Stephen Kaufer

Former President & Chief Executive Officer, TRIPADVISOR INC

Reinventing hotel booking with AI - Holden, Kaufer and Vakil with Chris Hemmeter - #Phocuswright

🎥 Nov 15, 2025 📺 Phocuswright ⏱ 23m 👁 545 views

As generative and agentic AI reshape search, discovery and conversion, the rules of hotel distribution are up for debate. Two industry veterans bring decades of perspective from the front lines of travel and search, joined by an entrepreneur reimagining hotel distribution technology for the AI era. Expect a candid look at how AI could rebalance power across brands, OTAs, metasearch and new intermediaries; what “direct” really means in an AI-first funnel; and where value, margin and guest experience may shift next. Speakers Richard Holden Former General Manager, Google Travel Stephen Kaufer Fo...

Watch on YouTube

About Stephen Kaufer

In a recent interview, Stephen Kaufer discussed the early challenges of building Tripadvisor. He recalled that the company initially had a "failed business model" and described the travel industry as being "in total disarray" at the time. Kaufer also reflected on a difficult financial decision the company faced, stating that while it was "a crazy bad decision" financially, it was "absolutely the right decision" given the risks and the life-changing experience it provided for the company's employees. Kaufer also recounted the origin of the Tripadvisor idea in 1998, which came after he spent a significant amount of time online trying to research a resort in Mexico. He described the early business model of sending users to Expedia for bookings, noting that the company was not initially aware of how many of those users actually completed a booking.

Source: AI-verified profile updated from Stephen Kaufer's recent appearances. Browse all interviews →

Transcript (46 segments)

Moderator0:00

All right. Thanks everybody. Thanks for sticking around so late. I appreciate it. We've got 22 minutes and 30 seconds. I don't know how we're going to do this. We're going to try hard though. So, this is a very exciting panel for me. These are three real iconic players from the travel business. I'm going to just give you a quick overview. You know, Richard, 20 plus years at Google, vice president, general manager of Google's travel products. Basically built Google Flights, Google Hotels, long fantastic history. Now no longer with Google, retired from Google, so he can actually tell us what the hell's going on. I think that's going to be really fun. Steve, longtime builder, started building software companies in the 1980s and then founded and took to an IPO and a billion reviews TripAdvisor. The 20-year CEO of TripAdvisor, and I believe in the hall of fame, travel hall of fame of Phocuswright.

Stephen Kaufer0:57

Very exciting. I mean, super cool.

Moderator0:59

And then Sanjay, founder of his own new startup that I'm sure we'll talk a little bit about, but also long stint in product at Google and at TripAdvisor as well. Very intimidating these three. Magna Cum Laude undergrad at Princeton, Stanford MBA, Harvard engineering undergrad, and an MIT rocket scientist, PhD in aerospace engineering. So, this is a panel that confirms that just because you're smart doesn't mean you're right. So, Steve, let's start with you. Look, you've had a long career. You've lived through the disruption in travel. You've seen the arc of the digitization of travel and now here we are in the age of AI. What if anything is different about this moment?

Stephen Kaufer1:52

You know, it's a fun question. I've had a little chance to reflect on how much really has changed since I first got involved. Travel agents went to OTAs and that was really empowering for consumers. There was a tremendous amount of inventory and the consumer, the traveler could actually do stuff themselves. Then you went from, hey, there's only a handful of experts, there's guidebooks out there to the whole user-generated content revolution where the wisdom of the crowd clearly won in that transition. And then you have companies that everyone admires, the Airbnbs and their Vrbo's bringing whole new sets of inventory online. And as I go through that, I'm like, okay, you got three covering 30 years. Wow, that's not a lot of change. And it's not to say there haven't been tons of improvements along the way. And so one asks like well is now any different or is this blockchain in travel again? And I'm like this I think absolutely transformational. A friend of mine said, oh my god I went to GPT. I put in give me a 5-day itinerary for Paris and I got this amazing list. I'm like yes Arthur Frommer did that before I was born. Right. That's not the impressive AI that I think is really coming. I stood on stage here 15 years ago talking about the amazing personalization that big data companies would be able to have like never delivered. No, just honestly never at TripAdvisor, not any other company. And so playing around and just thinking a little bit of how in the next couple of years AI really, any of a number of models be it big ones or the startup that will be able to take that ability to talk in English, have a conversation and get a really good result.

Moderator4:06

Absolutely. It's going to happen. It's here and it'll be number four on that list of totally transformative things.

Stephen Kaufer4:13

I think so.

Moderator4:14

But Richard, is that enough to change the power of the aggregators to change the front door of travel, the top funnel gatekeeper and then ultimately the aggregators themselves?

Richard Holden4:28

I do think along with Steve that we are on the cusp of something interesting happening and you're right for many years we all talked about personalization, other things happening. But we are at a point where and for many years we would talk about in my group at Google that we've enabled a lot of consumers to get all the data that they possibly could ever want but we've actually burdened them more than they've ever been before in terms of planning travel. I do think now with the new capabilities that are emerging as well there's a real opportunity to simplify that for users and crystallize it. And I agree, the experiences aren't great yet but you can see a sign of them coming and what you get, the data that you get back now is transformationally different than what you were getting before. And I do think new front doors are going to be created here for sure. And from a user perspective, and actually from a supplier perspective, that's an opportunity to change the dynamic that's existed for the past 20 years. Be frank, in many ways suppliers were out competed by a lot of the OTAs in terms of maximizing the front doors that existed before. With the change of new front doors coming along, new technologies being enabled, I think that's an opportunity for suppliers again to connect with consumers more directly than they ever have before.

Moderator5:45

But how is that actually materially different than what exists now? Because at some point, yes, there'll be multiple front doors. Let's say it's ChatGPT. ChatGPT spending a trillion dollars on capex not making that much money at some point investors are going to say hey guys you got to monetize this stuff that you're doing then it's the same old same old isn't it the aggregators live at the middle they have all that power all that inventory they've got engineers they're pretty smart why is there an opportunity to do something new for supply?

Richard Holden6:14

Look, I think one of the undervalued things about what changes with these conversational interfaces is they happen in one place. They happen on ChatGPT. What you see happening today on Google, we used to talk about this. People would visit 40 different websites before they actually book their hotel. They had questions. They need to get those questions answered either about the hotel or about...

Stephen Kaufer6:36

There's a lot of data that suggests that people actually like that. Planning travel is fun.

Richard Holden6:40

They absolutely... some do, but you want to chase that in. You can help people find that information within the conversational UI and have that in one flow and not have to wonder what tab that information was on that feels part of the deep transformation to me as to what's happening. Now the best answers for that stuff are usually from the suppliers not from the intermediaries. The intermediaries have worked really hard to create a corpus of data that's very efficient to walk through and find information on, but it's pretty thin, right? If you have a dog, there's a little check box that says pet friendly and you hope that your 55 lb husky fits within their list and they're not just looking for a 12 lb Chihuahua. I don't know, that's why I go visit all these websites. That's the information that suppliers, we believe, can bring to these AI tools and provide really helpful answers to people that are tracking down carefully right now.

Moderator7:39

Steve, you look, you grew up with I mean, look at Booking, Expedia combined, right? What have you got close to 200 billion of market cap, maybe 15 billion of marketing spend being dumped into the channel, right? That will likely be repurposed and used in different ways. Massive engineering talent, huge aggregators of the long tail now developing their own loyalty cases across those two things. Isn't that the kind of inertia that's hard to dislodge with a technical revolution like this?

Stephen Kaufer8:10

It is very hard to dislodge the aggregation and the results that they can give and that they have been giving to the AI engines. Right now, they're just a natural first in. But the question is like and certainly with direct bookers there's the ability to give the AI engines exactly what they want which is instant best availability, inventory, all the extra information, maybe additional rates that the OTAs do not have. OTAs do not have every hotel. They only have the hotels they sell, which is fine if you're looking for a Hilton in a town, but what about the rest of the broad set of places that a customer might want? And I don't know many people that today raise their hand and say, no, I prefer to book on Expedia or Booking versus the hotel itself. They are great companies. They do a job that is absolutely needed to help somebody find the right hotel in San Diego and they do that very well and no individual chain or there's only a couple of other sites that even offer that service. But when you think about where it's going to evolve if OpenAI is and I'm sure they will be paying very close attention to addressing a new need in a better way than what any OTA can do today. Part of that solution is going to be enabling that user to book directly on the hotel because that's what so many users want and they're going to listen to what their customers are looking for.

Moderator9:56

But you've tried that at Google. You had Book on Google, you had direct pipes to the suppliers, you were gathering that content that Steve talked about and it didn't really work. And to the point of aggregators, I do believe that I think Booking has a very very large universe of supply that's pretty close to most if not all of it.

Richard Holden10:22

I would rephrase a little bit what we were trying to do. I spent 10 years working on travel and first couple of years into it we decided we need to get much more serious about building a great organic travel search, hotel search experience and we spent a lot of time looking at the fact that it was very OTA dominant at that time. We spent a tremendous amount of time, that's what Sanjay worked on for many years there, was bringing more hotel content, supplier direct content onto the platform and our belief was by creating a better and more balanced ecosystem, consumers would trust it more and they would engage with it more. And our data definitely suggested that consumers were much happier, trusted the data more when they had a mix of data that was there. And we didn't, we got a lot of data from hotels, but I wouldn't say that we necessarily got all of what Sanjay is talking about getting here, which are the loyalty rates, other rates that OTAs aren't showing as well over time, the amenities, other information, which again, these new models through AI are going to enable us to get that data to work with the hotels to surface that data. Google wasn't doing that at the time. Now, they may be working on that now as well, but we didn't have access to that data at that time.

Moderator11:31

So, you're all three of you share the thesis that this shift to AI as the front door will accrue to the benefit of supply. And your thesis is that that is a sustainable position to be in.

Sanjay Vakil11:49

Just to be clear, I quit my job to do this. I had a sweet gig at Google. It was great. I feel pretty strongly.

Moderator11:56

Yeah. Fundamentally though, how then do you build a business to do that, right? I mean, because you have to now go out and aggregate that supply and you have to collect information that many of them don't currently have or the data that they have is a mess. You have to somehow get to then escape velocity as a company because you're dealing in a space where it's either go big or go home. And meanwhile, you're gonna have two very interested observers in the big aggregators watching how it's going. Okay. How do you raise enough capital to do that? And how do you defend yourself as you begin to win?

Sanjay Vakil12:37

Let's put the capital aside for a moment. If we get this up and running, the money will come in and I'm confident with that. There's two interesting things we've learned in this process. One is that hotels have a latent set of content that they want guests to know about. So we're working now with, we have contracts being signed with the top six chains in the world. Okay, six of the top 10. And to a group they have turned around and said we have this giant set of marketing material that nobody wants. We don't know where to put it. How do we get it to you? And we're down to FTPing stuff back and forth. Just give me a bucket of this information. I will have my AI work over it. We'll present it and make it available to consumers. And this is stuff that is down deep in what we would used to consider longtail queries. What temperature is the hot tub? Some people are digging around trying to find...

Moderator13:32

Is it a defendable thing that you can get that information? I mean because if the hotels have that information and want to give it up can't others get it including the existing aggregators or is it a different type of architecture?

Sanjay Vakil13:44

Structured information, that's the really interesting part. This is what AI can do that's quite different than what we used to do before. You used to have 82 attributes that you went through and checked off on a list and said yes I have a spa yes I have a hot tub there was no place to put the temperature of that hot tub and it doesn't exist today.

Moderator14:01

Richard, with Gemini 3 of course now and Google's incredible Q3 earnings results, hundred billion in revenue. I mean, just an unstoppable economic powerhouse. How do they play into this? Is Google's intention to be that gatekeeper to maintain its position as the dominant gatekeeper, but then work with aggregators and direct supply? I mean, what's the how is this going to play out?

Richard Holden14:30

I'm no longer at Google, so I don't know.

Moderator14:31

You can tell the truth, baby. Come on. Spill the beans.

Richard Holden14:34

What I would say is I think that the trajectory hasn't changed from the perspective of trying to build a better and better consumer experience which in my mind when I was on the team and I don't think it's changed is that much more supplier engaged content needs to be there in addition to aggregator content and I think Google is well positioned to do that extremely well over time. No doubt about it. I felt confident of that when people were talking about AI being a demise for Google and others. I didn't believe it and I certainly don't believe it today. I think they're doing great in terms of leveraging that.

Moderator15:08

Do you think we could wind up in a world where Google is the only dominant player? Do you think players like OpenAI have a place in the ecosystem?

Richard Holden15:17

I think there's opportunities now for many front doors. As I was saying before, I do think there's a lot of opportunity for others to play in this space.

Stephen Kaufer15:26

And absolutely true, there's going to be multiple with different specialties. The specialty may not be around what the LLM is trained on. It might just be how it is presented to the user. Right now, if you're using GPT or Gemini, honestly, the interface just sucks for planning travel. But that's not to say there aren't a dozen startups looking to use those underlying capabilities to give you a travel experience that would blow what Expedia or TripAdvisor or Kayak or anyone else can do and then we'll see where it goes. Do the big guys buy the smaller guys in order to bring that to the audience. And to the question or what you might have been getting to and is OpenAI going to then charge hoteliers for the privilege of sending I have no doubt that they will be adding the ad units but they need to prove themselves better than the Googles and everyone else. So like Gemini versus OpenAI Google's over here on the side. I know Gemini is part of it, but they're going to be going at it and may the best product win. We think that will be beneficial to Direct Booker because obviously all of these folks will want the best prices and the best content and the best material to train their engines on.

Moderator16:52

Sure, quick show of hands. Who thinks that this era is going to play to the benefit of supply? Who agrees with what the panel is saying here? Who disagrees? Oh, interesting. And everybody else who didn't put up their hand like me has no clue what's gonna actually happen. Is that right? Okay, good. So, what's the hard part here then? What has to go right for your thesis to play out and turn into an enduring business?

Sanjay Vakil17:20

That's a great question. So, this industry is fascinating and frustrating at times and sometimes it fumbles opportunities when they're presented. I think that AI is an opportunity to connect guests much more directly with hotels, but it does in fact require hotels to work together a little bit and get aggregated in some fashion, which is what we're trying to do as a pipe that aggregates this information together so that people can see the entirety of the inventory. And so we need people to get on board and get us the information and help out more. One of the early conversations on the Direct Booker board level was like how easy can we make this for every single hotel chain, every single aggregator who's collecting the independents and I loved the answer as a board member which was basically look point your Google feed that you already are doing over to Direct Booker. You don't have to get in that black hole, right? That Google feed that Richard and I built, right? So we know exactly what it...

Moderator18:33

So the value proposition then to the suppliers, is it an economic one or is it a quality of information one?

Sanjay Vakil18:40

It's both. It's both. We're just a pipe. We're not an OTA. Our job is to make this a low enough number that it's a no-brainer that you want to help out here. So, we're trying to get the price point below what you're going to pay your credit card processor, right? A tenth of what you'd pay the OTAs in the same situation.

Moderator19:01

You're basically going to them saying, look, we've got the API purpose-built to talk to the new gatekeepers come through us and the transaction comes to you direct. Now, they'll have to deal with the gatekeeper on economics, but they would anyway.

Sanjay Vakil19:14

That's right. Well, that's a pass through in some sense. And the open question is, does this look like a direct booking to them? Are they merchant of record? Yes. Do they get the email address and phone number of that user? Absolutely. Do they get to repurpose and contact that user again? Yes. Does the user get loyalty points because they went direct? All of these things are what I think of as a direct booking from the standpoint of the hotel and the consumer on the other side.

Moderator19:41

And Steve, you've been the CEO of a public company and you've had to drive that company through many storms and, you know, engineer strategy shifts and so forth. Take a second and imagine you're Glenn Fogel, right? And we're 5 years down the line and Direct Booker is starting to actually bite and take a little share. Is there a pivot? Is there a move you can make that destroys or hobbles the Direct Booker story?

Stephen Kaufer20:09

If I'm Glenn Fogel, I'm super happy because Booking.com is going to continue to grow through this transition and look at their growth over the past decade as hotels have taken back more control of their inventory. All the chains talking about a growth in their direct bookings and that company is still flying. It is not an either or. I would never short Booking or Expedia now. They're going to do just fine. We think Direct Booker really has the opportunity to help the individual hoteliers, to help the chains with this brand new technology that no one really knows where it's going to go. And it's super easy to play that card. Get in the game. We're not Direct Booker. We're not in control of what OpenAI is going to do to make that travel booking process easy. We know they're going to play there. And so here's a chance to hotels in San Diego. How can you as an individual hotelier hope to influence what they're going to build? They're not. But if you give us our technology, you want to win some more business. How about your member rates? Put that right up against, hey, Glenn Fogel's unhappy about that, but it's not materially affecting his business for a while, and we'll see how it plays out.

Moderator21:34

Right. So you think there's a chance for you to play as an additive, a value added player in this ecosystem without picking a fight with the big dog?

Stephen Kaufer21:42

Yeah, totally.

Richard Holden21:43

There's still a lot of travel that doesn't get planned effectively online. They still could grow within that as well too when the tools get easier to use and richer. And again, going back to my original point of consumers having all the world at their fingertips, but all the labor involved in it. Now, this is a way to streamline that labor and bring more people into online travel planning.

Moderator22:04

Final question because we're almost out of time. Do you think that we are in what a lot of people are calling an AI bubble? And is this like what you and I lived through back in the day, Web 2.0 all over again, or is something different going on here? Is this more constructive, Richard?

Richard Holden22:24

I think this is more constructive. I mean, who knows? I guess that would be my first statement. But my second one is that I think real stuff is being created here. Real substance. And it's just I think the growth is there.

Stephen Kaufer22:39

I'd have to echo that there's more hype than I and anyone else care to deal with, but there's real value being created in a ton of these companies. If you're using them, you can actually see it and taste it, which I couldn't say for a lot of the other things.

Sanjay Vakil22:53

Yeah. I agree. It's pretty frothy at times. And I think there's some startup valuations that are completely wacky. But I think there's a core of really interesting work happening right now that just wasn't available to us 5 years ago, 3 years ago, 2 years ago. And that'll be an enduring benefit going forward. It's not going to disappear. It's not a Webvan situation.

Moderator23:14

Well, look, this is an exciting and interesting adventure you guys are on and to have the three of you up here for 20 minutes was a real honor. So, thank you so much for taking the time and thank you all for sitting through it.

Stephen Kaufer23:25

Pleasure.