative LLMs, like it knows—like your example is perfect, right? It knows about you. It doesn't know everything about you, but it makes guesses. And the more it knows about you, the better those guesses are, right? And if you think about that in the context of not just the conversation, but imagine like, you know, you stay at a hotel and you're writing a response and then now you've got AI writing response to reviews or something like that. It knows a little bit more about you, so it can write a better response.
But if it gets a little something wrong, right? It's okay, right? If it recommends a restaurant that, you know, maybe isn't a speakeasy, but isn't perfect for a six-year-old, like you kind of figure that out and move on, right? But when we're—and that's cool, that's—AI is cool. It can do that, but that's not really solving the real problems of hotels, right? That's like, that's, you know, a nice to have, right? It makes things a little easier, smoother, but we're not—that's not like the real revolution because the real revolution requires deterministic, right? It requires knowing, okay, a reservation is either for two nights or three nights, not "either or." You're either paying $179 a night or you're paying $259 a night, not "either or depending," you know, it has to know what you sometimes refer to as the ground truth, right?
And so you can plug in or sprinkle some AI fairy dust with a cool chatbot or do some kind of stuff like that. And, you know, deterministic, it'll do some cool stuff. But if you want to think about how you use AI in your operational aspects, you need ground truth data and that only comes from unified data, right? Because you can't have some of your rate decisioning happening in a revenue management system and some of it happening in a channel manager and some of it—oh, there's that special rate that's only available in the PMS—like, no, you need one source of truth unified, right? And so I think that's the real context that goes even deeper that hoteliers can take advantage of now.
[30:00]
Kin Sio: Yeah, yeah. And, you know, just to explain that a little bit more, right? So if we remember talking about Josh and I probably like five minutes ago talking about, you know, when you have many different technology solutions, right? You know, lots of the problems that we run into operating hotels nowadays is that like, you know, where is the source of truth? Because, you know, your PMS might be, you know, saying that, you know, you have a certain rate for, you know, this room for these certain dates, but your revenue management system or your channel manager may have a different source of truth that says completely different things, right?
So when you want to use AI to improve your operational aspects to start, you know, recommending rates and whatnot, when AI solutions that you have are given two completely different sources of truth, right? AI is going to make the best guess based on the conflicting information. This is going to be when you start getting really bad results versus, to Josh's point, when you have one unified data, right? Everything, you know, this is one set of data that's not going to conflict with each other. This is the ground truth. Then when you feed this to AI, then AI is going to give you a lot better success rate when it comes to doing certain tasks.
So this is, you know, if we make it simplified, right, this is kind of the garbage in, garbage out situation. So to make AI really useful, the first step is really about not even using what AI solutions out there, not choosing between Claude or ChatGPT or whatever, right? You have to really look back at, you know, do you have really clean set of data that any solution that you have could operate on top of it? And this is the part I don't think people have been talking enough about right now, because everyone's chasing the shiny objects, right? You know, hey, AI is cool, but they are missing the part of like, okay, how are we going to make sure the data feeding into AI is accurate in the first place?
Josh Graham: 100%. And it's sort of like this idea that, you know, if you really want to effect change in a meaningful way, right now, I'm not talking about getting rid of humans, but a lot of people like to talk about, oh, AI is going to like, you know, free your front desk up from doing stuff, right, so that we can, you know, provide the real important hospitality. But let's be real, if all AI is doing is queuing things up for your front desk to approve, that's not actually saving them any time. That's just changing what they do, right? It's kind of like saying, well, I used to write things down in a notebook, and now I write things down in Excel. Okay, but then everybody just started hating Excel, right? That didn't actually make anybody's lives better, right? It just kind of changed how we worked, right?
So if you want to actually free up people's time to be more hospitable, you need to start having systems that actually make decisions. Not every decision, there's going to be thresholds, like there's reasonable sort of things here, but you need systems that actually make decisions that are deterministic. And for that you need deterministic AI, stuff that actually knows the real truth.
And so that's the kicker, I think, that a lot of people haven't really fully wrapped their arms around in the sense that, you know, there's still a lot of people out there going, oh yeah, yeah, it's just, it's not going to replace people. It's going to, you know, free people up. Okay, but what does—if you go down a level deeper, if it's going to free people up, how is it going to do that? It's going to do that by making decisions for them.
[35:00]
Kin Sio: Yep. So if I were a listener listening to this show right now, I'm so convinced, totally, I say, hey, I agree with everything that Kin and Josh have to say. What should I do now? Like, you know, I think lots of people are very confused and overwhelmed with lots of different news and all the advancements out there, right? But, you know, for someone to start making practical actions to really take advantage of the evolution, what's the first thing they do? Like, you know, they have probably a set of technology they're using already. What will be a good way to start doing an assessment on readiness? Like, you know, what can I do like today, tomorrow, the next week to start making positive impact to running the business?
Josh Graham: Well, I can tell you, the first thing you don't do is just go out and buy a bunch of AI tools so that you can tell your boss, yes, I have AI, right? Like, I mean, it's not a thing, right? And again, at Cloudbeds we pride ourselves on being a little bit of a rebel. So I'll say out loud what might actually—so I'm going to HITEC in three weeks. So hopefully I say this and I don't have people coming at me with knives, but there's a lot of vendors out there that are just like slapping an AI logo on old machine learning processes or, you know, just trying to make it seem like, you know, that what was old is new, but with AI, right?
And so there's, you know, there's a lot of pirates out there, right? So, you know, what I would suggest is that they start thinking about a couple of things. If you're talking to a vendor, right? Ask them—I mean, I can think of like five questions that you should ask every vendor, right?
Can your AI take an action on its own or can it only suggest one? If it can only suggest one, that's great, but you know, you're buying a smarter notebook, not an employee, right?
Ask them, where does the data live? Not where does it sync, but where does it live, right? If it creates a data silo, right? That's less valuable, even if it has ten more features than another platform where the data lives with all your other data, right?
Ask a vendor to show you a customer that looks at least somewhat like you, where their platform improved their P&L, not a guest satisfaction score, not something—a P&L. That's a difference maker, right?
What does that platform know about a guest after they come into the hotel? Because before is easy, right? We've got the before, there's a million ways you can know about guests before they walk in your door. The part that matters is after they walk in the door, right?
And lastly, ask if this vendor is trying to help run the hotel they have better or to help them build the hotel that they're becoming, right?
So those are five questions to ask, but like, look, we said this earlier, switching systems isn't easy. People have contracts, different things, right? And I'm not suggesting that you can take a 15, 20 vendor tech stack and then just go find a single vendor that's going to do everything you need, right? Like that's not a realistic thing.
So hotels need to start thinking about taking an honest look at their tech stack, counting the systems, counting the cost. Where can they start to consolidate? What can they consolidate this year? What can they consolidate next year? What vendors are moving towards not just a platform held together by integrations, but an actual platform with a single data model, right?
[39:00]
So, you know, if that vendor—let's just take an example. Everybody knows I work for Cloudbeds, so this isn't like a secret, right? We have guest experience, right? Guest messaging, you know, all of that before, during, after that side of things. If you can bring that in to the data model, a single data model, that's a huge win. Even if guest experience doesn't do every single feature that another system does. Or if you're going to work with another vendor, find a vendor that's dedicated to building integrations that share all of that data, right?
At the end of the day, PMS is history, but it's also that interaction layer that's there. So that's not going to change, right? A guest experience platform isn't going to suddenly start checking guests out and processing payments and printing invoices, right? So if they're not going to have that data, which they're not, right? Are they going to share their data back with the system that's doing that? And so that's just, that's a real example of like a question to ask.
Kin Sio: Yeah, yeah. And I think it's a big shift in thinking about technology and procurement, right? Because I think the past 10, 20 years, when people, when, you know, when hotel owners and hoteliers are comparing all these different technology solutions, right? The tipping point is always like the number of features, right? You know, solution A has these features, solution B has all those other fancy features, right?
So with now the current evolution of AI, the creation and development of new features are becoming so cheap. Like, you know, everybody can build some really easy features overnight, right? So that's becoming not that fancy anymore. And which to Josh's point, I think really the treasure, right? The goldmine is really about the data and back to the integration, right? You know, there's now opportunity to consolidate, to kind of build that one unified source of truth, because any other fancy features, which can so easily be built nowadays, needs to act on this one unified data.
Like this is your new currency, not about features, just worried about like, protecting the integrity of that unified data. And once you have that, like there's just so more possibilities and potential that you can do with it. And this is like, which, you know, to one of the questions that you mentioned, right? Where the data lives, because you got to make sure that you as the business owner, as the authority, the ownership of that data, because now that becomes your currency in this AI era. This is the part where any other tuning is going to be operated on top of that. And that belongs to you. No one can take it away if you treat it the right way.
Josh Graham: Well, Kin, I mean, I just thought of another sort of real life example, right? So to expand on it. So you're in Hawaii. So I imagine you have, you know, hotels you work with in Hawaii. And by virtue of geography, a hotel in Hawaii is probably talking to guests anywhere from three hours to eight hours worth of time zone away, right? If you're talking to someone on the mainland of the US, you're probably talking to them via text or something like that. Or if you're talking to a guest coming from Asia, perhaps it's on WeChat or, you know, if they're Korean, KakaoTalk, right? I think Line is Japan.
And so you've got these different things. And so you can go in and you can find a guest messaging tool that can answer, right? They can, you know, can do these things. And you're like, very cool. Now, when it's three in the morning for me, and the guest is already awake, I've got this AI that can answer. But what if that guest now asks a question—asks not just a question, like, you know, where is the best luau near the hotel, but says, can you do something for me? Or it says, you know, when is check-in? Well, check-in is at three o'clock. Well, can I get an early check-in?
Can you—this is like, if all that system can do is say, well, I need to—let me—I need to check or you need to call back, or you need to send an email. Well, okay, now. Okay, cool. What did you really do? You didn't really do anything, right? That's not an employee. That's a notebook, right? It's easier.
But the real revolution is when that messaging system can look at the check-ins and can say, okay, this guest is checking in on Friday. On Thursday, we're running 60% occupancy. So yes, I can probably—I can tell the guest coming in on Friday, that there's a good chance that they can get an early check-in. Oh, but wait, they're in a, you know, an ocean view room. Okay, well, we only have three of those. And those are all checking out on Friday. So maybe I shouldn't—like these are the kinds of ground truths that that guest experience platform now needs to know if you want that guest experience platform to actually go the extra mile and do that, right.
And so that's a real life example of the difference between deterministic and probabilistic and having that true context in one place for that system to access and actually make a decision.
[44:00]
Kin Sio: Yes. And actually, I have another real example to back it up. So some of the hotels in the portfolio, they kind of being the more early adopters, they tried all these, you know, phone agent AI, right. So, you know, you might realize sometimes when you start making a phone call to the hotel, it is an AI answering, you know, all those common questions.
So lots of those solutions right now, because they are siloed, they're not directly integrated with the PMS, right. So like, you can ask all these AI questions about amenities and things and whatnot. When you try to ask the AI to actually make some changes, or if you want to do like, let's say, early check-in, late check-out or change your room types, things like that, the AI couldn't do that for you, because it is not integrated to where, you know, the reservation data lives, right, in the PMS.
Versus the world that Josh is, you know, painting the picture about, right, you know, if it is a solution, right, I don't know if Cloudbeds has it already or maybe some other solution that's integrated to it, that the AI system can actually make changes to the reservation system, so that when somebody calls and says, hey, you know, they have all these, you know, requesting changes or even upsell, right, you know, hey, I want to request a late checkout that will cost, you know, $40, $50 more. But the guest is willing to pay it on the spot.
And if the AI now has the capabilities, and with the right data to make that action, now it becomes a, you know, an AI employee, because it actually makes changes, not just suggestions, make changes on the fly without rerouting to, you know, the human staff team. Then this is the golden path that I don't see huge adoption or any major case studies for that use case right now. But I think that should be the golden path that people, you know, hotels and tech companies really need to move to, right, because this is the key unlock of new opportunities, to give the time back to the human staff team, who can actually do a better job on site with the guests coming through the front desk.
Josh Graham: Right. And listen, I want to be really clear, I don't want to discourage people from experimenting with things. I don't want people to feel like, oh, my gosh, this is so much I might as well do nothing, because doing anything seems easier. And you're telling me, well, if I do these little things, it doesn't really like—there is some value, we just need to be really clear.
But let's just take this example, right? You know, you could have said, well, maybe 10 years ago, you would have had two people on the overnight staff. So if somebody, you know, came in or called in or whatever, like a person could answer them and walk them through, but let's be real, right? Like, you know, post COVID, post all of these things, like the last, you know, especially the last six years, it's been a meat grinder, right? So those people aren't there anymore.
So all we're doing is we're if we turn on those sorts of things, we're easing the pressure off just a little bit, right? You're giving the guests a little bit more, but it's what they would have had before, right? This isn't like a dramatic, massive change. This is just incrementally better, right? And so is there some value to that? Yes, right? But it's not the revolution. It's not the fundamental change. It's not giving, you know, again, if you do that with a guest and the guest can be like, oh, okay, cool. Like, what's next? Like, okay, like, you know, that's not going to excite them, right?
And so, you know, keeping those expectations in check while not discouraging people from experimenting. One of the things that I tell hotels, I've been saying a lot this year is that, you know, and I hear this often, well, Josh, you know, when, you know, I'm hearing all of this stuff about agentic booking, right? And like that the agents are just going to book hotels. So do I need to go buy an agentic booking engine or, or, or like, what's this MCP? I don't know. I heard about this somewhere, right?
Look, the successful hotel in 2026 is not going to be the hotel that has an MCP and agentic booking engine and all of this—no. But the successful hotel in 2027 might, right? This is all unfolding. And that's the other thing, you know, we talk a lot about MCP, but Google just unveiled in the last couple of days, universal commerce protocol. I mentioned this at the beginning, right? Where they put travel into the same rails or the same bucket as retail. So we're going to see how that plays out, right?
So, you know, it's important that hotels are staying on top of some of these things a little bit, but what is absolutely critical is that unified data, you're starting to make those steps to clean your systems so that whenever all of that noise kind of filters out, right? And we kind of move through this, then you have the context because I can't tell a hotelier that ChatGPT plugin is going to be more or less successful than, you know, Google running UCP through whatever. I don't know, I'm just making this stuff up at this point, right? But I don't know.
But what I can tell you beyond any—with a hundred percent confidence is that no matter what acronym or platform or whatever is the eventual winner or who's the next Expedia or Booking or maybe Expedia will be the next Expedia. I don't know, but hotels need to prepare for it the same way: cleaning up their data, having unified data, having that single source of ground truth. That's the critical thing.
[51:00]
So start the movement towards that, start preparing for that. And as a hotel, especially as an independent hotel, where you have that flexibility to make those kinds of decisions, you will be well prepared for 2027 and beyond. If you start making those changes and thinking about those changes to be made over the next 6, 12, 18 months.
Kin Sio: Yeah. And I think it's this hidden work that people are not talking enough about, right? So I think the key takeaway from today is that no, you're not going to be going out and trying to buy seven other AI solutions tomorrow, but instead start looking back at what you already have and do an audit of your tech stack.
We had a past episode with our hotel consultant, Dan Wacksman, based in Hawaii—a shout out to Dan—he actually mentioned about like, look, start doing a tech audit. Before long, you will find out a bunch of features and over-the-top things that you never used. Start taking that approach, start consolidating that, start building a more unified landscape of your data. Because if you do that right now, next year when better solutions, better AI stuff come around, you are going to be more prepared to ride that wave than now trying to buy seven other solutions and you have to do it again.
I think recently one of the major brands, I think it's Hyatt, announced now they're building all these new AI workflows on top of their data. People are all having discussions and conversations about that AI move. What they did not notice, the corporation spent the last few years just fixing up their data, fixing up their infrastructure to a point that they can start trying out with AI solutions for their operational improvement. So like these, honestly, years of work for a much bigger scale. If you are not doing that today, you're never going to be successful even if you buy seven other tools. So that's definitely something that anyone can take action on today, tomorrow.
Josh Graham: 100%. And Kin, you just said it again, scale. So first of all, Dan—shout out to those guys, great work. Those guys do really fabulous work. But you just said it right. Hyatt has to think about that at scale. A couple thousand plus properties, all of these different owners, all of these different things. They can't make a change without worry. It's like a Jenga block.
So they can't just swipe it clean and rebuild. You're an independent hotel or a small group of hotels. Maybe you don't have as many resources and that's when you bring in somebody like Dan or yourself to help you work through some of these things. But this is not about scale. In some cases, scale can actually make the challenge bigger and harder. Context. Context matters. And so independent hotels can leverage their lack of scale. And so that's huge.
Kin Sio: Yep. Thanks, Josh. And I do have the last question for you. We talked about OTAs a while back. OTAs definitely won big time back in the internet age. And I know Cloudbeds a while back launched a research report about now independent hotels. OTA shares of booking take up to over 60%. It's always that frenemy, necessary evil situation with independents.
[55:00]
So with everything that's changing in terms of technology, do you see a way that that landscape with the OTA dynamic is going to change? What should hoteliers start thinking about? Is there going to be still a heavy reliance situation? Is there going to be opportunities where they could do something else? What's your take?
Josh Graham: Yeah. So what's interesting, so there's two things. So one of the reasons that OTAs did so well was they were able to own the discovery phase. So let's say that whether you were just traveling for a quick business trip or whether you were traveling for the family trip of a lifetime, maybe it's somebody coming into Oahu for a single meeting or somebody taking their grandparents and their kids—taking their grandparents for the 50th wedding anniversary. The discovery process was really hard. Tons of you get to travel. OTAs made it really easy to take all of those and say, okay, I want to go to Oahu. I want to look for some parameters and I can find some things. So that made the discovery process a lot smoother. That surface—people use the word surface a lot now. And OTAs also made the transactional surface, the commerce surface really smooth. So you could very smoothly go from discovery to transaction within the same platform.
Now here's the question. Is and will LLMs, whether it's ChatGPT or Gemini or Claude or whoever it is, rapidly becoming the discovery surface or layer of choice? But choice for who? For geeks like you and me that live online and feel very comfortable with technology, right? We can go in and find—I can tell it, it knows about me and so it gives me three options. Well, are those the right three options? Are we really sure? Where's the FOMO? How's that going to play out?
And chat is not necessarily the greatest surface or interface for commerce. Again, we're selling—it can give you a lyrical description, but for years we said people don't read, because nobody takes the time to read anymore. So that was where Airbnb and their app experience was great or those kinds of things.
And what about upselling? Like I'm coming to Hawaii, I want to understand if that room, can I smell the ocean air? If I pay a little bit more, can I see the ocean? How does all of that work? And does that work in a chat interface?
So there's a really big question. And here's the thing, you asked me a question and I'm saying, I actually don't know the answer. And I would argue that anybody who tells you they know the answer is either trying to sell you something or full of stuff, right? Because nobody knows. It's all in flux. We are in that moment. It's pirate season, Kin. You don't know which way the wind's going and who's coming. Are they coming to help you out? Are they coming to kill the crew and take the gold? I don't know.
So yeah, so that's just as honest as I can be, Kin, that will the OTAs be around in five years, 10 years? We'll see. But whatever it is, you're going to want unified data. I'm going to come back to that same point. Unified data. It doesn't matter if it's an OTA or an LLM or if it's an agent talking to another agent. You're going to need unified data. So focus on that. That's my best advice.
Kin Sio: Awesome. Thank you very much, Josh. Where can people learn more about you and Cloudbeds?
Josh Graham: Well, Cloudbeds is cloudbeds.com. You can also look us up on LinkedIn. You can look me up on LinkedIn. My contact info is there or josh.graham at cloudbeds.com. So there you go.
Kin Sio: Awesome. Josh, thank you very much for joining us today. It was a super fun session. So thanks a lot for your time.
Josh Graham: My pleasure, Kin. We'll see you around.
Kin Sio: All right. See you around. Signing out.
Transform your group booking strategies with Lights On and watch your occupancy soar.
.svg)
