What if your content didn’t just rank — it resonated? In this episode, Melissa Popp reveals her exact AI-powered workflow for creating content that’s strategic, scalable, and unmistakably human.
https://page2pod.com - In this episode, Jon and Joe sit down with Melissa Popp, VP of Content Strategy & Innovation at RicketyRoo, for a deep dive into content that connects — not just ranks. With over 20 years in the game, Melissa has mastered the art of blending storytelling with strategy, using AI tools like ChatGPT, Claude, and Perplexity to enhance (not replace) human insight.
Melissa shares her exact workflow for integrating LLMs, the client onboarding questions that unlock authentic brand voices, and why internal linking deserves more love. She also explores the balance between creating at scale and crafting content that truly moves people. Whether you're an SEO pro, a content strategist, or navigating the noisy world of AI-powered marketing, this episode is a masterclass in doing content the right way.
🛠️ In This Episode
• 💡 Why content should be human and honest, not just optimized
• 🤖 How Melissa blends ChatGPT, Claude, and Perplexity into her content process
• 📋 The key questions that help brands find their voice
• 🔗 Using AI to uncover hidden gems for internal linking
• 🧪 The power of A/B testing and regular content updates
• 📉 Why Melissa avoids shortcuts like listicles for the sake of quality
• 🛠️ Custom GPTs, spaces, and workflows explained
• 💬 Why AI can't replace real empathy, nuance, and storytelling
• 💥 How SMBs can stand out even in “unsexy” industries
• 🎯 Redefining content success: from quantity to action-driven quality
Whether you're updating legacy blogs or starting from scratch, Melissa offers an approach that proves good content is still the best strategy — even in an AI world.
👉 Subscribe to stay ahead of the content game.
💬 What’s one way you're integrating AI into your content process? Drop it in the comments!
📚 Resources Mentioned
• Follow Melissa on LinkedIn → https://www.linkedin.com/in/melissapopp/
• Follow Melissa on X → https://x.com/poppupwriter
• Follow Melissa on Instagram → https://www.instagram.com/poppupwriter/
• Follow Melissa on Bluesky → https://bsky.app/profile/melissapopp.com
• Follow Melissa on Threads → https://www.threads.com/@poppupwriter
• Melissa's Personal Website → https://www.melissapopp.com/
• RicketyRoo → https://ricketyroo.com/
• Melissa's Human-Led AI-Powered Playbook Presentation at WTSFest Philly: https://ricketyroo.com/blog/wtsfest-philly-2025/
Jon Clark (00:00)
What happens when the person responsible for just getting content done becomes the one asking whether the work should exist at all? That's the question at the heart of this conversation with Melissa Popp, who has spent two decades turning messy, complicated ideas into stories people actually care about. Melissa is the VP of Content Strategy and Innovation at RicketyRoo, where she sits at the intersection of local SEO, content operations, and an increasingly noisy AI tooling landscape.
Her job is not just to crank out blog posts, it's to decide when content is the right answer, how it should sound, and how AI fits into that process without flattening a brand's voice into generic marketing speak. This episode dives into how Melissa is integrating large language models into the content creation process, not just to save time, but to actually make the work better. She walks us through her exact workflow across ChatGPT, Claude, and Perplexity, how she uses AI to uncover hidden gems for internal linking.
She walks through how she feeds those systems with brand questionnaires, customer reviews, and internal docs so the output feels like a real business with real people behind it, not a content farm. And she makes a strong case that whether you're writing for Google, for AI overviews, or for someone on their phone at 11 p.m., the only durable strategy is still quality content with a clear point of view. The tension for Melissa is pretty simple: Her clients need efficiency and scale,
but her entire career has been built on content that feels human, specific, and earned. She has to decide piece by piece where AI replaces grunt work and where a human has to step in to preserve nuance, empathy, and a brand voice that doesn't sound like everyone else in the SERP. I've been thinking a lot about the balance between scale and soul and content strategy, and Melissa had a lot to say that's going to stick with me. 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 (01:57)
Welcome back to episode 103 of the Page 2 Podcast. I'm your host, Jon Clark, and I'm joined as always by my partner in crime and Moving Traffic Media, Joe DeVita. I'm really looking forward to diving into all things content with our guest today, Melissa Popp. Welcome to the show.
Joe (02:07)
Hey.
Melissa Popp (02:14)
Thanks for having me. I'm always excited to talk about content.
Jon Clark (02:18)
Your background is so rich. More than, I think it's 20 years of content marketing. And Joe and I have so many questions about content. As always, I'm sure we won't get to all of them. But we asked our guests for a short bio before the episode. And I just want to read a really short snippet from what you added, because I think it's so succinctly perfect
that I can't imagine a better framing for how good you are at what you do. So this is Melissa's words. What I want listeners to know is this content isn't about chasing trends or pleasing algorithms. It's about telling the right story to the right people in a way that feels human and honest. That's the work I love and it shapes everything I do. I love that. I'm sure in your own voice, it would be even better. But what was it about content that grabbed you in the first place? How did you know
this was going to be your sort of thing?
Melissa Popp (03:04)
Yeah, I mean, this goes back to me being itty bitty content strategist, content writer, four or five years old before I knew where life would take me. I have always had an incredible imagination, a love for storytelling, a love for reading. By the time I hit high school, I had already read everything that was being assigned to me in English classes. It's been something I grew up with in my house and was fostered by my parents.
Deep-seated love of learning in general. And so I've been writing, I mean, since I've picked up a pencil back in the day. And so it started as creative writing, everything from poetry to short stories to plays, and eventually kind of led me into figuring out, I can make money with this. So let's see what I can do with it. And so that love of storytelling is really what drives me in everything I do.
Jon Clark (03:52)
I love You got started, or I guess you went to school for comparative literature, which I actually hadn't heard of before as like a major. And.
I was thinking about sort of the application of AI. Does that sort of background in sort of analysis really help you with the concept of maybe information gain or ways that LLMs synthesize content? Does that make sense?
Melissa Popp (04:17)
Yeah, and I think, you know, it's hard to answer that question because we're still learning so much about information game and how LLMs are processing information and what information are they processing model to model, right? So it's still very early in figuring out what works and what doesn't. But I think in general, my literary background, I was a comparative literature major and the journalism minor,
in the hopes of eventually go on to become a journalist. That was the big goal at the time of college. And I think really literary analysis gives me insight into what works, what doesn't, what actually drives a story forward, what pulls people in emotionally. And I think a lot of people in our industry when it comes to content marketing, we don't necessarily have time or get into that nuance with something we're writing for a client.
Ideally, we would all want the time and the budget to be able to really dive into what are competitors doing, what are informational sites doing, what do our clients have to offer that can dive into that nuance and that lived context of what content should be. And I think that background and analysis definitely, I look at content, I think a lot differently than my peers and the industry does. And I think it makes the output stronger and also the connections between
your clients and their end customer, I think it makes that stronger too.
Jon Clark (05:32)
It's interesting, right? As an industry, we often start with a keyword, and then we just throw content at that topic or at that keyword without a lot of thought around, well, what's going to pull someone through this full piece of content and hopefully even take an action, right? Like that's the really the trick behind doing content well. And I think you'd
you really talk a lot about that concept. One of the things I noticed about your title at you have this innovation layer to your specific job title. I was really curious about that. I guess, traditional content job title, you don't often see innovation sort of tied with content. And I think in the age of
AI, LLMs, etcetera. It's really needed, isn't it, to be able to think about content a little bit differently and how you're applying it to many platforms, not just search.
Melissa Popp (06:22)
Exactly. And that title change is relatively new, happened in the fall. And one of the things I've been spearheading at RicketyRoo for the last, it's almost two years now of working on these projects is figuring out how to integrate AI into not just our content workflows, but everything from the sales process to helping our SEO team, helping our testing departments. I've, gosh, two plus years, I can't believe how quickly it's flown by, you know, it's ChatGPT came out and everybody jumped on
Jon Clark (06:35)
Okay. Okay.
Melissa Popp (06:48)
trying to figure out how does it work? How do we use it? How do we show up in And I, from a very early standpoint of adoption for
AI, realized that if we as marketers in general, regardless of discipline or our niche, want to survive and continue in this industry, we have to learn how these tools work. We have to learn where to place AI in our efficiencies, what works, what doesn't,
what our kind of line is of how we use it. And so that's really where the innovation part has come from. I would say it's probably 70 % more on the AI driven side and 30 %
Jon Clark (07:21)
Hmm.
Melissa Popp (07:22)
on still the content marketing side. Because to me personally, like the content marketing fundamentals aren't changing because AI is in the picture. It's just another channel that we have to figure out what works, how do we help our clients get there, just like everything else that pops up, TikTok
blew up a couple of years ago and everybody jumped into is TikTok a search engine? Is it not a search engine? How does it work? And I feel the same way about AI. And that's really where the innovation part of my title comes in. It's supporting those efforts at our agency, figuring out what we feel good about using AI.
Jon Clark (07:39)
Right
I definitely want to come back to that sort of AI process and application. But I was curious, in your role at RicketyRoo, do you sit across all channels? Or are you specifically focused on the SEO side? Are you doing social content as well?
Melissa Popp (08:11)
We're primarily focused on local SEO and content marketing to support that. We do advise clients if they are doing social media, email marketing, those sorts of things, but really our bread and butter is the content marketing and local SEO.
Jon Clark (08:23)
You presented an awesome presentation around like human led sort of AI powered content and sort of the workflow that goes into that. I'm guessing maybe that's part of what you referenced a couple of seconds ago. But can you take us through that process a little bit? How did you like where did you start? Did you learn about the LLMs first and then figure out which LLM
mapped to the process best, or was it really just sort of trial and error and sort of learning the platforms as you go?
Melissa Popp (08:51)
Yeah, for me, I feel like it's the opposite of what a lot of people have done is I started with the platforms themselves. Like I said, I was an early adopter of ChatGPT, when really it was kind of the only one really publicly accessible at the time. Now, you know, we have hundreds of them that you can keep up whit. And so,
Jon Clark (08:57)
Mm-hmm.
All right.
Melissa Popp (09:08)
you know, I started playing around with it and not just on the marketing side of things. To me, the figuring out AI and marketing, that's the easy part. We do marketing every day. We know what we can get AI to do or not do. We just have to kind of figure out and push it and prod it and yell it in and those sorts of things. So I actually started learning it two different ways. One was through meal prepping and cooking at home. The old classic of this is all the crap I have in my fridge. What can I do with it? Things like that. And through trial and error.
Jon Clark (09:29)
Yeah.
Melissa Popp (09:35)
of that like
refining how I would prompt and how I would chain prompts and all sorts of stuff like that. And then the second I actually was inspired by Marie Haynes, who is using it to up her Fortnite game. Everybody who knows me knows I'm huge into Fortnite. And it's a way I can just blow off steam quickly after work. And Marie was using it as she was doing things and losing in matches was talking about that with
with AI and getting tips and tricks of how to overcome those things and then improve your game. And so I started doing that not just with Fortnite, but with other games as well. And so I think a lot of what I've learned about prompt engineering has been through trial and error, but more so doing it outside of marketing and seeing how far you can push it in things outside my comfort zone. And then being able to implement that into what I'm actually doing with creating outlines and building buyer personas and writing content itself.
Jon Clark (09:58)
Hmm.
Super interesting. Yeah, like real world application to what you want it to produce for business. Makes a lot of sense, sort of learning on the fly. So if you sort of look across the ecosystem, if I'm remembering correctly, the main LLMs you use for this content process are ChatGPT, Claude, Perplexity. And are you using Gemini? I can't remember.
Melissa Popp (10:44)
I am the least biggest fan of I do not understand them. I have a lot of friends who swear by it. I have other people I work with that use it. And I just do not see what they're seeing even with 3 plus coming out. Like, I don't see, you know, and that's the interesting thing I think about LLMs right now is like, you know, everybody is using them differently. You're not necessarily using the same prompts or custom instructions throughout
Jon Clark (10:48)
Yeah.
Thank
Melissa Popp (11:10)
and the outputs can be so radically different. But I primarily lean into ChatGPT, Claude and big huge fan of Perplexity.
Jon Clark (11:17)
Yeah, I've come to use Claude and Perplexity a bit more than ChatGPT, especially after some of the recent models came out. But take us through how you see each one of those sort of mapping to a content process. Like, how do you apply each one? And are you going back and testing them to make sure that it's still the best fit for that piece of the process?
Melissa Popp (11:36)
Well, let me talk about testing first. Yes, like I'm constantly testing and just because you know, I'm talking crap about Gemini doesn't mean I'm not using Gemini every day and testing prompts and testing things to see if things have changed, right? These LLMs are changing things on the fly every day, not just big model releases. So you never know how any of them are necessarily going to interact with you when you wake up in the day. So I'm constantly refining and trying to make sure that we're using the correct tools we might be doing.
Jon Clark (11:59)
you
Melissa Popp (12:04)
could
be a whole other four hour podcast. So long is very short. So perplexity is my researcher. was also an early adopter of Perplexity. I'm a huge fan of it. For anybody listening, if you haven't tried Perplexity, they typically offer free one year deals through a variety of things like PayPal. And I know Airtel did a deal with them. So there are opportunities to try it without spending any money. And if you have that opportunity
Jon Clark (12:06)
Ha
Melissa Popp (12:28)
please try it. I'm not an affiliate or anything for them. This is how much I love Perplexity. So it's my researcher. Perplexity is misunderstood as a tool. A lot of people think that it works just like ChatGPT or Claude. And really, Perplexity's engine is based solely on research. Now, yes, you can get it to do other things, but that's actually what it's built for. And so most people, when they first start using the tool, even though they do have access to other models within it,
don't realize that and then think it just sucks and moves on. But once you figure out how you can build it with research and leverage it that way, it's a deeper search in particular to me is the things it finds or the random questions where I'm like, here's a quote or a statistic, where is the actual source of this? And you search something like that on Google and you'll get every press release that's ever used it that has more topical authority than the source.
And Perplexity is great at actually finding the direct source for you and cutting through that noise. And so typically, I have a whole space set up for there in different niches that I can go in that have their own custom instructions, their own knowledge bases that they can go in and then use that research through the web to find things. So that's typically where I start. And then ChatGPT is both my outliner and my first drafter. So
Again, I build a lot of projects and custom GPTs for use cases, depending on client and depending on niche. It's not a one size fits all that I have a content outliner and that's what it does. It's much
Jon Clark (13:44)
you
Melissa Popp (13:57)
more nuanced into particularly with clients, what information can we share about them, branding guidelines, buyer personas, just knowledge bases in general. Sometimes you can be surprised by what clients have that they can share with you. And so I use ChatGPT as both my outliner and drafter.
I will say 5.2 sucks and I still will not use 5.2. I am still using 5.1. That can be a little bit frustrating with ChatGPT and how quickly they try to push out models to compete with others. You do kind of have to go back and forth sometimes, especially in the writing side to get the best output. And then I use Claude Opus 4.5.
Jon Clark (14:16)
Agreed.
Melissa Popp (14:31)
I do believe Claude Opus 4.5 is the better writer, but unfortunately the limits that they place on both their individual paid accounts and their business accounts still is just, for scale is terrible to try to build a process when you're, you know, might be writing 200, 300 pieces of content across a client, you know, multiple clients in a month. So I would love to move to Claude full-time as a writer, but unfortunately until those limits change, I just can't recommend that
for my team or others. What I've done with Claude is built both a fact checker and a copy editor. And so fact checker is based on, I built with custom instructions on a knowledge base, based on like the actual analytical skill of fact checking. And so it has this whole basic and pull on of how professional credentialed fact checkers would go in and look at things and pull things out. And so I do a first pass of that.
And then after that, I've built a copy editor that's right now, mine is currently based on AP style. that's typically what I lean into. But it also has an internal style guide that in my Women in Tech Fest presentation, if anybody checks out the blog post in the RicketyRoo blog, I actually give away that secret sauce and what my internal style guide and custom instructions for that is. And so that becomes my first pass copy editor and
Jon Clark (15:35)
You
Melissa Popp (15:48)
not only adhere to my internal style guide, but Also pull out a lot of those very common AI language points that we see. Our beloved em dashes and semicolons and
delve, let's delve into everything. And so I use this kind of workflow for any piece of content that comes across my way.
Jon Clark (16:11)
And all these are set up in custom projects and spaces, right? So you sort of take one output and move it to the next project or space, and that's sort of the workflow.
Melissa Popp (16:21)
Yeah, it seems much more complicated when I'm speaking how the workflow works, but once you actually get into it, it makes sense intuitively from going from one tool to the I try to, everything I build for myself and for other people, there are people in this industry that understand the technical aspects of LLMs and all of that. That is not me. So everything I want to build is very much the keep it simple, stupid kind of method of
of just simplifying and making the access as easy as possible. And then that's where you get a foundational output that let's say is 60 to 70 % where you want it. A human can now come in, save three, four, five hours of work and really focus on the creativity, the empathy, the human connection that AI doesn't necessarily always get.
Joe (17:07)
I've read, you've written a few times about how you've helped small and medium sized businesses find their voice. I've read where you, know, just because this brand has a logo and a color scheme and a font, they don't really have a voice in the market that makes them different from their competitors.
And it sounds like you help those companies find their voice as you take on a project with a new client. What's the minimum you need to know about a brand to give ChatGPT as a foundation for writing in that brand's voice?
Melissa Popp (17:41)
That is a really good question. And I want to say that I don't know the answer exactly to that question. Part of the reason for that is like, is that, you know, when you're working with with small medium sized businesses, most of the time you're working on it, you're usually working almost
almost always with an owner, with people who are actually running the business day to day or doing the work of the business, right? And so they know they can do the work, but for them, you know, it's very much putting food on the table, sending their kids to college. They're not thinking about the fact that they were able to start a business. It's been successful on a level, which means what they're doing works. They're connecting with people. They don't necessarily think about that connection. And so...
When I work with clients, like I said, I wouldn't necessarily, there's a minimum amount of knowledge or documents or information you can share with ChatGPT or other LLMs to speak in that voice. But the more you get to talk to not just the person who owns the company, but the people who are on the front lines doing the work, the customer service, people who are answering the phone, even a receptionist, you can start learning why are they there? Beyond just this idea of a paycheck, right?
Why did they decide to take this job versus? So even if a client doesn't necessarily know who they are, by talking to them and getting to know them, you can start pulling out those storytelling elements. Some clients, their brains do not work that way and you can't necessarily pull out. But there's other ways you can compensate for that. And it's things like looking at customer testimonials. What are customers saying of, especially in those positive reviews, right?
What are they saying that drew them to there? You can pull data from that. Talking to anybody who answers the phone at a business, what are the questions? What are the problems people are coming from? And then figuring out, you know, we work with, at RicketyRoo, we work with a lot of home services businesses. And let's be honest, you know, HVAC, plumbing, all of that stuff, there's nothing magical or special about those industries. You know, these are, you know, fixing a toilet is probably about the same as it was 20 years ago as it is today, right?
But there's a reason why people go with one plumber over another. And sometimes if the client doesn't know that answer, we can usually just having conversations like this. It doesn't have to be a questionnaire. It doesn't have to be some, let's figure out, look at your GA4 and figure out your demographics and then pare down from there and figure out what makes you special. It's having conversations like this. And this isn't something that we normally do in an SEO driven industry.
Right? You know, it's, we're not having conversations. It's very much who's your ideal customer and what are their pain points? And it's like, yeah, everybody has that. That's there's nothing special about that, but it's, finding that nuance in those conversations like this, that you can pull out that info. And then that is things that you can continuously be feeding as your seed content and ChatGPT or elsewhere.
Jon Clark (20:31)
So, you know, in a lot of presentations, not just really just industry wide sort of generally speaking, a lot of people are like, yeah, set up spaces or set up custom GPTs. But there isn't a lot of, I guess, insight into what is the training documentation or what are those custom instructions that really need to be considered in order to get a great output. You listed a whole bunch of really
probably ideal pieces, right? Testimonials and just transcripts and things like that. Is there something that you always try to include in your custom instructions when you're working with a new client or trying to get a better output? Is there anything that, you know, is just, it's like the first thing you start with?
Melissa Popp (21:12)
Yeah, so when it comes to custom instructions, I'm going to touch on that first and then talk about our seed content. So when it comes to custom instructions, I would say 80 % of the time, you'll have a base set of custom instructions for a client. And maybe 20 % of that will be nuanced to who they are, what they do, and little tidbits there. Your custom instructions are going to guide more what your space or GPT does versus necessarily inform it on what your knowledge base will be.
So you have that aspect of it. On the knowledge base side, it always comes down to data analytics has always had the concept of garbage in garbage out. So the more garbage you feed in AI, of course it's going to spit out garbage to you because you're not giving it anything to work with. You know, people want to pretend AI is on the verge of sentience and figuring those things out. But AI is number one goal is to answer the question based on the answer you want, not the answer you need.
And so the more garbage you feed into it, the further away you're gonna get from the answer you actually need. So my biggest thing is any collateral we can get from a client, we wanna add to this, whether it's branding guides, buyer personas, internal knowledge bases, anything they feel comfortable that's more publicly facing, that like if because of course we can't control what AI does with the content we give it. So we wanna be mindful we're not feeding it proprietary data or anything like that.
Jon Clark (22:30)
Right.
Melissa Popp (22:30)
So from there at RicketyRoo, what we do is we have a branding and content questionnaire. It probably makes most of our clients want to throw themselves out a window because I'm a very thorough person. And so I believe right now it might be close to 30, 40 questions in this. And a lot of these questions are not things that I think a lot of our clients are used to being asked. And some clients stumble over it because they're things they're not used to thinking about. Others start
really getting into it and give us great details about things. These questions are everything from the simplicity of who in your industry as a competitor has better than you? Because you can have your competitors, but then that doesn't necessarily mean content-wise. They correspond to who you are, how you want to do things. Than you? And we get into more nuanced questions of like,
What words and phrases do you hate? Do you not want associated with your brand? What do you want associated with your brand? Things like this. And so once we get that, that's a knowledge base that we can then upload that we can use and feed into ChatGPT. And anytime our clients, you have clients who change services, change taglines, all sorts of stuff. As soon as we get that, that's a great thing about AI is we can go right in and update that live.
It's not we have to wait a week for it to process or anything like that. We can get that information. So really, there's no checklist of how much we want or need, but it really is whatever we can get from our clients that might help, we can then add. And ChatGPT, Claude, all of them have very robust knowledge systems that you can upload a lot of stuff to. And especially if you...
I'm a big fan of markdown files for everything I upload into it because they reduce space and they're easier to process on the LLM side. So really like the sky is the limit of the more you can get from a client. And it can even be, you know, what are your 10 favorite blog posts across the web, not even in their industry. Take all the content from that, use that as a guide, you know. You can really get outside the box of what could potentially help create better outputs.
whether you have official documentation or not.
Joe (24:35)
We have a similar process with a new client. It's like an intake form. We don't ask them to fill it out. We schedule a call and like we type out the answers or record them as they speak them. But the hardest part always seems to be why are you different than your competitors? What value do you bring to the market that they're not bringing? And the answers we often get are like some little thing about their product that makes them a little bit different. But nobody uses like adjectives
Jon Clark (24:36)
I'm.
Joe (25:03)
is to describe themselves. Help me get to, how do I create your brand voice? It's not with that little product feature that's a little bit different. It's something about your company or your leadership or the drive of your employees that's gonna help me understand your company's voice. We only work with marketers. They have a tough time understanding that. Do you see that? Do you see that often?
Melissa Popp (25:26)
Oh yeah, I mean, especially with smaller businesses for sure. Like again, this is a mindset, they start a business, they realize they can make money, they realize they're good at what they do, but it's almost, sometimes I wonder if it's a psychological issue versus a marketing issue, that these business owners and the people that work for them
don't actually have the confidence in what they're doing compared to their competition. And so you ask them that question and like, I mean, I would say 80 % of the people I work with, I could ask that question too. And they're just like, well, we're good at what we do. We have good customer service. We have friendly people. And it's like, no, those are good things. There's nothing wrong with that. But everybody has that and everybody markets themselves that way. So what else?
And then most of the time they sit there and they scratch their head because this isn't something they thought about. They just were like, OK, we need to make money. We're going to go do this. We're making money. That's good enough. And so this kind of circles back to what I was talking about, where it's as you get to know your clients more, you have more of these conversations. You can start figuring out first who they are as human beings, right? Because just because they're paying you and they're your customer as an agency.
you know, sometimes there's a power imbalance there, right? They're paying you so you feel you need to do, you know, what they say, right? And you have to kind of push them out of their comfort zone of that relationship to get to what makes them better. And frankly, sometimes, honestly, there might not be something that makes them better. That's the reality, right? Especially in more saturated niches and industries. And so that's where it's like, okay, like, what is
the best customer service you ever gave to a client that was above and beyond that cost you money? It's a question I like to ask. You it's like, you you didn't make anything off of that. It probably cost you money to help that. But what made you decide we're going to eat the cost of this, do the right thing and help this person regardless of what industry or service it might be?
And so that's a question sometimes I will ask when I get into those places, because I firmly believe that every individual on the planet, every business, you have a story. There's a reason why you're in this world and why you're here doing something, whether it's work, whether it's play, whether it's relationship, whatever it might be. But we don't look at our lives as stories to tell.
You know, we're going through emotions where we have things to do off a checklist. You know, we have kids to take care of and school to go to and work to go to. And we don't think about the story that we're telling in our own lives. And then for businesses, it's the same thing. They're not thinking about the story that they're telling for their business because they are telling the story. They just don't know how to look at it that way. And that's where we can come in and help them figure that out and help them build that.
Jon Clark (28:08)
I love that you're really almost giving a personality to, you know, in some cases, businesses that traditionally haven't had one, right? Like HVAC is, you know, something every house needs, but there isn't like a, you know, a sexy story sort of behind the implementation of It sounds like through your questionnaire, you have a very unique opportunity to help
these companies sort of define, well, maybe two parts of the question. Are you dealing at all with like this concept of like semantic triples and sort of helping brands determine what those are for themselves? It sounds like that questionnaire would be ideal to sort of
lead that brand to what they want to be known for in LLMs and otherwise? Is that something that you guys are focused on at all?
Melissa Popp (28:53)
So that's something that we're A-B testing to figure out impact, both with our clients and honestly with our own brand as well. You know, there, and to kind of like jump on that, like there are a lot of things about how people can show up in LLMs that a lot of our industry either swears by, swears off. We can't even decide what we're calling this whole era of evolution. Yeah, yeah, you know, I mean, that's a whole other episode of just the nomenclature of the world we're in.
Jon Clark (29:14)
You must spend time on LinkedIn.
Melissa Popp (29:21)
You know, and so, you know, we are testing different things with clients, both with their help and without, both internally to try to figure out, you know, whether it's semantic triplets, whether it's query fan out, all of these things, right? Cause you know, we don't know as an industry yet what works and what doesn't. We have ideas and we're testing. And so, but yeah, I mean, the concepts of semantic triplets is really fascinating because that has been around much longer than we've talked about it with AI. And we've...
I think the way we write for social media in particular, not so much long form content or video, but social media like Instagram, Facebook, Twitter, those sorts of things, we have changed how we wrote to be so much more simplistic than our old school, how we dissected sentences in elementary school and all of this stuff. And so it's fascinating everyone talking about the concept of semantic triplets because that's always been a writing tool.
But just now people are like, Oh, this makes sense here. We need to jump on that. And the reality is, most of these businesses that we work with, couldn't in one sentence using semantic triplets identify exactly who they are, what they do, boom, in that sequence. And so I think it can be a really good exercise whether you're using it to help write and show up in LLMs as a tool.
but also to have someone focus and narrow down. It's like your 30 second elevator pitch, right? How many businesses can give you a 30 second elevator pitch? How many of us as individuals can do that? And so I think semantic triplets can be a powerful way, not just to test and see what might work within the context of content, but also an exercise for clients of figuring out exactly who they are and what they do.
Jon Clark (30:50)
All right.
Yeah, it's my daughter seven and she's, you know, deep into learning how to read and she'll bring home these, you know, very simple books. It's like the hat is black and, you know, the cat is... like very simple, basically semantic triples. And that's like the foundation of writing a good sentence. And then, of course, as you get older, you add in adjectives and adverbs,
descriptors and all these different things that make it enjoyable to read as an adult. We had Duane Forrester on and he was talking about one of the unique, I guess, components of a good SEO writer or guess writer in the future is to be able to balance those two things where you're sort of need to write
so much simplistically for LLMs today, but you also need to be applicable to how humans like to read. It sounds like there's a balance there to sort of toe where you're trying to meet the needs of both things and the user.
Melissa Popp (31:55)
Definitely. I mean, what's interesting about that is that's how it's always been in our industry, right? Like this is not a new problem. You know, when SEO first started, gosh, whenever I say I've been doing this for 20 years, I mean, it just dates me and ages me so much. But you know, when I started, there was no Google. For those of you out there, believe me, when I say this, we had things like AltaVista and Lycos and Ask Jeeves in its early iterations. You know, as Google emerged and... ⁓
Jon Clark (32:15)
Right. ⁓
Melissa Popp (32:22)
the ad revenue behind Google in particular, as people started jumping on board, SEO was very simplistic. At one point, if you had 51 keywords stuffed into a piece of content and your competitor had 50, you were number one. And so since the beginning of the idea of SEO, we have always had that tipping point of how do we write for search engines and algorithms? How do we write for humans? And it's dipped back and forth. And we've had Panda and Penguin and...
and all sorts of stuff, helpful content update, and Google plane catch up to try to eliminate how SEOs are manipulating search results with these tactics. So this is always been, you this is just the next iteration of that with AI, right? And the biggest problem is, is it keeps coming back to the same thing. Quality content is what drives showing up anywhere, whether it's organic search, whether it's Google, whether it's Bing, now whether it's AI.
Jon Clark (32:59)
Mm-hmm.
Melissa Popp (33:19)
You still have to have a foundation of good content that's written with intent. The problem with AI at this point though, people have learned that Google is an evil entity and you can't trust them, right? More people are educated and understand that what you see on the first page may not necessarily be the highest quality content. You hope it does. Google obviously wants it to as well, but you know, it's an imperfect system.
And now we have people switching to AI where they ask a question and they automatically believe that what the AI is spitting out and what it's sourcing is correct. But it can only give you, just like search results, the best it can find, right? And since we're in these early stage of LLMs, it drives me nuts seeing people out there saying, okay, well, we're gonna optimize for AI. It's the same for every AI out there. Is it?
Jon Clark (34:09)
Hmm.
Melissa Popp (34:10)
Please somebody out there come yell at me and tell me that I'm wrong there. We have no idea how Claude surfaces stuff compared to ChatGPT. Gemini We have I think a better idea. think it would possibly works a little bit closer to how Google search works in some ways on the backend, but we really don't know, are we dealing with different AI ranking factors per engine or not? And so, you know,
Jon Clark (34:11)
Right.
Melissa Popp (34:35)
We talk a lot about citations and internal linking and what it takes to surface
Jon Clark (34:39)
you
Melissa Popp (34:40)
and be shown in AI. But you still need quality content for them to find crawl surface to show up there. So it's funny how many circular cycles we see of this. To me, quality content is always the first thing you start with. It doesn't matter what the medium is or what your goal is. That has to be the foundation of how you start.
Jon Clark (34:57)
Okay.
Melissa Popp (35:00)
And everything else lifts it up and helps it, whether it's ranking in Google or it's ranking in an LLM.
Jon Clark (35:03)
Okay.
Joe (35:07)
Do you have a good like bullshit
Jon Clark (35:09)
you send it back to the
Joe (35:09)
meter for the slop people are producing with AI for your own team? Help us develop a bullshit meter for Is is there a few things that like you send it back to the editor and say, it's gotta be better'?
Melissa Popp (35:11)
yeah.
Yeah.
So I put a lot more trust to my team. I don't want to say that I probably should. We have a great team at RicketyRoo. It's very hard with AI and how quickly things are going to the slop is getting harder to see, I believe, especially for those who aren't using AI daily. Obviously marketers are using it daily. They're, you know,
If I see more than three em dashes in your content, even if I know you like em dashes, I know you used AI. There is no reason for there to be that many em dashes for the most part, if you want an em dash to be effective. The same thing with the use of colons instead of full sentences that yes, again, that's another literary way to break up content, but at the same time, it's overused by LLMs. There are certain structural ways that an LLM speaks, which makes sense.
It's studying so much writing, it's basically saying, okay, you know, over a hundred pieces of writing, I see em dashes everywhere, I'm gonna use em dashes, that must be correct. You know, that's my theory of how these patterns emerge. You know, in the very beginning, when ChatGPT first came out, it was the word delve and ever evolving. And you'd see, I don't know, I couldn't tell you the last time I used the word delve until I started using AI. And then I was like, oh yeah, this word exists. Nobody I know uses this word.
And so there's patterns like that that you can see that have. you know, I mean, just last year it was emojis everywhere. Yes, people use emojis, they don't use 52 in a piece, a LinkedIn post, you know, that kind of thing. And so, you know, really the more content you're reading online, whether you're a writer or not, the more I think you can spot those patterns. But if you're not reading as much as you normally do,
Jon Clark (36:44)
Yeah.
Melissa Popp (37:02)
and you're not actually seeing those patterns, then it becomes much harder to spot that swap. And unfortunately, we're at to a place where you can almost create infinite content at this point you know, it's only going to get worse. Just like keyword spinning in the beginning of Google, where people would throw in a software, create a thousand programmatic type pages, put them out there. And people got, unfortunately, the people who aren't in our industry and don't understand how this works, they get used to that.
And that becomes the new normal. And then they don't see those patterns anymore. And it becomes an education piece. Because unfortunately, it just seems most of the people using AI and the internet in general, they want quality content, but they don't really care if they get it or not at some points. And I think the AI slop is that new kind of place where if it kind of answers the question, who cares how well it writes.
Jon Clark (37:50)
It's interesting, like every client wants more content. And I think the beginning phases are always, yes, it needs to be quality, it needs to align to my tone of voice. But the ease of creating I think sometimes outweighs that, right? Because it just becomes so easy to just put out more and more content. And so much of legacy SEO was:
'You need more content. You need more content'. And so the concept of producing more to generate more was the mantra for a long time. Unfortunately, that's not the case anymore. Really quality wins, even if the volume might be less. You definitely have a great process for AI into content and still producing
high quality content. Is there anything that you found AI to be really bad at in that process? Like is there something that you just don't hand off to an AI and you like own that piece of it from a human perspective?
Melissa Popp (38:44)
Yeah, so I would say that when it comes to niches and industries, even your money, your life and more complicated topics, we will always try the process with AI to see what we can get out of it. You know, there, I have worked in health niches, very complicated nutrition based health niches that I've been incredibly surprised
by how well AI has actually researched and we'll get 80 % of the facts right when I actually have a doctor or a nurse look at it and vet that, right? So we always start by any industry we wanna try and see what happens because I have been in the last probably year, I've been very surprised by how much better
all the LLMs, even Gemini, have gotten at synthesizing information for that output. That's definitely improved. So from there, what we'll do is we'll take a look, especially if it's a more complicated industry, what does the output look like? And more importantly, how is it distilling more maybe legal, health, technical information? Is it able to actually synthesize that in a way that
Jon Clark (39:30)
Thank
Melissa Popp (39:51)
when you get your foundational output, it makes sense. It flows right. It's actually written well because that's usually where the disconnect comes from in more technical topics is that it seems like it spends its time more trying to get the facts right than trying to actually write it correctly in a way that you can run with and edit and all of that. So sometimes it depends on the industry. Sometimes it doesn't.
Legal definitely seems to have, legal and accounting actually seem to have some of the most problems I've seen. And mostly that's because the laws around them change so frequently that you may not be necessarily be getting the most up-to-date information. And that's another problem with some of these, you you're trying to write news content and stuff like that, you know, depending on what the knowledge cutoff is and whether or not you know how to prompt to get live data
Jon Clark (40:41)
Mm.
Melissa Popp (40:43)
which a lot of people don't realize you can do or how to do that well, you may not be getting accurate information like up to date as well. but yeah, no, I definitely like, we try every niche that we can to see what we can kind of milk out of to get a foundational output that our writers can then run with.
Joe (41:01)
Our agency doesn't, we don't have any, we don't have writers on staff. We don't do a lot of writing. We do some, we work with lot of freelancers to develop content when we need to do it. I'm curious if you see a change in pricing now. You guys create a lot of... you write a lot of content knowing, we have some clients just say, yeah, I want two blogs, just go to ChatGPT and do it. They don't even want,
the quality, like they're coming to us say, I don't care about quality, just get me that quick output. We don't want to do that. We're not going to do that. And I assume that you don't want to, and you're not going to do it either. But when it's so cheap just to develop content with an LLM, can you ask for more if you guarantee that it's all human done? Is the pricing at all changing for you guys and how you create and deliver content?
Melissa Popp (41:49)
No, if anything, our pricing has actually gotten a tad bit more expensive since we've heard. Yeah, yeah. And there's a couple of reasons for that. One of them is, you know, I don't like using the term proprietary for anything, right? But it kind of makes sense a little bit of this. You what we're building is not what anybody else is building out there. Of course, we're using other, you know, this isn't 100 % proprietary. But gosh, I couldn't even tell you how many hours of experience for me
Joe (41:53)
That's what I was hoping to hear, yeah.
Melissa Popp (42:15)
it took to be able to build what I've built for RicketyRoo. or what I've built for freelance clients, because I also assist companies building their own AI workflows as well outside of what I do at RicketyRoo. And so, you know, there's, you know, this is just like, why does a doctor charge insurance so much when you have open heart surgery? It's because they can do open heart surgery and you can't. And so that, you know, there are things that I can build and assist clients with, not just with RicketyRoo's processes, but in general, like,
We'll have clients who come to us. We just had a client come to us who, gosh, I forget, I wanna say it was a thousand plus location pages that they wanted to build. It was the same mindset. Let's use ChatGPT. Let's just, it doesn't matter. They don't need to be unique. Trying to give me a heart attack. So I need that open heart surgery. And so we told them, we're like, no, we are not going to do this for you. Here's the reasons why. Like here, yes.
Jon Clark (42:51)
Wow.
Melissa Popp (43:06)
Short term, it probably will work. You'll rank in those thousand cities and it'll be fine. And six months from now, Google is gonna hate you, LLMs are gonna hate you, and you have to start all over, right? Very common quality versus quantity problem that's been around since the beginning of our industry. So instead what we did is we built them a guidebook of here are all the best practices. If you wanna use AI to do this, here's the pros and cons, but here is the guide to doing this correctly
and kind of found the middle ground for them, that it wouldn't be pure crap that's getting put out, but it wouldn't necessarily be the perfect ideal of what we would want. But it was the middle ground. So we will work with clients in that way as well. We were never going to say yes to 1,000 ChatGPT, five minutes per post, boom, boom, boom here. There's no way. What we stand for as a company,
who we are as individuals and what we believe in with the output of what we're doing, the work we're doing. But it is a very unfortunately common problem in this industry because, you know, let's say that you're charging, I'm gonna use really conservative numbers here. Let's say you just, for all those location pages, you charge $500 for one of them, right? Even to do that internally for that cheap, not AI, not anything, right? Most companies can't afford that, or they're gonna see that number and think you're insane.
But the reality is this particular business has not many locations and is not showing up. It needs to figure out a way to do it. So how do we hand off something that makes us feel good, that we don't feel sleazy, right? And also doesn't set them up for failure down the line where they're gonna have to spend that $500 per post anyway to fix that because it's not sustainable. And so, you know, there's, our industry has always had
the good guys, the bad guys, the middle, right? Everybody has their line of ⁓ black hat, white hat, gray hat tactics they wanna use, whether it's content or SEO. And I think now more than ever, we're in a place where we are operating a little bit more in the gray, but the reality is our experience can push us closer to what we feel good in that white space, but it is still gonna possibly cost more to do that.
Despite the use of AI because you have a human that is involved in a much more nuanced and oftentimes intense way than they were if they just picked up a pen and started writing a blog post, right? So yeah, we have not reduced our prices at all since integrating AI. And in a lot of cases, I think we're producing better content
Jon Clark (45:30)
A critical way, right?
Melissa Popp (45:42)
with the time constraints and budget constraints that we have with clients than we ever have.
Jon Clark (45:46)
Yes, I think one of the worst things that Google did was come out and say that AI content was OK. I feel like there are so many asterisks that were beneath that statement and everyone just latched on to the overarching sentence. Anywho, so before we jump to rapid fire, I did have one other question because I was really curious about internal linking. It's so critical for websites in general, but especially content, especially when you're creating
you know, lots of blog content or you're building out a hub page, right? Are you using AI or who knows, maybe even like a project or a space to assist with internal linking as part of your content process?
Melissa Popp (46:22)
So I have built a space in Perplexity in particular that I use. It's not client-based or it's very much its own internal linking expert space that I can then run deep research for to kind of find like hidden gems, right? Because like it's super easy with internal linking to look at your, however you're gonna break down your anchor text, right? And find the very obvious ways to interlink not just to the page you're trying to rank, but elsewhere, right?
You got to remember internal linking is twofold. One is user experience, which we often forget, and the other is for boosting SEO, right? And so oftentimes we're only looking at it through that lens. I like to look at it through both lenses. And so I've built a space to help me find hidden pieces of content, especially on larger websites, that I can link to ⁓ and that both make sense, may not necessarily have exact anchor text or things like that, but make sense of where you might want to lead someone
on a website.
Jon Clark (47:16)
Got it. Yeah. I have not, ironically, I've not tried Perplexity for internal linking suggestions. So I'm definitely going to do that after we get off the pod here. All right. Let's jump to rapid fire and wrap things up. What is one Star Wars character that would be the best content strategist?
Melissa Popp (47:34)
Without a doubt after Andor, it's Mon Mothma. Her speech, probably one of the greatest political speeches, even though fictional, ever. She understands her audience, both the good and the bad guys. She understands timing. She understands the stakes of when to speak and when not to speak, absolutely. I should build a custom GPT of her character and see what I can get out of it.
Jon Clark (47:56)
Hey, you should. That's awesome. What's the content format that's overrated heading into 2026?
Melissa Popp (48:02)
Oh gosh, everybody's going to get mad at me in the AI space. It's listicles. Come on, how many listicles can we still make? Yes, I get it. They show up in featured snippets and AI overviews and LLMs, but come on, how many more can we do?
Jon Clark (48:06)
Ha
Yeah, I would second that one. ⁓
Joe (48:16)
Or at least stick
to the top three, top five. It's like the 17, the 23, it's like, ugh.
Melissa Popp (48:19)
Yeah!
Yeah,
don't got time for Yep, exactly.
Jon Clark (48:23)
And then their company's listed number one.
How about underrated? What's an underrated format?
Melissa Popp (48:29)
Comparison pieces. We neglect the middle of the funnel for everything we do. And comparison pieces are probably the most powerful way to convert someone into the bottom of your funnel. We just, we ignore them because they don't typically have great SEO value.
Jon Clark (48:33)
Hmm.
Yeah, that's true. If you're unsure that a piece is useful, do you have... like what's your litmus test for that?
Melissa Popp (48:50)
I do a lot of A-B testing with content in a lot of different ways, sometimes big, sometimes small. I will often want to do title tag test, headline test, header tests, even shaking up where content goes in a piece. I have been known to take images out, add images, like sometimes I'll have to go in and be like, what is happening here? Because I'm just rapid fire of that. So I'm a big believer in continuing to test what works and what might not work.
And I'm a big believer too in content updating. That's a huge part of my process. I don't just believe in creating new content like a lot of people do and A-B testing those pieces you're not quite sure of once they're live. I think is a way to really figure that out and then adjust if needed rather than get rid of.
Jon Clark (49:35)
Content updates, we love that as well. It's a big part of
our sort of content audit strategy for a site as a whole. Do you have like a, this is not maybe rapid fire, but do you have like a general timeframe in which you always go back and just take a second look or is it really dependent on the topic?
Melissa Popp (49:55)
I think it's twofold. I think it's really dependent on the topic and whether or not that particular topic or niche is kind of evergreen, you know, kind of going back to plumbing and HVAC, right? Not much changes in those industries. That information is a lot of it's static. There's not a lot of innovation or anything there. But, you you get into something more medical or health focused and, you know, you might daily be having different updates to things as the government or healthcare providers, insurance companies, medical breakthroughs happen.
Jon Clark (50:06)
Mm-hmm.
Melissa Popp (50:23)
Right? So ideally, you know, if I'm in a, in a fast moving niche, I like to look at content every six months, do content audit, figure out what's working, what's not, what can we adjust? I am, I'm known to take a scalpel to content audits and make clients, clients cry. not, not to be mean, but, the reality is, to, to move the content that's going to work, get rid of what doesn't and, and keep pushing things forward.
And so, you I think it depends on that. And then it also depends on the amount of content a website has. You if you have, you you think of something, Amazon may not be known for informational, but billions of listings, right? You know, the reality is you're not going to be able to tap into something that big all the time, right? So that might be something that's more every year, every couple of years. But really for me, is much more topic focused of how fast that topic moves.
Jon Clark (51:11)
Got it.
Gosh, I wish we had more time because I feel like I have so many questions around your approach there. We'll have you on again. How about that? Who is someone in either the content or maybe just the SEO space in general that you're really paying attention to right now, especially as we're in this sort of inflection place as an industry?
Melissa Popp (51:18)
Yeah, I love you.
Yeah, I mean, it always goes back to Brittany Mueller. She, the education that she's putting out there to teach people freely in many cases of how LLMs work, how machine learning works, and then the constant encouragement and push to build and to fail and to try things. She has been a great resource and friend to me throughout the last couple years of this. And it's, we need more people in our industry to build and to play
Jon Clark (51:39)
Mm-hmm.
Melissa Popp (51:57)
and to fail and share those things. It's not just about results and, this worked. Well, what worked about it? You don't have to give away all your secrets to do that, but the reality is the more that we can put ourselves out there and work together as an industry as opposed to against one another as everything is changing, the better off we're all gonna be. And I think Brittany just really encourages people to have that type of attitude in this kind of every day it's changing kind of chaos.
Jon Clark (51:59)
Great.
Yeah, I love Brittany, if you're listening, there's a DM in your LinkedIn inbox. We'd love to have you that's a great one. She puts out such great stuff. And you're right, willingness to teach and educate is phenomenal. All right, last one. If to bet
RicketyRoo on like one thing as it relates to content strategy for the new year, what would you earmark as that thing?
Melissa Popp (52:49)
Yeah, gosh, let's see. You know, Content used to be a publishing problem. How much can we publish? How much can that help us rank, show up, convert, etcetera? I think now
2026 and probably the future of our industry, we need to stop thinking about how much we can produce and really start asking how what we produce is changing someone's mind. And whether that's getting them to fill out a form, buy something, act, push them further down the funnel, I think we really need to start changing how we look at content and what that end goal actually is. It can't be quantity anymore. It can't be half
half baked quality. You know, it has to be something that really is pushing and driving someone emotionally to make a decision.
Jon Clark (53:35)
Yeah, it really is about the quality and saying something different, right? Like what's that thing that you can add to the conversation? Yeah, I love that. this has been phenomenal. I've definitely learned a ton and I appreciate your being so candid. But please let our listeners know where they can find you. LinkedIn, your site, other podcasts.
Melissa Popp (53:54)
Oh gosh, ⁓ I pretty much live everywhere as a marketer. I'm still very active on Twitter and Bluesky and LinkedIn. You can find me on Instagram either at Melissa Popp or Poppupwriter. And then yeah, I blog on the RicketyRoo blog, sometimes on my personal website, but usually social media is a pretty easy way to find me.
Jon Clark (54:14)
Okay, perfect. Well, 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.