Loading summary
Transcript114 lines
- [00:01]
Kip
This is a can't miss episode. Today we are breaking down the AI Search engine optimization playbook. Everything you need to do to get visibility, awareness and traffic from services like ChatGPT. We've got an amazing SEO expert, Asia, joining us today and we're going to break down the playbook. We got some amazing data to share with you all and we're going to give you the step by step approach at the very end. So you want to watch this whole show. Let's get into today's episode. We'll be right back to today's show.
- [00:32]
Kieran
But first I gotta tell you about a week that you're not gonna wanna miss. It is a highlight of my year every year. It is inbound, which is HubSpot's annual event where we bring together the best attendees, the best speakers, the best experts in sales, in marketing and customer success in AI. We all learn together and we all figure out how to grow our businesses better in the coming year. And this year, it's no exception. It's a fire lineup. It's going to be at the Moscone center in San Francisco, California, September 3rd through the 5th.
- [01:07]
Kip
So it's a great way to kind of close out Q3 and kick off.
- [01:10]
Kieran
Q4 really strong and have your heads right for 2026 because you're going to learn and hear from Amy Poehler, Dario Amadi of Anthropic, Darkesh Patel, Sean Evans from Hot Ones, Marquise Brownlee, Glennon Doyle, and the world's best speakers and experts across marketing, sales, customer service, AI. They are all at inbound. Hundreds of amazing sessions. And you know, it's more than just the sessions. It's about the networking. It's about learning from your fellow attendees, your peers, understanding what people are doing really well, where people are getting stuck. That is how you break out. That is how you are going to grow better and faster in 2026. You're not going to want to miss it. You can go to inbound.com register to secure your ticket and your spot for Inbound 2025.
- [02:14]
Asia
Okay, so you are busy doing SEO. It's where I started my career. It's where Kip and Asia has spent a lot. We used to get a ton of traffic from Google Search, the blue links and all of that traffic is disappearing into the AI ether and it's disappearing into AI search. And what we are going to go through today is how to do AI search optimization. That really is the new thing you have to focus on if you want to get visibility for your brand. But the thing is, it's so new, it's really hard to understand. Like, what is the playbooks? What should I do? How do I make sure that my brand, my product appears in AI search? We're here to give you the full playbook for AI search optimization. We are lucky to be joined by Asia Frost works at HubSpot, one of the best minds on this topic. And so I wanted to kick off with this. I was like, looking at something this morning. I was actually looking at the growth in AI companies in general, just how quickly they are growing revenue. And I thought this was a pretty amazing chart just to kind of help people understand what's happening. Because for the average business, I hear like two types of stories. I hear, wow, this is really starting to impact us. We don't know what we should do. We have decline in organic search traffic. What do I do? Do I do SEO? Do I optimize for Google AIO views? Do I optimize for LLMs? And then you have other businesses who say, no, I don't believe this. This is going to take forever to actually impact us. It's going to happen over a very long period of time. And what this chart shows here is the time to reach 1 billion users. ChatGPT took 2 1/2 years. Google took 13 years. So ChatGPT did it 10 years earlier. And so it feels like this is telling us that users are voting with their thumbs and fingers and actually going to type all of their deepest, darkest thoughts into ChatGPT versus Google.
- [04:04]
Unknown
Yeah, I think that's right. The more and more that I talk to our customers, the more and more I'm just eavesdropping in on people on the street. I'm hearing them say, I am going to ChatGPT. And my next question is always, and how much are you using Google? And they pause and they say, you know, I'm using it a lot less.
- [04:22]
Asia
Yeah, everyone is using a lot less. This is the other one I really caught my attention, which is what I'm showing here is a chart that shows the estimated LLMs versus organic search value. Like if you had told people in 2025, not even 2025. I remember when we did the first show, like, I think around this in 2023 and we were talking about the fact that this is going to really disrupt Google because it's a much better experience. So even if you pull this chart back to 2023, how quickly traditional search has become unbundled into these AI chat LLMs, it's actually one of the Wildest disruptions that I can ever remember.
- [04:58]
Unknown
Yeah, we were talking about this the other day. It's happened incredibly quickly, probably more quickly than we had anticipated. The other really interesting thing about this chart that you're showing is the total LLM value, which we can see is more than just the traffic that's being attributed to LLMs. In other words, someone coming from an LLM is worth more to us than someone coming from traditional search. And why do we think that is? It's because someone is completing their entire buyer's journey in AI search. They're going from I have a problem to this is the solution very quickly. And because people feel like these LLMs are giving them very objective, unbiased results, I think they're also trusting those answers a lot more than the answers than they would've found on, say, Google Asia.
- [05:50]
Kip
I think that's right. And I think if you're watching the show today, that means you're pretty in the know and you kind of have a sense of what's going on, but you're probably working with a lot of people who don't have this level of context and you're trying to get them on board. The most simple explanation is like, Google and those 10 blue links, that was an answer engine. And if you were a company, you could create some articles that provide some answers and get some traffic for it. LLMs and AI overviews and all the ways AI is changing search, that's an action engine. People can go and take very specific action. They can buy, they can research, they can take the next step of the problem they're trying to solve in a much more actionable way. And that's one more valuable to the in business, but also takes a very different approach, which we're going to talk about in this show, of what you need to do differently. But that's like the simplest heuristic that I can provide. If you were just like talking to your CEO to try to explain what is happening. But I think what Kieran is showing here is the thing we have to talk about, is that the LLM visitor is worth way more than a traditional search visitor. And Asia, you've really dug into this. So I want you to kind of like break that down for us. But like, one of our early hypothesis is like, hey, we're gonna get less traffic from LLMs, but that traffic is going to be worth more. And Semrush is saying it's worth over 4x more, but, like, break that down for us. Help us understand what that actually means.
- [07:11]
Unknown
Sure. When you have traditionally captured someone from organic search, you know, you're capturing them at all different parts of the buyer's journey. Maybe they are, to use a HubSpot example, just trying to learn what content marketing is. So they go to our blog, they read a content marketing guide 101, maybe they download a kit to help them start implementing content marketing techniques. And we hope that over, you know, two to three weeks, maybe two to three months, maybe two to three years, as we continue to educate them about concepts related to our product, they eventually decide to talk to our sales team or become a free customer. So that entire journey takes a very long time. And along the way we are giving them helpful resources to help them understand that HubSpot is the answer to their content marketing need. Now someone is going to ChatGPT. They're saying, ChatGPT, tell me everything I need to know about content marketing. And ChatGPT is doing that in an incredibly personalized, helpful way. They are reaching the understanding that they need a content marketing solution a lot more quickly. And they are also learning about all the potential options for content marketing solutions much more quickly. By the time they decide that HubSpot is the content marketing solution for them and go to our website, all of that earlier nurturing has already happened and they want to talk to sales really quickly. And we see this actually in our gong calls. I have a habit of looking at our sales call transcripts every week and I see over and over and over again, yeah, ChatGPT said you were the best, so I came to you. ChatGPT is my best friend. They said HubSpot is the CRM I need to use.
- [08:58]
Asia
And there was a pretty incredible stat because I think we know this is happening, which is 80% of the B2B buyer journey has typically started with Google and over time that's going to become LLMs. They believe it's going to reach around 95% will start their journey with an LLM. But I think what you're talking to is it's not just they start their journey with an LLM. It's like they don't really end it. Like it's all self contained within the LLM. And so it gets much, much harder to actually reach that consumer because they can do so much more in the LLM and they trust LLM. They believe it can give them the right answers. It's unbiased. The other, you know, I think part of that is, and we're going to get into giving you some actual actionable things. You can do. But I think this was the other one I think is related to that, which is like a challenge for brands, which is they're in the LLM, they're doing a bunch of that research. And what this has shown us is what the LLMs reference are not really vendor websites. They're all of these kind of third party websites that aggregate together different vendors. And so like when they did this study, Jason Tabelen, he looked at a bunch of data and he said that the brand's actual website was only mentioned 9% of the time. So only 9% of the times for these queries was a brand website mentioned. And the links, that's pretty incredible. Like when you look at what they're citing and the sources they're citing, it's not really like vendor websites. I'm not sure. Have you seen that Asia in your analysis or has that kind of been one thing that has stood out?
- [10:25]
Unknown
Yeah, I think that this is really interesting data because I think what we don't know yet is whether LLMs prefer third party websites because they consider them less biased than vendor websites or if they prefer that content because that content is inherently more ingestible for an LLM. So when we transition to how do you optimize for AI search? A lot of the AI search emerging best practices are things that these third party websites have already been doing. It's just how their content is set up and displayed. So I think that there's a real advantage if you're a vendor to start applying these AI search best practices and become a much more sighted domain in your space.
- [11:12]
Kip
Okay, before we get into the how, there's a couple things I want to talk about. Part of the topic that you're talking about, which is like who is showing up in LLMs? I think it's important for us to give everybody the context is that Google has been on a decade plus long journey and fight to do two things. One, deliver a personalized search experience. And things like ChatGPT with memory have kind of trumped that experience and I think have been able to deliver a personalized search experience in the way that Google always wanted to. The second thing is that Google laid this roadmap where it was totally embroiled in all these lawsuits with publishers like Google stealing our traffic, Google stealing our revenue. And I think it's safe to say that OpenAI has seen that and wants to avoid those problems. Right. And they, they're going to likely make decisions and I think are making decisions in terms of content partnerships and things to avoid those problems. And so as you're thinking about your strategy, I'd encourage us all to have those two factors involved, which is one, search has gotten way more personal and two, if OpenAI has a partnership or has lower risk with a site or a service that that is probably gonna have some likelihood it's gonna be included in results versus something that they have a higher risk for. Is that right?
- [12:36]
Unknown
Yeah, I think we've seen that in the Data about which LLMs prefer which sources. If you map those preferences by the media and content sites that each LLM has partnered with, you can see clear relationships. Like for example, Both Google and OpenAI have partnerships with Reddit and Reddit shows up very frequently for both Reddit and.
- [13:00]
Asia
Quora are the two biggest sites. Like, who would have thought Quora would be such a linchpin of. Well, I guess it is, because if you look at the data, what they're really relying on is user generated content. Quora was never able to make a real business out of its actual business. They probably are making a pretty good business out of selling training data. But they're the two biggest sites that appear in these LLMs. And then as you're saying, so like the example there would be, if you go to ChatGPT, you're likely not going to find anything from the New York Times because they have an upcoming court case. And I do agree with you, Kip. I'm not sure if OpenAI are doing that good of a job at avoiding copyright lawsuits and they're going to have a lot of. I'm saying that they're kind of your influence.
- [13:37]
Kip
They're incentivized to do that, is what I'm saying. And more often or not are probably going to at least try to do that, is what I'm saying. The follow up question I have to both of you is can you have an effective AI search optimization strategy without engaging on communities like Reddit and having a strategy on communities like Reddit?
- [13:56]
Asia
No, I think it's a good question, but I think the setup here for Asia is okay. I'm a business. I can spend my time on the blue links, which I just want to say, even though I said from the outset the blue links are going to die, the best times in my marketing career were messing around with the blue links.
- [14:10]
Unknown
Everyone.
- [14:11]
Asia
That was like a good time.
- [14:12]
Kip
Are you blue link nostalgic? I'm going to mess with the nostalgia part of blue links.
- [14:16]
Asia
I think we're going to miss the times where you could really scale an engine that was predictable you could generate a ton of revenue and you could do it in a way that actually you could have like limited amount of resources but pretty good knowledge and actually make some real impact. And I think that's going to be a tough thing to.
- [14:38]
Kip
I think you're being a little cure and negative on this one. Go ahead, Asia.
- [14:41]
Asia
I'm not being, I'm not being negative. Let me finish the setup. Right, so like I think that the blue links go away for informational content. We can talk about that. But like you can spend your time optimizing for blue links and we should talk about what, if anything, the person should do there. Then they can optimize for Google AI overviews. So we're going to talk about this. AI overviews are becoming the prevalent way for Google to start to surface up search. They said that AI mode is going to be the default mode and lots of SEO people went on X and LinkedIn and cried a lot because that is a tough thing to navigate as well. When an AI overview apparently appears for a search result, the number one blue link gets around 33% less traffic. So there's some real lowering of the traffic. And then you could optimize for LLMs. Right? So you have your search traditional blue links, you have your AIO view, which is Google, and you have your LLMs. And so what are you telling people? Why do you think I'm overly negative around the blue links? And what are you telling people to do, Asia?
- [15:35]
Unknown
I understand the nostalgia for the blue links. I spent some time in the grief period as well. Here's what I'll say.
- [15:41]
Kip
This is hilarious.
- [15:42]
Unknown
I used to daydream and this probably gives you a peek into my psyche about what it would be like to go back to the early days of SEO and just have an open field of experimentation. I will be totally honest. In the past couple years, SEO had gotten pretty stale. We knew what worked and we just did what worked. And it was really predictable and it was really boring. And now we have a brand new open field again. And as everyone is starting to figure things out, that means that no matter where you're starting from, you could be a startup, you could be a big company like HubSpot. No one has a big edge right now. The edge goes to the people who figure out the new tactics the most quickly. And I think a lot of it is going to rely on AI powered content systems, which as y' all have talked about on this podcast, pretty much every time you talk means you don't need a ton of people. You don't need a huge team or a lot of money to get these systems going. And so I think that people in 20 years are going to be nostalgic for this period.
- [16:43]
Asia
Yeah. So I think Kip and I agree with this. I agree. I would have swap out SEO with marketing. I thought marketing was really boring, actually.
- [16:49]
Kip
I was so bored it was really driving me crazy.
- [16:52]
Asia
I was gonna go and open a coffee shop. I used to tell Kib about that. I was like, this is boring.
- [16:56]
Kip
I mean, we did have a lot of conversations about like, oh, maybe we should just like do something different. It's a little boring right now. I agree.
- [17:02]
Asia
I'm all back in now because it's all being rebuilt.
- [17:05]
Unknown
Yeah. So back to the core question of all right, well, I have a bunch of different audiences or platforms. What do I optimize for? If I was a business starting with very little organic traffic, I would be all in on LLMs because I think the amount of time that it would take you to establish an SEO strategy that works, by that time, all your Personas are using ChatGPT or Google's AI mode as their core way of getting information. If you have an existing SEO infrastructure and you have a lot of traffic to protect, then this is not a flip switch scenario. I think you need to protect what you've got while investing and scaling your AI search optimization tactics over time. And the best AI search tactics will not be bad for Google and I think that that's a little counterintuitive. But if you are doing AI search well, it will not hurt your Google rankings and hopefully it will even help.
- [18:17]
Kieran
Hey everyone, if you're like me, you love learning from the best and you just love learning, period. And a podcast that is all about learning is one that I love. It's called Billion Dollar Moves. It's hosted by Sarah Chen Spellings. It's brought to you by HubSpot Media and Sarah's awesome and she covers really what it takes to build a remarkable business. She's a venture capitalist, she's a strategist. She has really hard questions and wants to find the highs and the lows from the most successful builders and entrepreneurs of our time so that we can all learn from them and make our own billion dollar moves. She's got a great episode around the Big Bet mindset and how you really build bold change for your company. It's awesome. You're going to want to check that out as well as all of our episodes are awesome. Highly recommend. You can listen to Billion Dollar Moves wherever you get your podcasts.
- [19:10]
Kip
I think that is one of the most important points of the entire show is that these are not mutually exclusive things is what you're telling everybody, right? It's like if you're really focused on AI SEO, you are going to get LLM and AI overview traffic, but you're also going to get some traditional blue link traffic from those efforts. Exactly the one thing I want to tee up before we get into the deep how to on everything because I want to hear from both of you because I have a controversial take because we're talking about how this is a new era and there's a bunch of new opportunities. And I think what stops people from taking advantage of new marketing eras opportunities arbitrage is because they look at the world in the old way versus the new way. And one of the things about the old way is that visits were highly correlated to customers and revenue. Because what would happen is you had much more control of your visits through advertising, through traditional Google search, and your conversion of those visits to customers would stay pretty flat, pretty consistent, right? You didn't have as much control of the conversion rate. And so it's like, oh, I know that 1% of my visits are going to convert into customers. I just got to go get some more visits. Was kind of the old way of thinking of it. And I think everybody watching the show and I get emails from founders, financial analysts, all these people like, oh, what are the visits doing here? And I want to come out and be like, I think visits are fundamentally less important and in some ways unimportant. And that's like the hottest take I've had in a while because of the new ways people are discovering sites and brands. And I want to see if you all agree or disagree with me there.
- [20:46]
Asia
I want to hear from Asia. Given that she has spent a lot of her career 10x in visits to websites.
- [20:54]
Unknown
Talk about something I'm nostalgic for. No, I completely agree. And not just because, you know, that would be easier for me if we all felt that way over the next few years. Here's the thing. If we go back to what we were talking about earlier, how someone was going to visit HubSpot many times over the course of a few weeks to a few years before they eventually convert. If we say all of those interactions are now happening in LLMs, but they're still eventually coming to HubSpot and buying, then yeah, the value of an individual visit matters a lot less. What we care about is that ultimate visit, that visit where they decide to become a user or a customer. And what we what HubSpot's growth team is increasingly shifting to is visibility in LLMs. How often are we showing up as the recommended solution for the questions for the challenges that our Personas have?
- [21:53]
Asia
I think this is a great segue into a couple of things. Why you should listen to the next segment of the show because Asia and her team have done a pretty great job of making sure they're kicking everybody's ass. Let's be real, that's not really visible. So what are we looking at here? We're looking at one of the metrics from the tool we use to measure our awareness in these LLMs. And I think one of the things that will jump out of you straight away is share of voice. This is InquoteUnquote, AI search optimization tool, and that is a brand metric. And I think that's one of the things to wrap your head around is that you have to kind of think much more like a traditional brand market here. You're looking for impressions, you're looking for visibility, you're looking to make sure that your brand is more visible than other brands in these LLMs. And maybe just walk us through this chart, Asia, just.
- [22:40]
Kip
I'm representing the comment guys and gals. They'd be like, what tool is this? What tool is this? What are you showing? Yeah, we just need to give people a little information because there's going to be questions.
- [22:49]
Asia
Why don't I pass it back to you, Asia? You can kind of jump into the tool and maybe talk a little bit about how we use it.
- [22:54]
Unknown
So this is xfunnel. There are a bunch of tools out there that will show you LLM visibility and which websites are being cited in LLMs. We chose X Funnel because the founder and his team, they're really experimentation, centric, and I think that this era is all about experimenting as quickly as possible, learning as quickly as possible, and then feeding those learnings back into your playbook.
- [23:21]
Kip
Hey, Asia, on the tool selection.
- [23:23]
Unknown
Yeah.
- [23:24]
Kip
Because you and I have had a little bit of conversation on this because we're going to get questions about it. There's a bunch of people doing it X Funnel hall, like There's literally like 10 companies doing this.
- [23:33]
Unknown
Yeah. If not more.
- [23:34]
Kip
The differentiation I have been told from you is that it's pretty low and like, you should pick the one that you just kind of like the best and you think is going to evolve at a fast rate, which is one of the reasons you picked X Funnel. But, like, they're all using kind of the same approach and kind of the same data. So it's not like that there's one that is far and away better than everything else at this moment.
- [23:53]
Unknown
Yeah.
- [23:54]
Kip
Is that right?
- [23:54]
Unknown
There's no moat in this space at this moment. I think if someone can figure out how to really simulate all the context that an LLM has about a person or a user when they're talking, that would be a real competitive advantage. If someone can nail the volume piece. Thinking about, you know, what we were saying around, it's getting a lot harder to measure. We have a lot less direct visibility between what made someone convert. If someone can figure out, okay, the number of times that a question is being asked in LLMs, that would be a huge competitive advantage.
- [24:29]
Asia
But, yeah, I think the tool that will win here, and I think it's only doable if OpenAI or these kind of models decide to have a partnership, is the tool that can actually log in with your ChatGPT and integrate your memory, because then they can mimic the memory in the tool. And I think that's the big thing is like, because memory is just drastically changes what our results are going to be. I don't think it's a business model that they are going to particularly want to get into. Unless the only kind of wrinkle here is if they do a freemium tier and they have a paid platform, then they will give optimization tools and I don't know if they get. When they give those optimization tools to businesses, will there be something there that would help you better track visibility across that freemium tier?
- [25:09]
Kip
Look, I would bet my house that they're going to do ads and there's going to be some analytics around ads is what I think you're saying, Kieran. Right. And that will probably be a next important step of the journey of understanding of what needs to happen.
- [25:22]
Unknown
Yeah, I am very much looking forward to getting some data. Google has started introducing ads into AI mode, and we are a Google advertiser. And so we should start getting some data from, from AI mode, which I'm, I'm excited to dive into. Yeah, I think the memory piece is really hard. And what Veer and the team at xfunnel have done to try to simulate that as much as possible is forcing the LLM to think as a Persona. So as a VP of sales, as an enterprise AE, we have about 10 Personas that we've loaded into here across our core products. It's a very crude solution, but, Kieran, I think it gets to what you're talking about. So, yeah. So over here you can see our visibility across all the LLMs that x funnel tracks. I would say we're really focusing on Search GBT, ChatGPT and Gemini right now, because that's where we see the majority of usage happening.
- [26:25]
Kip
Tell people what Search GPT just so that people. Because I think that's not a common moniker and so I'm just trying to represent the viewers here.
- [26:34]
Unknown
Sure. SearchGPT is the searching system that ChatGPT uses, so you can have regular ChatGPT responses where the web is not actively being searched and then you can layer in search. Now, I think as the LLMs get more and more sophisticated, you probably won't have a choice about the merge. Exactly. So this we will probably, at some point just see as ChatGPT. This will probably just be Google. And then what we were looking at a little earlier is HubSpot. Share voice. Yes, Kit.
- [27:06]
Kip
I feel like I'm at school. Sorry, I'm just trying to ask all the questions. Does Gemini include AI overviews in Google or is that actually just like I need to be in the Gemini experience?
- [27:17]
Unknown
Fantastic question. Just in the Gemini experience. So we actually use a completely separate tool.
- [27:25]
Kip
This is what I'm getting at.
- [27:26]
Unknown
Yeah, we use a separate tool to track AI overviews, which are those AI generated summaries at the top of a serp. You don't have to be logged into any special experience or really do anything to opt into that. You're getting it by default. And the third Google experience is AI mode, which users in the US are now seeing as a tab on their Google experience. At some point, AI mode will become the primary Google experience. That's going to be like a 7.7 earthquake on the marketing Richter scale.
- [28:05]
Kip
Yeah, yeah, you're probably underselling it.
- [28:07]
Unknown
In actuality, how far up does it go? Like, are we talking.
- [28:11]
Kip
It goes to 10? It goes to 10. Yeah.
- [28:13]
Unknown
All right, we'll say it's a 10 earthquake. It's going to be real. Yeah. So, but I think the question under the question is how much can we learn just from looking at Gemini? And Gemini is the system that powers AI mode. It powers AI overviews, and it obviously powers the Gemini specific experience. But it is only a. A piece of the puzzle right now. I think we should just, you know, blow this up, print this out, hang it on our walls. Right now, HubSpot has the most share of voice for the queries and the Personas that we are tracking. Now, if I was going to be skeptical about this chart, if I was Kieran? I would say, well, how are those queries chosen? Is that for every conversation that HubSpot could ever show up in. And we hand selected these queries, we said, okay, these are all the queries that we think our Personas are asking, which is obviously pretty different from how SEO works where you can literally go into ahrefs or search console and see the actual queries that people have asked and how much volume is behind them. So super imperfect. But I think that this tells us that we are definitely on the right track for the topics that we care about.
- [29:30]
Asia
I know you have a point here as well. We should just quickly touch on this because this is like a version of a aided slash unaided awareness study where you have to kind of guess the survey questions. Like it's just so much different from what we were able to do in search where you could see the keywords and just figure out what are the ones you should actually go and look at to see if you appear. And the other thing is, and we're going to, I think you're going to give some guidance in this, in the AI, when you get into like how you actually manage to do, do this. How did you manage to get HubSpot to be so visible? The actionable takeaways is people, how they ask questions in AI is going to be so much different than search. Like search taught us to ask in key phrases. Now AI is just how you conversate. That is hugely different in trying to predict how people talk to another person, that's how they treat AI.
- [30:17]
Unknown
I think it's different not only in how they talk to the LLM, but in, in terms of the type of content that we need to create. Maybe this is a good segue into those tactics.
- [30:28]
Kip
Yeah. I would just say as we transition before traditional Google, you would search, something wouldn't be quite right and you would keep refining your search. Right. And so one search would end up becoming 20 searches. Right. Where now like you have one search in ChatGPT, it asks you some clarifying questions and it gives you like a curated, actionable response. And that's just a holistically different experience. And I think what we're trying to tell people is that experience is different. How you measure that experience is insanely different. And you need to start aligning your company to look, think about measuring this differently. And I like what you were just showing there with like the share of voice stuff. Asia, is it perfect? No, but I look at it as like it's you're defining where we're Playing here are like the types of conversations we want to play and are we winning where we're playing or not? Like, it used to be an SEO, it's like, oh, we're play everywhere and we're going to win. It's like, no, we have to really focus a little bit more. And are we winning where we're focusing? And I think that's a big evolution in how this work is happening.
- [31:28]
Unknown
Yeah, absolutely.
- [31:29]
Kip
All right, let's tell people what they should do.
- [31:31]
Unknown
Okay, so I think the first big thing, it used to be that the quality of your content was why you won. Now I think it's about the specificity of your content and that doesn't mean that quality doesn't matter. But if you think about how much more context someone is able to give in AI search, they're able to tell ChatGPT all these things about them, their business, their needs. Now you don't just need an answer for what is the best X, you need an answer for what is the best X. Considering that I am a manufacturing company in New Jersey, we've been around for 10 years. Our growth has recently plateaued. We're spending a lot of money on paid ads. And so that is an excellent explosion of content for every Persona and every problem that you have. And it means then rather than thinking about, okay, how do I create these one to three really good pages on these head terms, you need to think about, how do I create these 100 to 300 great pages on all of these super long tail terms? And no one, including HubSpot, has the resources to do that. If you have a blogger sitting at their keyboard typing out every word, which means your job as a growth marketer, as an SEO, as an aeo, whatever you want to call yourself is okay, how do I build systems for great content? How do I build AI powered systems that will produce really great content at scale that is highly specific?
- [33:06]
Kip
I think that last minute was probably the simplest articulation anybody can get around what is changing and what you need to do differently. And what you're really also telling people is like, you're going to need AI to solve the AI search problem. You cannot solve the AI search problem with manual old school work.
- [33:25]
Unknown
Yes, exactly.
- [33:27]
Asia
So I'm a traditional search person. I hear you say that. And I think two things. How do you make sure that content is unique and does it matter? Right? Like, does it matter? Like, if you just change one data point for each industry and then have thousands of pages where every page had like a singular data point so how different do the pages have to be? And one of the things you said I think was really good for folks earlier on. The things you have to do for AI search will not be a drag on your traditional search. Now I could say that is going to be a drag because we always hear that Google doesn't like, you know, paper thin content or lots of duplication of the pages. So maybe talk through how you think about that.
- [34:06]
Unknown
I think if we were in a world where we didn't have to care about Google at all, then you could experiment with changing one data point on thousands of pages and see what happens. There are a lot of indications that LLMs care way less about what we've historically thought of as thin content or doorway pages. Basically the skill content idea, we know Google cares about it and we know that Google still matters. So I would not just publish a bunch of pages with one thing different. What we are doing, we are publishing very unique pages with a lot of original data. And I think that original data is the key to making this work.
- [34:51]
Kieran
Yes.
- [34:52]
Unknown
Now that original data doesn't need to be this huge study of, you know, something relevant to your Persona across the United States doesn't need to be big or expensive. You can use the data that you have on your own customers, which is probably where you should be grounding all of these content efforts to start. So if you know that using your product or service, that manufacturing company in New Jersey is likely to have 3x higher close rates because you have seen manufacturing companies get that type of return. That's exactly the type of data point you want to bring into this content. And so start with your original data on your customers and use that to populate these pages.
- [35:34]
Asia
Another important point, I think one of the important things for people to take away here is you're also talking about content that is really mapped to. You're in buy in mode. This is not informational content. This is not like, how do I teach you something and do thousands of pages you're specifically talking about? Hey, you're already in that buy in process and we're gonna create lots of different variations of our pages that would help you understand if we're the right solution for you. Maybe talk a little bit about that. Like, do you buy the fact there's little to no point anymore invested in the informational educational content and instead you spend all your time on content that is much more mapped to someone who's in buying mode?
- [36:16]
Unknown
I think there's a way to do the informational content, but I think everyone Including HubSpot should start with the bottom of the funnel content and work your way out.
- [36:23]
Kip
Which is the opposite of ten years ago.
- [36:26]
Unknown
Exactly. You've always had the most top of funnel content because that answers the broadest number of questions. And that's again where everyone has been starting their journey. And then you would have the least bottom of funnel content. And now I think it's flipped. I think you want the most bottom of funnel content. And yeah, Kieran, I think it's a great call out because we went to data and it's probably hard to conceptualize where that data would even come in before you understand the type of content that we are talking about. We are talking about content that says here is why this product or service is the best solution for this Persona with this context. And the idea is that when an LLM is formulating an answer for a user, they are pulling on that content so they can say here is why this product or service is the best for your problem. Yeah. And that's when you get that visibility in the LLM4 the queries you really care about.
- [37:20]
Kip
The way I think about this is you went from starting at the top to starting at the bottom. The leverage in AI search is around the breadth and ultra long tail of product related information and queries. And if you don't have a lot of public product information, pricing information, buying action data and information, you're not going to be successful. And so you're going to need to change your business. There's at the top of the funnel, we've gone from teaching people how to do things to needing to give them tools to do the things themselves. And Kieran, you and I have talked about, we're going to do a whole tools episode. We're going to build a couple mocks of if you were like the average B2B business, what are some tools that you would try to build and use to capture people maybe a little further up the buyer's journey?
- [38:07]
Asia
I do want to talk a little bit about this as a segue into we should talk about citations because that really is your on page strategy. One quick clarifying question we should get into there. Is there any optimization in the structure of those pages we should quickly talk about. But the thing I wanted to quickly reference on the citations was because we're talking about it here is I was doing a ton of research on putting together this whole what we know about LLMs, everything you could want to know about how you can appear in LLMs. And so one of the questions I've Always asked, I go back and keep asking is like, what are all of the signals that have been wrote about that would actually tell us how LLMs recommend products? Like is there anything be released? And like why an LLM recommends products, a certain product. And it gave me back this answer. And I do think ChatGPT is starting to reference websites much, much more. And so in every answer it gave me, it had these websites referenced. There was one website I referenced several times that kind of really stood out to me. It's like, wow, this website has got itself in the answers a lot for this informational query. And I went to it and it was like a pretty newish website. It's actually an agency who does AI search optimization. So kudos to them. It's called Growth Marshall. I don't know anything about them, but the reason it really stood out to me is like because of this chart. It is a brand newish site and it didn't appear in Google searches for those same queries. And what this is telling us here, the chart we're looking at is the ranking positions of LLM cited search results. I think this is pretty interesting in that these AI search assistants are referencing sites that are not in Google's top 20 results.
- [39:37]
Kip
Can you explain this chart? I don't think this chart is obvious to everybody. Explain this means because this is a really interesting chart.
- [39:43]
Asia
Yeah, it's basically saying ChatGPT search primarily cites lower ranking search results and then.
- [39:49]
Kip
What a traditional Google search, right?
- [39:51]
Asia
Yeah. The ranking position in Google of the sites that it's pulling into its resources and you can see here the majority are outside of the top 20. And I think that's fascinating.
- [40:02]
Kip
So fascinating because we have been taught.
- [40:04]
Asia
Google works on page rank trusted sites. So what is ChatGPT doing differently where it is referencing sites that are completely different than how Google think about Trust? Like obviously ChatGPT think about trust in a very different way. I don't know if you have insights or not, Asia.
- [40:20]
Unknown
I think it's all about specificity. I think that these websites that are being cited are targeting much more specific queries that have historically made no sense to target interesting. And that's why they're getting pulled in. ChatGPT does not care about site authority.
- [40:34]
Kip
I think what's interesting is this minute before it's into the show is the glimmer of sunshine, the glimmer of hope in this whole episode. It used to be if you were starting a business or you wanted to really change the trajectory of your business, somebody would come to you Asia. And you'd be like, okay, well, all right, you gotta set up a website, you gotta create all this content. You gotta hire these people to create content. It's gonna take you six to eight months to build your authority. And I think in a year we can get you to this level of traffic. Is that roughly Right. And now I think we're saying, hey, you can start a new site, have a very focused strategy for LLMs, and you start getting traffic from LLMs in days.
- [41:12]
Unknown
Yep.
- [41:12]
Kip
Which is wild. Like, completely different than like, we've been used to thinking about this problem.
- [41:18]
Unknown
Yeah. Are you still pessimistic, Kieran, or has this raised your spirit?
- [41:23]
Asia
No. Like, I want to be really honest, I would not be doing this role if AI hadn't happened. I was going to do something differently. I just had got pretty tired with the same old thing. And so I'm a person. I think, Kip, you're very similar. I need new stuff to go after and figure out or there's no point in me doing it. Right.
- [41:41]
Kip
Correct.
- [41:42]
Asia
And so I couldn't be happier with what's happening. But like, my nostalgia for search was more. I was doing search when it was being figured out and I just liked the fact that I could exactly see the effort go in and then the money come out. Right. And I think that's just the only nostalgia I have is like, oh, I do this thing and I get money coming out of here. But I just want to make sure we end on one thing because I know we're going to get screamed at if we don't. I do want you to tell people how they should think about citations because you've told them, I think a really great takeaway on, on page. But what is your off page strategy? How do you think about what brands should do to actually, you know, if we believe it's not all on your website, what are the things you should do to get mentioned off site and how should they do that?
- [42:21]
Unknown
Yeah, I think you want to be repeated as often as possible in your core brand terms. So we have several at HubSpot. But let's take CRM as an example because that's always been a really important association for us. We want to get as many mentions, as many positive mentions as possible across the websites that LLMs are pulling in as HubSpot as a CRM. And in the old world, for that to be valuable, it needed to be a backlink. It needed to be sending people directly to our website. But we're just trying to train the model that HubSpot is a CRM and ideally that HubSpot is a great CRM. So we don't really care about the backlink, we just care about the mention. And if I were starting from scratch, I would use a tool like xfunnel. I would figure out the websites that are being cited in LLM responses and I would reach out to those websites and see if I could improve the number of mentions I had for my product for my key terms.
- [43:24]
Asia
And we'll end on this. Mentions doesn't matter if it links right. It's just like co citations words and you want to make sure those co citations include the thing you want to be ranked for. And so one last question. You mentioned lots more pages in Google's index for very specific queries. Do you have to have that level of specificity in the co citations? Like do I have to have HubSpot CRM from manufacturing company in Ohio?
- [43:49]
Unknown
In an ideal world you would, but I don't think that that's possible for anyone to do. So I think you stick with the head terms, you stick with the big ones.
- [43:56]
Asia
Like I'll give you a current dark days of SEO used to be in black hat forms. I still think that one of the takeaways for someone like Me in the ChatGPT thing is ChatGPT doesn't care about trust. So what I could do in the interim where it doesn't really care about trust because it hasn't figured that out or doesn't care about that right now, is I could like buy up websites, I could create websites and I could just inject a huge amount of specificity into its index and co site a product or a brand. And I do wonder how it would figure out not to like include those things in its trained model.
- [44:27]
Unknown
Yeah, I think it remains to be seen whether third party versus owned makes a difference. Yeah, because if it doesn't then you might as well just do it all on your own site and that's actually a lot easier.
- [44:37]
Asia
You would buy a bunch of third party sites and then just inject all your co citations into them.
- [44:41]
Kip
My sense is it would probably work for a little bit and then it would not.
- [44:43]
Asia
Yeah, I think it would work for a little bit.
- [44:45]
Kip
My bold prediction is that most companies will have account specific pages on their site, meaning like hey, I'm trying to market or sell to this account, this company and I'm going to have a lot of content that is specific to that company. I think there's some tests Asia that we're going to run around with this and so we can come back with that. The last, last, last question. Give everybody the recap that if they're like, hey, I'm bought in. I want to win this new AI search game. What are the couple of things they should go start doing now to be successful.
- [45:17]
Unknown
Okay. If you don't already know your Personas, you need to figure that out immediately. Then you need to set up AI powered systems to create highly specific content for those Personas at scale. You need to populate that content with as much unique firsthand data as you have. I'm realizing we didn't even get into the structure of a page that's optimized.
- [45:39]
Kip
We got to do a follow up.
- [45:40]
Asia
Yeah, we need to do a part two.
- [45:42]
Kip
Yeah, yeah.
- [45:42]
Unknown
Well, pay attention to all the stuff we're going to talk about next and let your CMO know that visits don't matter how they used to and that we need to buckle up for a brave new world.
- [45:56]
Kip
Yeah. Well, you have to change how you're measuring the performance. Right. And get a team and company aligned on that. And that's a big thing.
- [46:01]
Unknown
Yeah.
- [46:02]
Kieran
All right.
- [46:02]
Kip
Kieran, any words of optimism or pessimism before we rock out?
- [46:05]
Asia
Yeah, I'm incredibly optimistic. For the people who are deeply curious, there's like a ton of opportunity here. I think the thing that's going to be a struggle in the interim is the attribution measurement. I think that is a hard thing for the average company to overcome when they're trying to figure out how do I forecast what I'm going to get back from all of this. That's just not possible.
- [46:25]
Kip
I completely agree. There's going to be lots of things we could cover on this topic. We're going to do follow up shows, leave comments below on certain topics that you want us to go deeper on or do follow up shows. Asia, we're going to have you back on the show. We're going to cover a bunch more of this stuff. Thank you for taking some time to be on with us today and good luck to everybody out there in the AI search game. We'll see you real soon on the next episode of Marking against the Grain. Alert.
- [47:06]
Kieran
Alert. We have a major announcement that you need to know. We've got some big news. HubSpot CRM and ChatGPT from OpenAI have come together for an amazing integration. HubSpot CRM is now the first CRM to integrate with ChatGPT Deep Research. Bring all of your customer context and data from HubSpot into ChatGPT now you can do truly remarkable things to better grow your business. Marketing teams can find the highest converting cohorts and create tailored nurture sequences and then use those insights to launch automated workflows right inside of HubSpot. Sales teams can effortlessly analyze deal pipelines, identify at risk opportunities, and receive AI guided suggestions for next steps to close deals faster. This is a game changer. I think this is one of the most important things that has happened to marketing and sales and customer service in years. The HubSpot Deep Research Connector will automatically be available to all HubSpot accounts. Free, paid, all of them. If you use HubSpot, you have access to this and it's available to ChatGPT team, Enterprise and. Edu subscriptions. Turn on the HubSpot Deep Research Connector in ChatGPT to get powerful PhD level insights from your customer data.