Heads In Beds Show - LLM & AI Tools Are Driving HOW MUCH Traffic To Vacation Rental Websites?

Episode Date: February 26, 2025

In this episode, Conrad and Paul discuss the impact of Large Language Model (LLM) tools on vacation rental website traffic, questioning the narrative of Google's decline and the takeover by AI. They explore whether LLM tools are truly changing search habits and trends. They dive into original research from January 2025, analyzing client data to determine the actual session data coming from client sites. Despite claims from online experts, their data suggests Google is far from dead...Enjoy!⭐️ Links & Show NotesPaul Manzey Conrad O'ConnellConrad's Book: Mastering Vacation Rental MarketingConrad's Course: Mastering Vacation Rental Marketing 101🔗 Connect With BuildUp BookingsWebsiteFacebook PageInstagram🚀 About BuildUp BookingsBuildUp Bookings is a team of creative, problem solvers made to drive you more traffic, direct bookings and results for your accommodations brand. Reach out to us for help on search, social and email marketing for your vacation rental brand.

Welcome to the Head to Med Show presented by Buildup Bookings. We teach you how to get more vacation properties, earn more revenue per property, master marketing, and increase your occupancy. Take your vacation rental marketing game to the next level by listening in. I'm your co-host Conrad. And I'm your co-host Paul. All right. How's it going? Conrad. I'm your co host Paul. All right, how's it going? Good afternoon. What's happening, Paul?

It's a, you know, it's a nice little middle of the weekend of the week here is I'm not forgetting kind of what days of the week is it's, it's been an interesting stretch since we came back from from vacation. So there's there's still some brain fog, I'm going to say. I remember recording but there are some parts as I've gone back through them. It was it was a little tight. I missed a little bit of it. So how are you sir? How are things going? Hopefully a lot healthier and a lot better for you.

Yeah, that the health has been fine. No major hiccups there in a little things here and there but not enough to warrant brain damage, which it sounds like maybe crept into your household. So that is a bummer. Those things do happen from time to time. So yeah, no complaints on my side. I text you a little video a few minutes ago of it snowing here for a very brief period of time, but the snow's already gone. So more cold weather throughout the throughout the southeast here, which is not what we signed up for. I've got my I've got my cozy RLX jacket on today. So we'll see how how that goes. It keeps me warm enough here in my in my office. And the electric bill is going to be wild. You know what I mean? Like I get these notices now for me, like your bill is going to go up a lot. You seem to be burning a lot of energy throughout

the day. It's like, yeah, I'm turning the heat on dude. It's like 25 degrees outside. And I'm not cut out for this anymore. I used to be, but you know, you live here long enough, you lose it. You know, I have it and I lost it for sure that, oh, it's not that cold. It doesn't bother me feeling. It's not even you. The houses down there are not built for this. You were hearty in your younger years. You've earned your stripes of cold weather. I'm not gonna like we can't dispute that. But you had the confines that were built for a cold weather environment in virtual beach. Oh, fireplace, a fireplace right about now, you know, cracking,

hanging out in front with the kids. That'd be elite right now. But now we just got to crank up the heat in there in our system and just prepare ourselves for the $400 electric bill that will find its way upon me in March. So that'll be fun. Well, anyway, speaking of things that are changing from a data perspective, not only is my electric bill going up, so too is the conversation. The conversation, Paul, that Google is dead, Google is dying, it's over.

Google had a pretty solid, you know, 25-odd year run somewhere in that time frame, and it's over, at least according to the pundits that you might hear on LinkedIn. So here's the premise of today's episode. Let me see if I can set it up properly and see if we can dive this in. Because I think that what this is going to be is there's

been a few series now on the podcast, because we've been doing this for a while, that it's become a little bit reoccurring. This one will be reoccurring, I think, a little bit too. And it's checking in on the pulse. So right now, I've got my fingers placed on the carotid artery or whatever that artery is in the neck. I don't even know. Not a medical student. The neck of Google. Okay. Yeah. I got, I got in the neck of Google and everybody's telling me that this patient is dead, that

it's over and that it's gone, but I'm putting my finger on the pulse right now and I'm hearing, you know, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum,

bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, bum, disconnect between what people say online and the so-called experts on LinkedIn or other places and what's actually happening in my client's actual Google Analytics accounts and what the actual data says. So today is an inward look at our client's data anonymized and not put out in any specific context about the fact that Google is dead, right, according to the experts, and these large language model tools have killed search completely. I'm going to gently push back on that notion as we get going here and perhaps present some interesting findings from some original research that we've done here in January for presentation here in February, all about the session data that's coming in for our, our client site. So that's kind of the context. Where do you kind of stand on this whole narrative thought process of the death of Google and the large language models are going to take over? where do you stand in that whole conversation going on today?

I think it's important to know that there's going to be traffic at some point from these models. And yeah, we should really start to try to create content that answers questions that... Wait, that's SEO. So like, there's just so much that happens on Google that you can't really discount 10 billion coming down to 9 billion, coming down to 8 billion. This is the number of searches that we talk about in given periods. That's still where a lot of the volume is. So yeah, there's movement on the AI side of things.

I think it's really, really important to understand what that means. And yeah, I think we should take steps to ensure that that is on any future roadmap of content strategy, content planning, anything like that. But for right now, the shift is not there. And then we'll take a look at the numbers. And I don't think we've come out super pro AI in alignment. It's more of a, it's not against AI because it's got some great uses. It's, is it going to immediately change the way we see search habits and search trends take place? And I just don't

think we've seen that yet. And again, this is, it early on or we're kind of a I mean depending on the wheel what you're looking at there's the beginning of chat GPT and the different you know models and everything like that and Gemini coming in Bard and all these other things depending on how you're looking at that we're a year and a half two years max into this but mostly we're like six months in the middle out of this. So let's understand, let's take into the entire scope and look at that time frame in a logical way and not just say, oh, one thing came through, two things came through.

Yeah, small movement is coming through 98% of what the movement is still going to move behind the scenes as it does right now. So that's kind of, I would say, I mean, we're coming on on the same side of things. And it's just contextually where we're landing on maybe how soon things are changing and where they're changing to. Is it chat, QBD search? Is it perplexity? Is it Gemini? What does that actually look like? Yeah, you touched on an important thing there that I want to say again. This might have been mentioned on previous episodes or things that we've done before. I do so many things now, Paul. I'll be honest. Sometimes I forget what I said on what format, if this was like in one

hour that I did or a podcast or something. I'm also doing two podcasts, so just to add to the confusion. So anyways, if you're a listener to the podcast, you've heard this before, I guess a small apology maybe do, but I'll say it again, for many people who haven't heard this before. A lot of the my use cases personally for Jack GPT, which is the AI tool of choice that I tend to use the most, I know Paul is more of a Gemini guy, are not replacing search because they are not what a search engine did a year ago or two years ago, to your point before, like a search engine never did something where it's like today, for example, it was take this transcript of a call, distill it down into the key points and then build a scope of work based on the sales call that I had with a prospect. That is something that did not exist before. If a search engine was an Apple, that is a giraffe. They literally have no, there's no contextual relevance between those two things because the fact that I'm using chat GPT for that function or purpose does not replace my need

to also search Google for other information about the client or other specific data points or something like that. So that's, I think, one important point to think about, which is that not all, just to state it very plainly and clearly, not all growth from Jack DBT needs to come or will come at the expense of traffic used in a traditional search context on Google, on Bing, whatever, right? Duck, duck, go pick your search engine of choice, right? So that's one, I think, important point to lay out, which is that, again, one's a banana, one's a giraffe. In some cases, yes, there could be some overlapping purposes, but my gut reaction is like the way that people are using ChatGPT,

myself included, if I'm any of a, you know, if I'm a sample of one, then that's my sample size, is that I'm doing things that I literally was doing in a completely different way before with a new large language model tool. So that's one thing to kind of pick out there. I would say the second part of what you're saying there that we probably do agree on mostly is this idea that if it was a baseball game, we're in the top of the first inning still. I don't even think we're in the bottom of the first inning yet. Maybe we'll get there at some

point next year, you know, in 2026, you know, or something like that will be the bottom of the first inning. So we have a long way to go and how these things operate. And I know I feel that way very clearly because think of how much work right now I don't know how you feel about this has to be done like a web interface like the chat tbt web or interface or the App whatever and then then copy and paste it and put into another system Like that's not how most knowledge work is done most knowledge work That's done is put from one system into that system like I don't typically work in data outside of like a spreadsheet and then put it into a spreadsheet and then

You know have to keep going back between the two things. That doesn't really make any sense. But that is kind of how I'm using Jaxi BT today. So my belief is that eventually all these AI tools will have a button where it's like summarize this and it just summarizes the whole thing. You kind of see little flicks and artifacts of this in Google products today, like in spreadsheets,

like the Gemini functions in Google spreadsheets, but it doesn't really work that well, if we're being honest. If you click summarize, it'll sometimes get a little bit off. I'll ask it to make a table with certain information. It'll say, I can't do that. I've asked Gemini to respond to emails, or at least draft the email on my behalf. And it's like, oh, no, there's personal information in there, so I can't do it. So again, very early in this kind of stuff.

And I think that the way it's going to look two, three years from now is nothing like it looks today, and how some of these AI tools work together. Also makes me wonder about, like, I don't know if you see, have you seen the comic where it's, I'm using a chat GBT to generate this email to send to somebody, and then on the other side, it's like, I'm using chat GBT to distill this email down into bullet points so I can reply back.

So it's like robots talking to each other, you know, with this AI layer on top of it, acting like people, you know, which I thought was kind of a funny thing. So, yeah, I just wanted to get that out there. And then the second piece of what you said there, the size of Google, I think some people don't know. And to be fair, I tried to hunt this down in preparation for today's show, and Google no longer releases data. Last time they released it, that is, you were right on there. The number of billions, eight to 10 billion queries that they're processing on an annual basis,

Google is so big, it's just almost hard to fathom how much search volume actually goes on on Google. And we're here in our little tiny weird corner of less than 1% of the search volume giving all of our takes on it. Obviously, we're seeing a very specific narrow look at it. There's no doubt in my mind that many search fields have been impacted heavily by things like algorithm updates, things like Google's choice to include featured snippets. We've talked before on the site about the celebrity network story is a good story until people just want one answer from some sites. Once they get it, they go away. But our little world, our little bubble of vacation rentals, we've got a little bit more of a moat around us with respect to these large language model tools, because people still want to go look at the website.

Because we have such unique inventory, it kind of works in our favor a little bit. I think of us lasting a little bit longer in these search functions as well. Because people want to go and look. They want to go look at 10 different units before they make a reservation. And they can't do that at the moment very well through these LLM tools, even the ones that are search-specific like Perplexity. So yeah, that's, I don't even know if I had a really strong agenda there.

There's just like things that I've been bouncing around that I wanted to get out there. And then we can get to the data here in a second, but any reaction to those? I think it is. I think the counter to that is, oh, it's going to accelerate really quickly. What has kind of accelerated quickly? I mean, you can't argue that it hasn't. But I think it's a very similar discussion to what some small business owners are having around TikTok right now. And

I think rightful outcry over the people who are using that as a huge sales tool and just brand exposure, brand awareness play is that it is, I think that at a certain point, you can't just move people over. You can't just automatically move eight billion people. Let's say five, let's conservatively say, say they lost at some point in life, five billion searches. You're just not going to see that go. And I think it is. It's the experience of looking at what the LLM does when you try to use the travel side of things. I think that's also to our benefit, is that it's not a great experience right now, or it's still sending people in some way back

to some type of result that we'll talk you know, we'll talk about here a little bit, but it is it's big, it's, it's, it's a giant, it's monolithic, these are things don't just disappear overnight, and the digital side of things. So right, right. Which is my other analogy, I'll do that real quick and then we'll dive into data, which is that Google is literally like a glacier, it's a glacier on the North pole that is massive. Like you can't even, you need a car to get from one end of the glacier to the

other end of the glacier. That's how much search volume that Google takes in on a monthly, you know, annual basis, whatever you want to measure it out and chat to BT, if it is in fact taking away market share from Google is doing so by running the, the chat to BT boat up next to the glacier and they have a shovel or a pick and they're picking away at the, you know, at the, at the corner of it. right? And it's like, all right, I'm taking chunks of this glacier off. It's like, yeah, dude, you might be like, I'm not disputing that. It's also a glacier. Like, you have like a long way to go, literally billions and billions of queries to go to

potentially kill Google. So that's just the context there. I put it in our chat here. This wasn't in our outline, but according to a similar web, chat GPT is now the eighth most visited website in, I believe this is US only, although there's some other data sources from a similar place in there. So it is bigger than a site like reddit.com, for example, or it's bigger than Yahoo already, which shout out to Yahoo for holding out the 11th spot, seemingly decades after their cultural relevance, like they still are such a large, ironically search engine and content and publishing, you know, platform, but you've got Google, it's the most traffic YouTube, Facebook, Instagram, formally Twitter, the artist formerly known as Twitter.com, WhatsApp, Wikipedia, chat, TBT, Reddit, and then Yahoo Japan, which is interesting, is your top 10

according to similar rub. So again, if if chat so chat, TBT is already eight, according to similar rub data, if it was much higher, presumably, again, Google will be taking more of a hit. That's not the case. Google is still by far the most visited website, the most visited search platform in the world. And it's not the case. Google is still by far the most visited website, the most visited search platform in the world. And it's not even close. There is not a close second. If anything, it's funny that they control two of the top two results, right?

Google and Yahoo are controlled. Google, excuse me, and YouTube are controlled by the parent company. So just to show you how much effort has to be expended to potentially make some damage, that's all us setting the table and giving you some context so that the listener, hopefully on the other side, can visualize how big this is and how much work it's going to take to chip away at that glacier.

So that's my thought process. So here's what we did. We set the table on the day of analysis. So I have my theories, I have my suspicions. You know, I kind of talked about it at the beginning. Here's the things I don't like. So I said, all right, I have someone on my team helping me with this. And I said, let's go into all of our clients data for January of 2025. Let's go into analytics looks specifically

for our vacation rental management clients, ones that we work with currently, and just ones we still have access to who never took us off analytics. Thank you to those anonymous clients, so we can get more data. Didn't do anything nefarious. All we did was go in there, download the search data from analytics, and then we exported it, look for how many visitors are coming through these different LLM tools. So here's kind of some of the numbers to share with you, the dear listener.

In January 2025, we recorded from vacation rental management sites, 734,376 sessions. So that's from all traffic sources, we're working with a sample size of roughly 750,000, a little bit less than that, just to give you some context. So of those 750,000 visits, which would roughly the population of a small state, not quite. I mean, it's a little bit less than that, but you get the point. A lot of people. The total number of sessions that were attributed directly to large language model tools, and I'll break down the ones here in a second of what they were, is 429.

So let me repeat that for you. 734,000 sessions came in. 429 of those sessions were attributed directly coming from an LLM tool. So to run that as a percentage, it is 0.0584% of total sessions were driven from LLM tools. Google is like 30% for context, depending on the client, sometimes as high as 50%. So again, if we put those up on a screen right now, and you were able to see it, you'd see 50% is like half the pie, and then less than, not even less than 1%, less than half of 1% or right around half of 1% of the traffic coming is from these large language model tools, according to the data set that we looked

at. All US property managers, all anywhere from like 10 to like a thousand units. And again, we had roughly 750,000 sessions to base this off of. Now you could say, hey, Conrad, maybe your agency just sucks at optimizing for LLM tools, so that's not representative of what I'm seeing in my data. If so, I would love to be proven wrong here. Please send me your data if you're willing to be added in. And when we do this again, we could update this down the road.

Obviously what I'm curious about is how does this change? We're recording this in February 25. Again, this is based on data from January 2025. Let's say we did it in June or December or something. How much will have changed this year from this current market share of around half of 1% compared to Google, which is getting quite literally 100 times more traffic in our data set for most of our clients. They're getting 50% sometimes of their traffic from Google.

So your reaction to these ideas and it beat me up a little bit on some of this data because I know when we were doing the outline, you had some flaws and maybe some of the ways that we're collecting this. So I'm open to that as well. What it is, and I think that this is where, generally speaking, and what we can attribute not as spot on. And we've talked about that, that that's just the reality, even on some big sites that should probably be I mean, it's big brands, big, you work with some big players that could and by right should be getting some traffic from those sources, and they're not. Or this

is very limited, clearly. So, no, I think that's spot on. Now, here's the thing that I talked about it in the LinkedIn post, and part of it is that ChatGBT only gives credit to the traffic that's coming from the actual click on the source itself inline within the chat there. If you look at the citations or look at anything that they're using on the side, there's no UTM parameter on that. I don't know why.

I think it's a silly oversight or something like that. Or maybe they don't expect anybody to be clicking that, whatever that is what it is. But I think that that's something where we're probably getting most of the referral from ChatGPT, but on the Google side of things, you're just not getting that. And there's actually something, one of the searches that I was doing, I was trying to use the new 2.0 with the learning, I'll get the actual model name, but it is, it's overlaying those over the apps. And what I thought was really interesting is that

instead of doing the searches and kind of finding the results, that happened in one of the queries, just kind of trying to plan trip from, in a beach market with pet friendly accommodations and a certain number of people that we needed. We got a few results, after a while we were just getting search results. So it was Google search for this and not actually giving any result, any traffic

to the actual websites and just coming back to a very bad search experience because you're searching for the URL. We know how that works in Google search. So So I do I think that that's something that Google's not attributing their own traffic properly I don't know if they're doing it giving it to organic I don't know if they're giving it to any other areas I think that's the nice part about when we have like a the AI overview or the Featured snippet, you know, it's going to give you a URL parameter that's actually bringing you kind of to that. I mean, it

is you can kind of parse through those little alphanumeric plus the query because those actually show up in kind of the parameter to get you to that anchor point before that content is Google smart enough to be able to create that. I just feel like it's a fundamental miss that they're not showing it, but from their perspective, do you want to show 1% traffic, half percent traffic from something that may not be producing the way you want it to either? So I don't think we're missing a lot of traffic from the LLMs, but I do think there's probably another percent out there that we're maybe missing, maybe 2%. We're not talking about 20%, 30%.

This is still at best your seventh best channel. Grouping would be AI models after social media, organic social media, probably paid social media, email, organic, general paid, general. I mean, this is kind of, I get it, put it in, put it in its, in priority within that conversation of this is how much it's driving right now. Maybe I should focus more on these other channels that are, if you're running paid social, focus more there. If you're running more email, focus more there because that's moving the needle right

now, more than anything you can do just to optimize for AI. If you're not building better content on the SEO side, then you're not optimizing. Right. And what we don't know today, I mean, there's a few things we don't know. So you're right, the breakdown of the LLM tools, by the way, is that chat GBT was 346 of those sessions, perplexity was 73. And the Gemini was 10. Now, the question that we might have there is that, to your point, how many people are clicking out of the sources that want to do these searches,

and how many people are doing these very long-tail, very specific vacation rental searches in these LLM tools? I think this data is pretty clear that that's not really happening at any significant level, right? You're exactly correct that pretty much every other traffic source is going to give you more traffic than these things right now. And the other side of it too, and we've talked about this before, we don't know the answer to this yet. So anyone that says they know the answer, I'd love to see their methodology and their data behind it. But what is the difference between quote unquote optimizing for an LLM? And what is the difference between quote unquote optimizing for Google, let's say, you know, because the way I see it, Google and

I created this years ago. When Matt Lando had me do a guest post on SEO, he said, people get so confused. Break it down. Make it simple. I go, cool. T-L-C-K. Technical.

Fix the website so it's in good technical shape, using the best practices. Link building. We need to make sure people know about the site. We've got to get links to it. They've got to be relevant sites. They've got to be sites that have their own links pointing to them that are authoritative, et cetera.

Keyword research. We've got to know what people are looking for that we wanna rank for, and then we gotta make the content. We gotta actually put the page in the site that targets that keyword, hopefully in a comprehensive way. That's it, everything in SEO fits in those four buckets. Technical, link building, keyword research, and then content, right?

Everything fits in those buckets. I'd be curious to know, once we get more data on this, what is the difference between those four activities and how an LLM might crawl a website? How ChatGPD is crawling your website, or maybe it's the parent company, right? Open AI crawling the website. Are they looking at technical ability, their ability to crawl the website? I bet. Do they or does their caller work the same way Google? Probably not. So

that's worth, you know, understanding. Maybe they have different, you know, improvements that we can make down the road to be like, oh, we're optimizing the site more for an LLM to crawl it maybe in terms of how we're doing headings or structures or something that's worth discussing. Does LLM tool use links? I think so. It's hard to know right now. But when you go do a search and you say what are the best car brands, it tends to look at car brands that are new and have lots of popularity and links and reviews about them. It seems to do index more on content, I would say. And then the keywords is almost like the questions you're

asking the large language model tool. So it seems like those raw ingredients, the way that we bake them up for Google is going to be pretty similar to the way, for the most part, that we're baking them up for the LLM tool. The difference seems to be more of the review and sentiment analysis that an LLM will do that Google doesn't really do. So Google doesn't, in my experience, have quite as much say in, I'm going to show a website ranking on the first page of a Google search,

even if that brand or that reputation that site has isn't amazing. It seems to still tolerate that, particularly if it has a lot of links. LLM tools don't seem to want to do that. Like if a brand has a negative sentiment about them, an LLM tool tends to seem suppressive more, you know, if it has bad reviews or bad sentiment about it.

That's one little thing that I've noticed that I can point out specifically across a few hundred searches, like, okay, I see what you're doing here. You're showing the brands that have better, higher quality reviews or better sentiment online that you're able to crawl an index versus brands that have a lower sentiment review. So if sentiment online, they're able to crawl an index versus brands that have a lower sentiment review.

So if you got bad reviews, you probably might not do well on an LLM tool. You're probably not gonna do well anywhere, by the way, if you have bad reviews. You're not gonna do well on Airbnb, you're not gonna do well on Verbal, you're certainly not gonna do well on Google if you have a lot of bad Google reviews.

And it's hard to believe you're gonna do well on any other platform. So that's a pretty logical container, but if maybe you thought you could hide behind bad reviews and I could just pay Google and get the traffic anyways, maybe that's gonna be a little harder in the future with these LLM tools. So that's a speculation. But that could be one thing I've noticed on my side so far. I mean, I think the one thing that just taking a look at some of the technicals on the robots.txt

files, making sure that there are some sites, some builders that just have it built in that they don't want the LLMs to scrape or they have the kind of restricted from the LLMs actually scraping the site or being able to actually crawl the site. So I think that that's something that you should be cognizant of. I think at the very least, just make sure that there's nothing there. You can just kind of look at that. Usually if you type that into the URL on the website or, you know, there are some other options, just talk to your web person as well. But that's something that I think it's a good check.

It's a good health check just to make sure that you know that you are being indexed because that's the other thing is that we can, we can game the system a little bit with these LLMs right now and we can kind of track the system to look into specific websites and kind of parse and crawl those websites a little better. So making sure that that's not the first time that they're they're running through the website is is something that I've taken to making sure any website that we're working with, that's what we do. We kind of talk to the website,

talk to these LLMs about the website, just to make sure that they understand. I think that's probably gonna be kind of part of a, I don't know if it's an SEO check, less moving forward, but I think you're going to want to, like you index a site map on the, on Google and Bing right now in the search console, you're gonna wanna do something very similar to that

and kind of talk up a website with these LLMs to make sure they understand and they're crawling all the pages that have key information because you want them to be able to answer questions with your websites, which you can moving forward. Yeah, I think that's very true. And like I said, I could see, I could see a future where we're trying to optimize, quote unquote, optimize for an LLM tool. And it might work a little bit differently than how we do it today for Google. But my gut reaction right now is like, when I go

do searches, it seems to be that there's a pretty reasonable correlation between ranking well on Google, and quote unquote, ranking well or returning results in the LLM tool, right? Specifically, as the search tools have gotten better, like on ChatGPD specifically, it's going out and gathering more data from the web. That's a little bit faster, a little more instant. Obviously, I think that is Gemini strength for sure. You can ask it, what was the score of the Lakers game last night? And it'll know that it's not LLM tools like on the ChatGPD side have been a little bit

slower in that respect. And by the way, this was the person on my team that helped me with this task when it looked for Claude data, not realizing of course that Claude doesn't really send any referrals out because it's basically, Claude is the one platform that really the least connected of these, right? Perplexity, ChatGPT, Gemini have a significantly more ability

to crawl the web, index the web, store the web than Claude, which has really none of those abilities. So Claude recorded a grand total of zero sessions because there's no sessions that would have occurred from Claude even if a brand was mentioned in there. So what might be happening, by the way, and I'm willing to admit this, is that people are using a tool like Jack GPT or Gemini to search and kind of what are the best companies in this market? I'm looking to go vacation here, et cetera. They might be going, oh, okay, that's a cool name. And then they might be going to Google and then

searching for that company, or they might be going directly to that company. So of course, both of those potential behaviors I suggested would bypass the tracking that we look for on the analytics side of things. So I'm fully willing to admit that my data collection methodology here is imperfect. I'm totally willing to admit that. And this came up too. I know there's a Slack channel that you and I are both on with a company that you used to work at.

And there was a comment in there about how'd you hear about us? And someone said ChatGBT for the first time in this lead form of the company that you used to work at. And that was kind of a watershed moment, right? That never happened before. Every other source had been new. So that was someone of a watershed moment, right? That never happened before. Every other source had been new. So that was someone discovering a product or a service through ChatGPT.

Again, who knows how they actually got to the website. Maybe it was through ChatGPT, maybe they Googled it, maybe they just went directly to the website, but they said, hey, I did some research on ChatGPT and that's why I was led to this particular service to offer this particular person that this thing was looking for, right? So that's an example of like, yeah.

And that's always been the case, for example, when we work with clients that have high social reach, they get all these referrals from Google and all the analytics will say Google, and then you go do the digging and you're like, oh yeah, you Googled it, but that's because you saw it first on Instagram. So whenever I work with these like so-called viral Airbnb projects,

I see that the majority of their traffic is from Google, that's a correct statement, but the majority of their awareness comes from Instagram and then people just Google it and they go, oh yeah, what's the name of this property? Let me Google it, find the link, find the website, and then get there really quickly. So that's another piece of the puzzle here that we obviously are bypassing or we can't really track in a way unless you're doing qualitative study of your guests when they book and asking them,

Hey, how did you hear about us? Which I think is a very valuable exercise to do if you have the bandwidth to do it, because you learn a lot about how effective ads are and they don't get any credit for them, right? Like that? Whenever I'm putting an ad as a port up for a client, by the way, I always feel like what I'm showing them when I'm showing them click attribution is like, hey, what I'm showing you here is the worst case scenario. If we're getting a 10 to one ROAS, that is the worst case scenario, because these are only the ones I can prove. All the ones I can't prove, like the 19 direct book news you got last month that just were marked as direct, a lot of those were likely influenced in some form or fashion by being a previous guest, by clicking on an ad before, but I can't prove it because of all

these privacy restrictions now that we have on LX tracking. So it's like, again, when I show you who is probably the worst case scenario, probably fair to say the same comment about my LLM data. It has to be the worst case scenario. There has to be a little bit more influence coming in there. But again, just to fly in the face of this conversation around the fact that these LLM tools are taking over, the data really does not support that right now in our world, that's for sure. I think that that's with the attribution path, that is still, even though Google has done their best to try to hide that or just put it, feel like you can only find it in campaigns, no, no, no, it's still there. But I think that's one of the most critical reports

to go over and just demonstrating that. I mean, it is, even if it's the 15 organic sessions, that's a lot of different times that people are hitting your website and just understanding that it's a pathway, it's a process, and we will. At some point along the lines, we will see that chat GPT shows up

in that attribution path report. That would be phenomenal. I mean, awesome. I welcome that day when it happens. But I think the other thing is getting outside of Google Analytics, there are some tools that you can use. We talked about it. We used Amplitude paid model. But if you want to pay and get additional data,

you can go at a higher level and kind of match people in understanding that they're using different devices. additional data, you can go at a higher level and kind of match people and understanding that they're using different devices. They're using the same email addresses attached to six different devices and 14 different accounts and things like that. You can start to put together a much more comprehensive picture. Again, you can go that deep. What are you actually going to do with it? Does that actually inform better decisions on your SEO or this and that? Maybe. Maybe retargeting, maybe audience building and stuff like that. But it still goes back to the fact that we can see. We get a picture

of how people are hitting the website. It's not nearly as good as it used to be. But I think that until you're consistently seeing Gemini or you're seeing a better trip planning, let's reverse even to the overall experience on AI, until you're seeing a better trip planning experience and more of the younger generation is taking advantage of that because I think it is. They're still looking now at TikTok. If you can get bookings set up in Facebook somehow, that is going to be a platform that I think you might see

direct bookings come through at some point along the lines. Then we don't have to worry about the attribution that we can't see through that channel. But there's just so much more to optimize more and think about as opposed to focus. I mean, it is, we want, I think it's beneficial for Google to have a better experience, but it's just not there yet.

I think it's beneficial for chat GPT, I have a widget that connects directly into Expedia. It's not there yet. And those are things that I think we thought were gonna be there in the first four to six months and 18 months later, they're not. So again, that's the other piece of this puzzle is that not even search volume, it's that the experience is no better when AI is incorporated into the trip planning experience right now. It's still better to go to an OTA or go to Airbnb or hopefully go to your

direct booking website and get that truly custom experience of this is what it is. This is what we've created. This is the experience that we want you to have and book on our website, please. One of my kind of main things that I believe to be true, the more I do marketing in general, is this idea that any channel can work. So there are people in our space who absolutely have figured out the TikTok channel, like TikTok has become their source of, you know, interest and demand. I think TikTok is a really tough one to tie your wagon to because as we've talked about extensively offline, it is a very fickle channel, you can have some reach there and then that reach can seemingly disappear in a very short period of time. If there's been complaints from people using, you know, getting their site penalized by Google, there absolutely has been complaints from people who have gotten all this

reach on TikTok. And then that reaches there and it's gone in the flash of an eye, you know, blink of an eye, which seems really problematic and really frustrating. If you would spend a lot of time, effort, energy, you know, cost video production skills to get well done on TikTok. And then all of a sudden, boom, your reach has just gone right away. That would be a pretty tough thing to swallow. I think if you're relying on that organic reach, I think the same thing can happen on Instagram, although it doesn't seem to be quite as, you know, discrete. Again, I'll mention I've been Wolf again, we had him on the podcast a few weeks ago, that dude is figured out Instagram,

that guy has Instagram down pat, right? Like he understands that format, right? From both the way that you market, etc. Right? That's, that's awesome to say. That doesn't mean though, that it's always the right channel for everybody, or that it's the best channel for you, you have to look, you have You have to look and say, what do I have? What am I marketing? What am I good at? Or what can I hire for?

What can I hire competency for? And then you have to come up with your own strategy, right? Like we can give you a lot of different ideas and fractals and pieces of the puzzle, but ultimately you have to pick it up and you've got to paint your own canvas. I think that's kind of where we are. With these LLM tools to kind of put a bow on this whole idea,

there's no one that you can point to today. There's no version of it where it's like, oh, Paul really knows a lot about homeowner marketing. Let me emulate some of his tactics with homeowner marketing that he's used for other customers and clients that he's worked with, and then see if some of those things would work for my business.

Let's see what Ben Wolf does with his Instagram account. Let's see what Conrad does with some of his clients on the SEO side. Maybe that's something that I'm decent at. I mean, we'll see. But we can point to examples and say, this is someone who's figured out this channel. Here's a company or brand or an individual, an agency, whatever that's figured out this channel. Let me go look at what they're doing and then modify it from there. Like, anytime anyone comes up in a meeting to you,

a marketing meeting, and says, should we do X? I think the first response out of my mouth now is, like, show me someone who you think is doing a really good job on that channel. And I think that it's one of those two things where if you're really good at that channel, and you can prove it and show it that you're really good at that channel, and you can do it consistently, not you had one little moment or one little flare up, but it's like, hey, month after month, year after year, we're running on this channel. I've been doing SEO now for basically 10 plus years at this point. I know what works. Again, TLCK, right? Technical, Linux, Content, and Concerned Research. That's been a pretty solid formula. Do I hit that a thousand? No, of course not. I'm not going to sit here and

say that I do. That's not true. But generally speaking, when I look over a long enough period of time, that seems to be a really solid path for getting more traffic, getting more relevant traffic, ranking for those high value keywords. When you get those clicks from Google or of course, you know, running a lot of ads over the years, when you pay for Google for clicks on those ad keywords and your site setups to convert, that seems like a pretty solid way to get to, you know, 10, 20, 30, 40, 50 or more direct bookings with those methodologies place.

This today is not that it could it be in the future? Absolutely. We can clip this in a year or two years, three years and say, oh my gosh, look at this new chat, GPT, vacation, rental search planning tool. It's unbelievable. This what people are using all the behaviors going there. We will gladly say that we're not saying that this is how it's going to be forever. But gosh, don't listen to these gurus, man, on LinkedIn that are claiming that

Google's dead because this is the 50th time that's been said. And yet, we'll bring up our similar web chart from earlier. They are number one number one, the most visited website on planet Earth today. And it's not even close. So chat GPT is different. Gemini is almost different even within Google. All these schools are different. Don't get distracted from the prize. If your prize is I want to get more direct bookings, I want to get more traffic to my website. Think of these other channels that you can leverage better think of how you you can leverage to your point. Like you said, TikTok, Instagram, you know, can I do better email marketing? These are all things that you can do that will provide you 10, 20%, 30, 40% of your traffic coming to your website. And they're likely attributing, you know, again, 10 to 50% of your direct booking. So don't get distracted. That's my last thought we can put a bell on this one. But I just wanted to, you know, break it down, do the data. This will now be my reference point, by the way, when people ask me like, Hey, what about I want to optimize for LLM tools, I'm just gonna copy and paste this podcast link fired off in a message and say awesome, Paul and I did a deep dive on that exact topic.

Yeah, it's gonna be a hot one. Here's the data that we're collecting. Like, what do you see? Like, again, if you see something different, if you have someone out there, who's listening, who's getting 50% of their traffic from Pluriproxity, fire it over to me. I'd love to see it. And I'll keep you anonymous if you want to. Believe me, I'm all ears. I want to learn about this stuff. But boy, when I went and looked across 750,000 sessions, this was my conclusion. I'll end on that note, which is 0.05% is where we stand today. Let's see how that changes. Let's keep revisiting this topic and see how things change and go from there. I have, I have a mind of play dough. I'm willing to let it be, you know, modified and changed. It is not a mind of brick and stone, you know, so if things can get changed,

that is the case. But man, there's some people talking that have no idea what they're talking about and have no data to back up what they're saying, which is super frustrating. So there we go. What do you think, Paul? How should we put a bow on this one? Or is that a decent way to know that? No, that's that's we I think that's that's the rage out we need right there. It's the heads of some as raging as I get it's yeah, you don't you don't go old man shut nothing clouds too often. So it's good to hear

all good. Well, all good. If you if you listen to a 30 odd minute deep dive into LLM tools and how they attribute traffic or send traffic to a vacation website. You are a weirdo. Thank you. And I love you for it. But we need something for you for being a weirdo, which is traffic or send traffic to a vacation website. You are a weirdo. Thank you. And I love you for it. But we need something for you for being a weirdo, which is you got to leave us our podcast review. Again, the data tells us Paul, speaking of data, the thousands of you listen, speaking of low review rates or

low, you know, referral rates from LLM tools, we got a low review rate like 10,000 plus you guys listen to this thing. And we don't have a ton of reviews. So we need more views, go to your podcast, have a choice, iTunes, Spotify listeners, we get the most downloads from you there for a review on those two platforms. Helps us the most. Click follow or subscribe or share so you get new episodes because we're coming out with new ones every week. And then, yeah, secondarily, make sure you pop in, leave us a five star review.

We appreciate that. And we'll catch you on the next episode of the Heads and Beds show. Have a phenomenal day. Thank you so much.