Aimee Jurenka breaks down how SEO is evolving in the age of AI search. From topic entities and semantic signals to prompt tracking pitfalls and brand visibility measurement, this conversation reveals what marketers need to rethink now.
https://page2pod.com - In this episode of the Page 2 Podcast, Jon Clark and Joe DeVita sit down with Aimee Jurenka, SEO and AI Visibility Strategist at Rickety Roo and founder of seo SUSTAINABLE, to unpack what SEO looks like as AI search, LLMs, and answer engines reshape brand discovery.
Aimee explains why modern SEO strategy is moving beyond traditional keyword targeting and toward topic entities, semantic associations, brand positioning, and what a company wants to be known for. She shares how marketers can measure brand visibility inside LLMs, why one-off prompt tracking can be misleading, and how SEOs can use competitor comparisons, branded queries, and content gaps to improve visibility across AI-driven search experiences.
🔍 In This Episode
• Why keyword research is becoming fuel for broader topic entity strategy
• How to think about what your brand wants to be known for in AI search
• The difference between what LLMs know about your brand and what they associate you with
• Why single-shot prompt tracking can create unreliable visibility data
• How branded search, mention rate, and competitive context help measure AI visibility
• The role of semantic triplets in clarifying brand identity and positioning
• Why side-by-side comparison content can outperform generic listicles
• How small businesses can learn from larger competitors without copying them
• What SEOs should understand about AI crawlers, bots, and technical accessibility
• How tools like ChatGPT, Gemini, Claude, Perplexity, NotebookLM, and Screaming Frog fit into AI visibility workflows
This episode is packed with practical insights for SEOs, content strategists, and marketers who want to build stronger brand visibility across both traditional search and emerging AI answer engines.
Subscribe to the Page 2 Podcast for more conversations with SEO leaders, AI search strategists, and digital marketing experts.
What do you think is the biggest shift in SEO right now: AI visibility, entity optimization, or brand authority? Drop your thoughts in the comments.
🛠️ Tools & Resources Mentioned
• Aimee Jurenka on LinkedIn
• Sitebulb article referenced in the episode
• Rickety Roo
• seo SUSTAINABLE
• John Mueller’s Q2 news mention
Welcome to the Page 2 Podcast where we uncover the strategies, systems, and tactical decisions that move brands beyond page two and into real visibility across search and answer engines. Today, we're talking with Aimee Jurenka, SEO and AI visibility strategist at Rickety Roo and founder of seo SUSTAINABLE. Aimee has been doing the kind of work a lot of marketers are still trying to define, building workflows, testing AI visibility strategies, and figuring out what it actually means for a brand to be known by a model.
She also recently got a very public nod from John Mueller, which turned into one of those rare moments where the internet validates months of very nerdy, very useful experimentation. Aimee walks through why keyword research is becoming less about individual pages and more about topic entities, associations, and what a company wants to be known for. She talks about measuring brand visibility in LLMs, why single-shot prompt tracking can create misleading data, and how mention rate, branded search, referral traffic, and downstream conversions may become part of the new measurement stack. But she also makes the case that old content, site structure, schema, semantic HTML, and even crawl logs may still have real value if we reorganize them around how AI systems retrieve and understand information. I really like this episode because it gets past the vague AI is changing search conversation and into the operational questions. What do we measure? What do we build? What do we stop doing? And how do we explain all of that to a client who just wants the phone to ring? If you learned something new today, take a second to subscribe to the Page 2 Podcast. Leave us a rating or review and let us know what resonated. We'd love to hear your thoughts. Okay, let's get into it.
Jon Clark (01:43)
Welcome to episode 119 of the Page 2 Podcast. I'm your host, Jon Clark, and I'm joined as always by my partner at Moving Traffic Media, Joe DeVita.
Joe (01:51)
Hello.
Jon Clark (01:52)
We're excited to welcome someone who's been doing some really fascinating work at the intersection of SEO, AI, and workflow building. So I'm really excited to get into this one. She's currently the SEO and AI visibility strategist at Rickety Roo and the founder of seo SUSTAINABLE, Aimee Jurenka. Welcome to the show.
Aimee Jurenka (02:08)
Great to be here.
Joe (02:08)
We got so many questions for you. There's no way we're going to get to the end of them. But I want to first ask by maybe you were surprised a couple of weeks ago when John Mueller mentioned you in his Q2 news. Did you know that was coming? If not, who's the first person to say, my God, you're famous?
Aimee Jurenka (02:13)
No, no. I actually spend a lot of time on LinkedIn. So when I decided to start learning about AI workflows, AI in general, AI search, I did the thing where you delete all your social media so you can spend time doing that. Well, that left me with LinkedIn. So now I spend a lot of time on LinkedIn. I was actually going through and reading the notifications and I was like, wow, that's weird. I was like, well, that can't be right. I'm just not looking at that right. I must just not be understanding something. And so then I watched the whole video and was like, my God, I can't believe it. So yeah, that was quite a scream, cry, running around the room, very excited moment to get that sort of not only visibility, but sort of that validation that working on the big picture stuff is really cool. It was amazing. It feels good, that's for sure.
Joe (03:13)
Yeah, I think it was a highlight of the quarter, or maybe even the year. But the article, the Sitebulb article, maybe we can start with that because it was great. I mean, it was a great article and we'll link to it in the show notes, but I just wanted to maybe start about the strategy and the research that you do before we develop a strategy. So in the article, you talk a lot about entities, traditional entities versus topical entities. And I wonder, because search engine optimization is still important too, there's some keyword research you have to do, some entity research, and you've got to come up with kind of a target list for both. What are the keywords that are important to me? What are the entities that are important to me? Can you talk a little bit about the approach you take to doing that research together?
Aimee Jurenka (04:01)
I'm very much a holistic SEO. So those keywords that you're looking at, those turn in. Those are fuel for those topic entities. While I'm getting to know a site and seeing what the site structure is, seeing what they're all about, you go ahead and you meet with brand and product. You see what they're doing and then that's how you figure out their core products or services. From those core products or services, you start asking, what do you guys want to be known for? What do you want to be related to? That keyword research feeds that. So that's that information that goes into that. Hey, in traditional SEO, here are the keywords and the type of things we want to be known for. Here's what we are known for. Here's what we want to be known for. How does that relate to what you want to be known for? Do those align? Do those not align? Are these all content gap opportunities? How do we look for that?
When I'm thinking about AI search, which is fast becoming search, because Google's shoving it down our throat no matter if we want it or not, I'm taking a really big step back. As you're doing that keyword research, that is purely research. We are probably not going to write, for me, content based on a keyword or a keyword phrase or anything like that. We're going to be taking those keywords, looking at the bigger picture, grouping them as a bigger picture and saying, okay, what was the topic entity about this that this fits into? So it's more like strategy down instead of keyword up. Once you have those topic entities, and of course you have your brand in there, you have your products in there, the same entities we're used to.
But once you have those topic entities, then you're able to go, well, okay, what kind of people are buying this and where are they at in their journey? What kind of content do we want to write? And so it's going to be a lot more based on the person and who we're trying to reach in that market because AI is so big. We can't cast a wide net like we did with traditional search. So we have to get a little bit more specific in there. And so that's how I'm going to be looking at that. The idea of keyword research, get our list of keywords, see what our highest opportunity is, we go and we write one piece of content on that, in my world is going to be gone.
Jon Clark (06:04)
There's another concept in the article around exactly what you're saying, transitioning away from traditional keyword targeting to more like association. I think in some ways to do that, you need to know what the models know about the brand as well to close some of those gaps. How are you doing some of that measurement? In other words, understanding what the models know about the brand today and then identifying what those things that the model might need to know are. Are you looking at competitors? Are you looking purely at the brand itself? We had Melissa Popp on earlier this season and she talked a lot about personas and all that sort of thing. So I would assume there's probably even some persona analysis there too.
Aimee Jurenka (06:52)
Yeah, definitely. When you get into how you identify where those gaps are or what's being said about you and what you want to be said about you, like I said, you first start back at the strategy. What is brand, what is product? What do we want them to know about us? Then there are different ways of being able to take that temperature. I like to do a big list of what I call branded keywords. That's going to have the name of your brand in it. Those are going to be the things that when you're specifically asking about your brand, you want to get certain responses back. You want to know what's been done with that. Then there are your entities, your topic entities, your associations. Those are not going to have your brand in it. That's going to be the "best CRM for blah, blah, blah." And then that's where you go and you start to see if your brand is being mentioned.
That's where you get those two sides of what do they know about me and also what do they know in general? So if I asked Gemini who Aimee Jurenka is, it gives a pretty good list. It says certain things. It actually gives me credit for things I haven't done. So thank you, Gemini. But if I asked who one of the leaders in AI visibility strategy is, I don't come up. So those are where you see those big gaps. How I measure it, 100% I look at what your competitors are doing. Who are your competitors? Again, I do things a little bit differently. So not only like who's a direct competitor, but also what is like an aspirational competitor? Kind of the shoot for the moon, hope to get a star idea.
If you're a small e-comm client that deals with mountain gear because you own a mountain shop at the bottom of Mount Vernon out there, then you're going to be looking to do what REI does. Because that's what your e-comm is about. Are you going to ever compete with REI? No, not at all. But we are going to want to see how REI structured the website and what they're doing that's working to see what we can apply to your small e-comm site that does mountain gear and skiing and whatever. So I do the competitors that way. We have your direct competitors and we have your aspirational shoot-for-the-stars competitors. And then that's how we start to get an idea of, okay, well, if AI knows this X, Y, and Z about your direct competitor, but they don't know X, Y, and Z about you, then we have that gap.
Or wait a minute, we're seeing where we are different. They know X, Y, and Z about this competitor. We don't offer those products. We're not in that market. We're not working for that clientele. That's my favorite. So it's like, bingo. Instead of doing a top 10 listicle, we can do a side-by-side comparison. And when we do that side-by-side comparison, we can specifically say we are not for this kind of person, those competitors are great for that, we're for this sort of market, we're for this sort of person. And then that narrows it down. So when people are doing that personalized search, then that's a really good thing that the LLMs can pull in. This is exactly who you're looking for. But we can also filter out that competition for those clients that we're trying to target. So being able to do that differentiation, not the big wide net and then try to move it through the funnel, but more of who are we specifically looking for and just starting to try to narrow that down as much as we can for the LLMs to let them know exactly who we are, what we do, and who we want to sell it to. That's what they're really looking at. But like I said, it's still going to be a very big picture because it's so big, they're taking it from everywhere. It's going to be very holistic. It's not going to just be your website. It's going to be so big in order to get that consistent information out.
Jon Clark (10:08)
Love the conversation around competitors because so often you'll start a project with a new client, ask who they see their competitors are, and oftentimes it's such a traditional set of competitors. Then you do searches for keywords that they want to be ranked for, and those competitors are nowhere, and it's a totally different set. It could be the aspirational competitors, or it could be a conversation with the client around maybe this isn't your target keyword that you think it might be.
But I was curious about the branded keyword selection. We had Everett Sizemore on the show last week, and he was talking a lot about semantic triples. Is that sort of how you're thinking about your branded keyword set, or is that sort of a different discussion with the client?
Aimee Jurenka (10:59)
I haven't thought about that in a while, but yes. Semantic triplets, I love them because I like the idea of being able to go through your brand and be able to define exactly who you are and what you do. I actually hadn't thought about applying semantic triplets to my branded keywords yet, but I'm going to now because that sounds really smart. It's also a great way to go over with the client or leadership and be like, here are our semantic triplets, is all of this correct and does it have all the information we are in? If the LLMs pulled this information, would this be the information we want people to see? So it's a really great way to go about doing that. Semantic triplets are just fun. There are a lot of ways you can generate a lot of them really quick and then you can narrow it down, you can edit, you can get inspiration from them.
That wasn't what I was doing, but I am going to do it now. I was doing a very generalized, "tell me all about Aimee Jurenka, tell me what you know about seo SUSTAINABLE, tell me what you know about if Aimee Jurenka does this." And I talk about myself in the third person right now, that's weird. So I've been really like, "tell me about the client, tell me what you do." So I've been doing really generalized, but like I said, I'm doing the triplets now. That's a great idea.
Joe (12:10)
You've written a lot about tough conversations you have to have with client leadership groups. It's like the same marketing executive who was buying an SEO service is buying an AI optimization service, but you have to convince them to shift their investment strategy for the next 12, 18 months. We're finding the hardest part of that conversation is in measuring success, but I wonder if you have a different take on this education part of our job.
Aimee Jurenka (12:41)
Yeah, because of inbound marketing, they all want traffic back. How are you going to get traffic back? Well, we're down on traffic. Well, we're just dying on traffic. You're going to have to give it up and that's the hard part. How do you explain to them that's not coming back? I found that it's really good to put a differentiator. SEO was about rankings and traffic. AI search is going to be about visibility and downstream conversions. So really making that strict line. These are two completely different things. Now the tactics might be the same. I'm an SEO moving into AI search. The people might be the same. A lot of strategies might be the same, but the customer journey or what we call the customer journey, like the funnel's gone. It got eaten by LLMs, gone, consumed.
So that whole idea of inbound marketing that everybody knew, from plumbers to CMOs, that is going away. Did you guys see that HubSpot changed it from inbound to unbound? I'm going to add divination to my LinkedIn skills. But they did. I was like, is this April Fools? Because it was right around that time, but they have. Because really what you're seeing is that is going away and clients and CMOs still want that. Traffic was what their measurement was. They understood keyword, content, traffic. And then it's our job to move it through the funnel. Someone with conversion rate optimization will move it through the funnel.
So going back and being like, hey, this is going to be brand new to you and we're not going to be getting traffic, so throw a funeral for that. What we're going to be measuring is that AI visibility for our topic entities, what we want to be known for. And then what we're going to be looking at is a lot of downstream conversions. We can get referral conversions in GA4, we know that, but the studies I've been reading say that a lot of people actually do their research in the LLM, get out of it, and then what they do is a branded search. So we're going to be looking for the correlation which we always hated, but it's coming back. Hey, we know that we increased our visibility for this topic entity that we know supports this product. What we've seen is an increase in organic and direct searches that have sold this product or have booked that demo. And so we're going to have to start looking at things that way because we're losing our attribution.
Jon Clark (14:53)
The old model of measurement was very straightforward and you could argue kind of easy: organic traffic, conversions, revenue, leads. And now it's hard. That attribution is tough. So if I heard you correctly, are you saying you're trying to correlate increases in AI visibility with increases of traffic on site, maybe that's direct, maybe it's organic also, maybe it's referral, and then sort of the downstream conversions based on that growth? Is that how you're thinking about marrying those things together?
Aimee Jurenka (15:27)
Yes, branded traffic. That was the key part is branded traffic is how I'm thinking about marrying those together. Because when I think about how the user is going through a user journey today, where they used to go, you put it in there, you get your list of 10 blue links, you'd make a list of five or six different things that you see there, then you'd research, then you'd come back. All these were different touches. All these were different times people could come to your website. And then they would get into the decision mode, and then that's when they're looking at your pricing sheets and doing different things. Then they might come back and actually convert. All that was our organic traffic and we could see that happening.
But now people are asking an LLM and they're getting a list of three or four things. They're just going with that list. They're not going out and looking for more different things. They're just taking that list. It reduces the cognitive load. And now they're starting to research that list directly in that LLM. So what can we track out of that? Not a whole bunch. We can track our visibility for what we want to be known for and we can track if we're starting to see either referral traffic or downstream conversions. I don't see a place where we have our magic marketing dust any longer that we can sprinkle on that yet. Things getting there.
Jon Clark (16:39)
Yeah, they announced some awesome stuff at SEO Week with some of the new data that they're going to be opening up for all of us, which is really exciting. I'm still waiting, Google Search Console, where are you guys? You had a great Earn the Stage presentation. I was really hoping to hear that talk. But you had a couple of great lines from the presentation. One was around "stop thinking like a rank tracker," which I think is perfectly aligned with what you were just talking about. So maybe to take that a level higher, how are you thinking about, I don't know if prompt tracking is the right word, but maybe visibility monitoring in the LLMs? Are there specific tools that you're using or workflows that you're running? How are you measuring within the LLMs themselves?
Aimee Jurenka (17:25)
All of that started with prompt tracking. I did what everybody else did. Okay, here comes AI search, so how is this different than SEO and how am I going to take what my brain understands about SEO and fill a slot over here? Rank tracking, prompt tracking. It seemed like it made sense. And then something just didn't seem right. There were just too many questions. How do you get that data? How do we know that's the right answer? How are you pulling things? How do we know that's going to be personalized search?
The thing that I landed on was because we don't have a SERP. So there's no single source of truth. In SEO there's a SERP, so we can roughly know that everybody at 10:30 on a Sunday night in San Francisco is going to see this result if they put in this keyword roughly. That does not exist in AI search. So I was like, well, how do we recreate a SERP? How are we going to do this in AI search? How are we going to roughly know what everybody's answer is?
I of course started clanking away like all of us are doing, NotebookLM, ChatGPT. ChatGPT, she's too nice, she gases us up. Let's go to Perplexity and ask, will this Claude agree? I'm very much that sort of person. I just love it. I was going around and around and I got all sorts of crazy formulas and stuff that I'm like, wait a minute, that didn't make sense, trying to find stuff to back it up. Finally, after I'd been clanking away at it for maybe six to nine months, I was able to put in the right question. I figured out what I wanted my prompt to be. And the AI was like, "yeah, use the Cochran 1977 statistical formula." It's like a standardized formula. It's the first page of every statistics book ever. If you take a statistics class, I haven't, but it is. And it's what they do when they're doing polling for voting. So it's how they get statistics for the world because I was like, somebody's already done this. There has to be somebody in the world that's already figured out how to take this large of a data set and then figure out some sort of statistically valid number.
I went ahead and just was like, perfect, checked it out, used that, and I applied that sort of theory and that method. I did a three-tier operation you can do. I'm calling it a mention rate. I don't have sentiment yet and I don't have accuracy yet. Hopefully, I'll layer that on someday. But right now, what you do is you put in your list of prompts, it runs it from an API, and then it takes that list of prompts. Depending on what you want your margin of error to be, it runs it a certain amount of time. So like if you want a 10% margin of error, you take 20 prompts, run those 20 prompts five times in a row, you take all those numbers, you aggregate it together. That's your mention rate for that topic, for that idea, for that entity. And so that's how I'm doing it. It's like a larger sampling of it. I did create a couple articles on LinkedIn about it since I didn't get it out there. I was like, gotta get this out there now, about to burst. And I actually created a mention rate tool on my website. It's not for bulk yet, you can't upload a whole bunch of different ones, you gotta do them one at a time.
I started running a test on it for myself for another thing I'm working on and it does respect the breaks in between that you're supposed to have. So I did it one step up where it was like, Oh my gosh, man, just trying to wait for that thing to go. It's not for the weak. Go faster, but it can't. It just runs and runs. As far as using that tool at scale, it's more of an inspiration. Get a hold of me. It's not going to be able to run things in bulk, but it is there for people to use independently for free. You just gotta have an API key. And it does back up that whole strategy. Again, my website talks about it a lot and where I got the idea from, why it's there and then the statistical value when it backs it up.
I am very, very much after this process anti-single-shot prompts. Okay, I got my 20 keywords or my 20 prompts and they're all about the same thing, let's run them once and then we take that score that one time you ran them and then we do that day over day like it's a rank tracker and we start to see how the line moves there. Because if you take those 20 prompts and run them that day, then run them again, you're going to get a different answer that second. So I feel like that's an imaginary number. You're just playing the slots, even worse than the slots because those are a little rigged, but you're gambling on what your results are going to be. So getting a large enough data set in order to feel like you have a statistically accurate number and then starting from there is where I want to go with it.
If somebody else comes up with a better idea, that's awesome. I know SparkToro did a big survey while I was kind of noodling this. I tripped, fell, broke my teeth trying to get into that because I was like, this is exactly what I'm working on, I'd love to be a part of this, I'd love to see how it works. And it came out and at first I was like, oh, they're saying prompt tracking doesn't work at all. But they said, no, as we pulled back, we saw that there were patterns when you ran it enough times. Garrett French just did one with Citation Labs starting to see the same thing. I'm working with somebody else who's trying to do it where you just do that single shot prompt 384 times because it's the same theory but just formulated differently. So if anybody else has ideas, I'd love, love, love to hear how people are trying to recreate the SERP in order to get accurate data for our tracking.
Joe (22:51)
I have this hypothetical question that there may not be an answer for yet. You're working with a company for a few months and they know their mention rate. They've been monitoring their mention rate for a few months and they have a really good idea of this is my baseline mention rate. And they can also see conversions and revenue as they're coming in. Hopefully, they can at some point make this correlation that when my mention rate increases by X, my conversion rate increases by Y. We've had multi-touch attribution platforms for 10 plus years which can't take an output from an LLM into their consideration. I wonder if they start to become worthless. What is the model we can create to try to project improvements in conversion with better mention rate? If we're going to try to talk executives into investing more in AI optimization, what are those data points we can use to say, do you see this, it turned into that?
Aimee Jurenka (23:57)
I love your idea because it's so new. We don't know yet. We're all at the same point as well as how do we go to our clients and bosses and say, "hey, all this work we've done for you in the past is useless." I was joking that we all needed to get together in a secret society, switch jobs and be like, "well, that guy," but it's me, so we could all collectively throw each other under the bus knowingly so then we can move forward. But really it's that shift of traditional search, you gotta throw that out of your brain. We're not going to be doing that. So getting that correlation of exactly how much money is this making us, the best I can think of, again somebody else tell me if you got a better idea, is those branded searches. Because we know you get a certain referral amount. People are going to start buying within the platform. So e-comm I think is going to get a lot easier if we can get data out of that. E-comm is going to get a lot easier to track if we can get data out of it. But that's the best I can think of for that connection. I really like your idea. We haven't had it long enough to know if we can project out because right now when I'm talking to people I can't make projections. I'm like, this is all brand new, here's what we're going to do to get started for a baseline.
It was purely an AI visibility approach to a strategy. And it was really, really cool because my strategy had all these pieces to it. You gotta do this, you gotta do this, you gotta make a content plan, you gotta add content, we gotta try to figure out some specialized schema. And this was at Rickety Roo. So it's a smaller client and they didn't have the availability for all that. They were like, "hey, the only thing we have is the site structure. We can do the site structure piece. You can organize it. Here's the one topic entity we want to be known better for." And I was like, okay, well, let's just run it. It was like my first official test. Celeste from RooLabs, director of RooLabs, was helping me do it and make it very official. I was like, this is awesome, let's get this set up. And so we did it. We ran it just to run it. And this was all their legacy content. So this is all that blog content, that informational content that I just wrote and said was no longer important. It was just that old content they had and it did improve visibility. Now we did not do revenue, we didn't do traffic, we didn't do things to see if it's going to increase revenue. The thing super exciting for me was that that told me, hey, all of us can now go, "hey, all that content we made for you that we now think the LLMs might want, we're going to organize that into your topic entities and it's still going to be valuable. We're going to be able to repurpose that content and still make it valuable for you." I think a lot of leadership especially is going to love to hear that we have a way to repurpose and utilize all that work we did in the past for the future. But no money's attached, no leads, nothing yet. Just that very basic core idea.
Joe (26:44)
Maybe we could talk about the crawl logs. Are you mining that information differently now?
Aimee Jurenka (26:49)
There you go, calling me out. I'll be honest with you, I have never reviewed log files. I just didn't need to for the strategies I was running for SEO. I could get most of the information I needed out of just a site health crawl. I need to do that now. So I am looking into how, and I haven't quite learned yet, looking into what are the steps and where the processes are to get those log files for all sorts of clients and all sorts of companies. When I create an idea, I really like the idea of creating it so the smallest solopreneur could do it on the smallest account and then also be scaled at the enterprise.
Of course for enterprise they all have Cloudflare or something. We can take them out, we can learn through Screaming Frog and then we can learn from there. Definitely 100% want to do that. There are like new error codes, or new to me error codes, that you only see in that, like timed out. That means the bot got, you're not fast enough for the bot, you're not fast enough for the agent. We want to be able to see that. I want to see your 404s because if we have made-up 404s and that's the model hallucinating, that means it's a content opportunity. It feels like somebody is asking that question, so that's a content gap we need to fill. We're going to start seeing which bots go where, maybe start seeing that early. Hey, we got this much bot traffic increase off of this is what we did. Now we are seeing, this week what I'm reading is saying that the crawling and the visibility really don't have much of a correlation yet. They're not seeing that, but at least we can say, "well, we're starting to get crawling, we're starting to see that." Maybe think of that as impressions kind of. And then the visibility might be our 1 through 10 as if you're popping up in the visibility. So see, again, I said throw all the old SEO out immediately and start organizing it in my brain that way. We're just there. It's so hard to change.
Joe (28:36)
I guess some have mentioned the importance of mining your server logs. And it's always felt really intimidating to me to export that. I'm not a database engineer, I'm not really going to know what to do with it. I guess I've never even really tried. Could I use AI? Could I build a tool? I could, Jon and I work together to build a tool that
Aimee Jurenka (28:56)
Go to Claude Code. You could. Yeah, I started vibe coding earlier. I was at a company and we were really, really ahead of the time where due to the P&L, we had to let our content team go. We had to let our entire content team off. And this is right after ChatGPT came out. So we were going to automate everything. Funny part, I was exactly 100% against it. I thought it was going to be a disaster. This is going to be a total nightmare. But I wanted to play with ChatGPT and I knew this was an opportunity for me to learn about custom GPTs and how it works. So I was like, okay, let's do it.
First month got them all set up. It was really, really cool. Content was ehh, but wasn't too bad. Second month, absolute freaking dumpster fire. We were having our old content manager, she was editing. So she was still working for us freelance. She was able to mark it all up. The owner saw that we now changed our process. So it was one of those live and learn type things. Experiment, fail fast, move on. I learned so much doing it and I actually turned into a fan of AI over that. Like what can we do then? Because I took it and was like, hey, with ChatGPT, I was able to figure out these content plans and I generated giant content plans with topics, optimized how we did our briefs. Freelance writers can always go and pick one every month they want for the next six months to a year. And so we were able to cut down quite a bit of overhead, being able to utilize it. I was like, this is freaking awesome.
And then I started seeing ones with like AirOps where they had their co-pilot. You just start typing stuff in. And that was really cool because if I wanted or had an idea for a tool, especially SaaS tools for AI search, they didn't exist yet. Just go in and start asking what they can do and how to do it. Start working it out in your brain. Lately I've moved over to Claude Code. So Noah Learner has got me into that a lot. And my God, you guys, if you're not on Claude Code yet, go do it. Go do it. It's a vibe coder's dream. So it's how I build my tools. I'm doing some reports on it. I redid my entire website with HTML and custom CSS. I've never devved in my life. I can read HTML, but that's about it. I don't know code or anything. It's amazing what it can do with the right prompt, being able to work through it and everything. Go for it.
Jon Clark (31:13)
We just built a tool so when you do a technical audit, image alt tags always come back as missing or as an opportunity. And usually what we say is, okay, client, we're going to put a stake in the sand today and any new images moving forward will put a process in place to make sure they always have an image alt tag, usually because there are hundreds or even thousands of images in the folder and it's just time-consuming to do. We just built a tool using Claude Code that basically will use, this is for WordPress sites, the WordPress REST API. It'll pull down every image that doesn't have an image alt tag. It'll use Claude to write it based on what it sees in the image. And then it'll give you the opportunity to approve it or not and push it all back up. Once they're approved, you can make edits in the interface. I mean, it's like insane. And this took maybe two and a half days to build. It's
Aimee Jurenka (32:04)
That is amazing. You could never do that before. It's a wonderful time to be alive.
Jon Clark (32:09)
Yeah, for sure. And I definitely want to talk more about your automation, but I'm going to make the worst connection ever. You mentioned you became a fan of AI. I wanted to ask about query fan-out as it relates to the topics and entities that we were talking about earlier. When you do that sort of analysis around your keyword research or even entity research, how are you thinking about breaking up those query fan-outs? I've heard some people say we're going to take all the query fan-out, build out an FAQ to answer each one on the same page and then push it back up. We don't think that's the right approach, but we'd love to hear how you're thinking about using that data because that's super interesting data as it relates to how you're approaching things around content and just entity alignment.
Aimee Jurenka (32:53)
So of course I thought like everybody else did. Okay, query fan-out, that's going to be our "people also ask." We're going to make more skyscraper content. And that Sitebulb thing says that's no longer going to work. But of course that's where my brain immediately went. This is how we get an excellent skyscraper piece. This is how we make sure we answer all the questions that AI has and fans out. This is how we're going to be able to make all that content that's behind the scenes, very much the "people also ask" tactic.
And then I started really thinking about it like, well, what does AI really need? Kind of even going back to that article, it's what does AI really want? When you're talking about Google's algorithm and search engines, when I first started, I really thought about what is their intention? What do they really want? What does the algorithm want? What is Google's number one intention? What are they really thinking about? What do they want us to write for them? And I started thinking about LLMs the same way. Start there. What do LLMs really want? What do they really need? What are they really looking for? And it's like, yes, query fan-out is there. It's a great place. I could look at it almost like keywords as well, but they really want personalized answers for personalized questions.
So if we're doing our query fan-out research and again we're seeing things that do not apply to our ICP, cannot apply to our personas, do not apply to our client base, throw it out. I'm really thinking about it from that perspective of we really need to know who we're selling to first, then we need to start doing this research, then we need to start to see what they're doing because LLMs don't want generalized content. They also don't need content about something that's not about you. Why would you waste your time and energy if it's not going to end up in a downstream conversion eventually? So that sort of idea for me is that's how I'm going about it. Now, once you have it, I'm not doing skyscraper. This might change next week too, it all changes so quick. I'm going to be doing individual pieces. I'm not going to be doing an FAQ at the bottom. I'm not going to be trying to answer everything in that block, especially with token availability. No, they're saying that the bots just read the first third and the last third. A lot of that stuff in the middle gets cut off if it's too long, it just gets truncated. So it just gets cut off and they don't read it anyway. AI wants that information, but also they want it as cheaply as possible. So the least amount of tokens they need to spend to get that information on the back end is what that actual system is actually optimized for. So if we could do smaller content that answers the questions that specifically target the people we want to sell to, I think that's where we're going to be able to see more of a lift than trying to do one long piece of content.
Joe (35:41)
The age of retrieval.
Aimee Jurenka (35:42)
Yeah, just scream it at the LLM. You know what, as a vibe coding gremlin goblin that I am, it does depend on the project. I have a really bad habit because I'm so excited about it and I enjoy doing it so much and I'm so like, "this is a great opportunity," that I get off and running. And then I kind of just gloss over a lot of those things. And then I end up like a lot of us where we send the wrong email or we have the wrong numbers or links that don't go to things and it's horrible. Quick, take that down from LinkedIn, quick, blah, blah, blah, because I get so excited doing it.
And so now there's a process, there are steps in place of like, all right, slow down. You're really excited about this part. You've got it all down. This looks great, but make sure you go take a break. You gotta take a break in between projects. You gotta maybe even take a whole day before I can come back and QA it because my eyes are seeing what my eyes want to see, what they're excited about. So that's a lot of my process for doing it because you gotta double check the numbers. You gotta actually read every line to see what it says. You can't just skim it. You gotta do a lot of things that I feel are hard for me to do when I'm caught up in the moment of, "this report's going to be so good" or "I made this whole presentation for this client, it's going to be so awesome," or "I optimized this and now my workflow is so much faster." That human-in-the-loop editing is definitely still required. Nothing great and fancy. I think I do what everybody else does. You get it, you go to another LLM, you ask them to proofread it, see what they come back from. ChatGPT will always gas you up, Perplexity is a little more hard on you. If you really want core answers, you go to Claude, but trying to jump around and see if whatever other LLMs say to try to help you sort of QA yourself.
It's so different. I didn't have a lot of local search experience when I came to Rickety Roo. So it's been really interesting for me to take a lot of these ideas that I have very SaaS, very B2B, that sort of thing where you've got this six-month roadmap, you can do this holistic plan and then work on clients that are like, "I needed phone calls in one month. What are you going to do for the turnaround? And we need conversions." They don't care about visibility. They don't care if they rank for something. They don't care if they're in local packs. That's not exciting to them. They're like, "are the phones ringing?" Very, very cut and dry. So at Rickety Roo, that's actually what we report on. Is that end metric for you? And then we'll back it up. If you want to talk about it more or if we are seeing early indicators that you care about, we will do that.
But going towards how is this going to work, how do I take my ideas and what applies to local has been really interesting. So working with Blake and Celeste on that, Melissa too, of course. How do we take these ideas and make it a local play? Because we know that this client is not going to have the budget for this topic entity whole caboodle plan that I've got going on. How are we going to be able to do that? And what are we going to be able to do right on their product pages? What are we going to do right on their service pages? How can we integrate this right now on their website for cheap, for easy? What sort of tactics can work? Right now we're doing a lot of comparison charts and tables. FAQs at the bottom of everything. And I know I said I wasn't going to do a bunch of FAQs, but true FAQs. Celeste came up with this, so I can't take any credit, but we're doing FAQs for AI answers. So we are noticing that local packs and all that sort of stuff have now got local in there. They're answering people's questions. So we are formatting an FAQ at the bottom of our pages specifically to get into that AI search answer. So that's kind of where I'm at. Unfortunately, I'm not really doing the maps. Talk to Celeste, I'm sure she's available. She's got some really cool things going on with that and how to mine sentiment and how to really look at that sort of stuff to be able to get in there. When it comes to reviews right now, we're just all concerned that they're all disappearing after this new guideline, and our valid real reviews, right?
Joe (39:37)
The local piece has been maybe for me the most challenging to think through how you help a client optimize for AI because the services are not really differentiated. One HVAC from the next, one plumber from the next, roofer to the next. You really gotta think hard about why your client's different than their competitive set. It's been really hard. At least with software, there's some feature that's different. And if it's not that different, you can make it different and talk about it differently, but it's been tricky.
Aimee Jurenka (40:07)
Yeah, and I remember that was one of the first things that occurred to me. What's going to have to change for informational content? I just got ripped to shreds too because I was like, personas, and people were like, "we've done personas forever." Well, you might have, you've had these large sites, you maybe work in enterprise, you have a brand team, you have a product team, you have a full content team. My guy doing HVAC in a small city that now wants to be ranked in Seattle doesn't have any of that. We don't have the money for that. We did it generally. We could just do it really generalized. Somebody who's looking for HVAC in Seattle, sells to these types of neighborhoods, maybe we'll just go with that sort of angle. Where now it's like we need to get specific. What neighborhoods do you go to? When were those houses made? Are they brand new and they need a maintenance program because they're first-time buyers and they don't know what to do yet and they're scared? Were they built in the '70s and we have DIYers that are going to be coming in and doing it themselves? Are we looking at a historic area where you need to come in and do a complete ductwork and a complete retrofit of it, plus do some poltergeist ceremonies to get rid of the ghost in there? Because every old house has one. I have an old house. Every project's like that. Every project is going to be more difficult. What do you know for your personas? I think again, it's going to be a lot more detailed, which is a lot more work that we're going to have to do, a lot more time and effort we're going to have to do for local SEO than we used to have to do.
Joe (41:29)
I like the idea of highlighting the neighborhood projects too. That's a good one to think more about. We only have a few minutes left. If you're okay with it, we'll move to a very short set of rapid-fire questions.
Aimee Jurenka (41:44)
Ready.
Joe (41:44)
I'm going to ask a question maybe back to tools because you're becoming a prolific toolmaker. Is there a tool you were using three years ago that you no longer need? Is there one subscription you've gotten rid of in the last couple of years?
Aimee Jurenka (41:57)
Well, in the beginning, there were a lot of AI tools for content briefs. There were a lot of content brief tools just in general. We're going to be able to make a content brief for you, we're going to be able to spin this up, we'll make the topics for you. All of that is scrapped. There's no reason at all for me to have any sort of content brief generation tool that you can't just create on an LLM or make in Claude Code yourself. So sorry content brief companies that did that.
I really think that was the first thing that we're going to just see go to the wayside. Because we really don't need that. And when it comes to my personal opinion, my crystal ball, to a lot of these really large SaaS tools that offer 25 different things, "we can do this and we can do that and our base subscription is this much money," I think we're going to see a lot of them struggle. Because a lot of people are like, "well, I only need two pieces of this. I only need this one piece. I only need to do this one thing for this one project. If it works out, I might spread it out to the rest of my clients, but I don't need any longer to buy this giant tool that offers 35 things for the one thing that I'm actually going to be using." I can just jump over and create that myself. So I think that's where we're really going to see the shift coming in.
Jon Clark (43:10)
Yeah, that's a good one. I'm curious, do you have any tips for underrated sources for prompt mining? Good old research.
Aimee Jurenka (43:17)
No. Here's my thing about prompt mining: it's so personalized. See, for me, if you're prompt mining, then it's the same as getting one keyword and trying to write a page for that one keyword to get traffic from that one piece of content and then work it through the funnel. I'm all about backing it up. So we're not going to be doing that one-to-one anymore. Prompt mining isn't really something in my strategy. It could be useful, we could look at it, but it isn't going to be necessary because what we're going to be doing is we're going to be backing it up to the bigger picture of what attributes, what topics, what things do we want to be known for, and what people do we want to reach. And so you don't need specific prompt mining for that. Again, keywords would work to help you get an idea. Prompt mining might be a good idea. But the idea of spending a lot of time and resources on that when you know that you're going to be writing for a personalized search which clients do we want to reach? kind of flips it around. Not what are people doing, but what do we want to do? Who do we want to be touch-pointing? Then it's not needed.
Joe (44:18)
Drown, we really need to work on our rapid-fire questions. I feel like all these questions require so much. Let me try this. I read your comparison article, you did a little comparison article, I guess you mentioned it, the misconceptions made with schema versus structured data. Is there a big misconception with schema you feel like still people are confused by?
Aimee Jurenka (44:27)
Oh, my Buzzword Betty series where I look at just the stuff that's kind of going around these days and really doing a little bit of a dive into what it actually is. I don't know about you, but I was using schema markup and semantic HTML and structured content and structured data, I kind of just used those all interchangeably. It was just one of those industry things where I also see a lot of other people doing it. I don't want to put anything on anybody else, but I know for me, I was just kind of plug-and-playing them around. But with AI search, not only are the buzzwords back, but also it's a different thing for them. The semantic HTML is the actual organization of it with the bullet points, with your H tags, with all that sort of stuff. Schema markup is the actual JSON in the back. And we don't know really which crawlers are doing what. We know Microsoft and Google have confirmed that their AIs use it, but that's about it.
Structured data, no longer do I want to use those interchangeably. No longer do I think that's okay for me in our industry, because we do want to start getting more specific about what exactly we are doing for AI. Not so much different, but exactly what are we doing? What does that mean? I'm a yapper. Leadership and clients are going to want to know more specifics. If it's not inbound marketing, if it's not keyword-to-page, and it's not all the stuff I already know, then what is it? The more specific information or tidbits that we can give leadership or our clients right now, I think that's going to make us the superheroes. That's gold. Any specific information we can give them for this make sure it's true about AI search and how it works, I think is really going to elevate us. And that was just starting my head spin off on that part of it when I was talking about not mixing up things anymore, getting really specific about our topics, what we're saying and what it means.
Jon Clark (46:40)
I think sometimes when the attribution and the revenue connection become a little less clear, that deeper explanation of the things that are happening behind the scenes becomes a lot more important.
Aimee Jurenka (46:52)
Thirst for knowledge.
Jon Clark (46:53)
Yeah, exactly. You produce Campfire Chat for Noah Learner, one of my favorite people. I think he's doing so much for the industry, really through the SEO community. What was the most interesting or maybe important thing that you learned?
Aimee Jurenka (47:09)
First off, I still know Noah Learner and I'm still learning from him. I absolutely love Noah so much. A big piece I learned from him is he was like, "I just share online about whatever I'm learning or whatever I'm interested in." Don't try and write thought leadership. Don't try to write educational. Don't try to write sales copy. If you want to start talking to people and build community, make some friends and do that sort of thing. Just write about what you're doing, whatever has got you excited that week, whatever your question is that week, whatever you're researching, whatever you're doing. Writing about what you're doing has been absolute gold for me. Nobody knew me two years ago except for me. So that's been an absolute gold piece of information that I take from him that I've just been running with. It's just fantastic.
Joe (48:03)
Well, you were rewarded with John Mueller's mention.
Aimee Jurenka (48:06)
He was the first person to be like, "Oh my God!" He was right on the SEO community Slack like, "Oh my God, Oh my God!" He's fantastic. If you haven't joined the SEO community, go ahead and do it. It's the nicest place in the SEO community. Instead of saying, "I don't know," we just say, "make me smarter." So then nobody has to put their ego in it. I really like that. Everybody just throws out ideas and it's a great place to, like when I was trying to learn what vector embedding was, they were right there for me to try to help me out. My brain got glitched on a part and it took me a couple of months to get past that glitch. We just kept talking about it and bringing it up where we're at and it's a really positive place to be.
Joe (48:48)
I laid in bed talking to Gemini for an hour about that one. Like, "no, what do you mean by this? What do you mean by that? Explain that like I'm a third grader."
Jon Clark (48:57)
That's a great prompt. Well Aimee, this has been amazing. Thanks so much for taking the time and digging in with us. Let our listeners know where they can find you online.
Aimee Jurenka (49:01)
Aimee Jurenka on LinkedIn and aimeejurenka.com. I am the only Aimee Jurenka. So if you spell it correctly, I'm the only one on all the internet. I didn't pick that branding by choice, it just worked out that way.
Jon Clark (49:19)
That is actually pretty helpful compared to Jon Clark where there are like thousands of us. Thanks again for joining us on the Page 2 Podcast. And if you enjoyed the show, please remember to subscribe, rate, and review. We'll see you next time. Bye-bye.