Joel Miller on AI, Human-Centered Marketing & the Future of SEO. A conversation with host Jeremy Rivera on the Unscripted SEO Podcast.

Unscripted Interview With Joel Miller

Jeremy Rivera

Hello, I’m Jeremy Rivera, your unscripted podcast host. I’m here with Joel Miller, who’s going to introduce his company. He’s going to introduce himself, but I want him to do so through the lens of why we should trust him as an expert.

Joel Miller

Thanks for having me on. My name is Joel Miller and my identical twin brother and I run The Sky Floor and we do digital marketing — that’s sort of the umbrella. But really what we’ve been focusing in on in this latest season is that we are problem solvers. We help you get to an outcome and a goal.

We’re not just here to push some buttons or apply your style guide to a website or whatever. We want to actually know what you want to achieve and then do whatever we can to help you get there. We really work across tons of industries — all kinds of work centered around the web, but it’s wide ranging.

One of the things that comes to mind immediately is that we had an ad in the Bears game day brochure that they pass out when you come into the Chicago Bears stadium. That’s distinctly not digital, but there was a digital component to help our client achieve their goal.

The Pyramid of Value: From Tactics to Transformation

Jeremy Rivera

I like the connection of marketing strategy and this evolution I had to go through myself with my own marketing and with the services that I provide — evolving past ‘what I do is edit meta titles and build backlinks and generate content.’

There was a fantastic interview I did where the guest Colby Wegter talked about this pyramid of value: at the bottom it’s executing on something specific, the next tier up is saving you time, and then a tier above that is enabling you to meet your goal or to become who or what you want to be.

That highest tier is enabling them to do or be the thing that they want to be — whether that’s a successful business person, a hybrid philanthropist and business person, or the highest selling person within the team. Enabling the business to do that — that’s a much better way to go about making sure your product and service delivers enough value that you’re going to be hired back.

Joel Miller

Yeah, especially in the AI era that we’re not only entering but living amongst. There’s a lot of fear and loathing — I’m not sure we’ll find out how true any of that is. But what you’re saying is a key distinctive: that’s something a computer will never be able to do quite the same way you or I can when we truly partner with a client and say, ‘we know where you’ve been, where you’re going, and we can use these tools to help get there.’

The human ingenuity to help you achieve those outcomes so you can be who you want to be — that’s going to be a distinctly human task still.

AI Adoption, Limitations, and the Gap Between Promise and Delivery

Jeremy Rivera

I saw an amazing graphic that showed of all of the millions of people — it was like 20 lines, and three-fourths of them were gray, meaning they had never used any sort of AI tool whatsoever. Then there were yellow and green lines of people who had tried it casually. And down in the bottom right corner, a small number of millions were those who use it daily.

We’re in that vanguard, that forward element adopting these things. But there’s definitely a moment of being the guy wearing Google glasses while taking a shower — so far ahead that it wasn’t yet useful, and you were embarrassing because you were thinking four or five steps ahead of people who really needed to mainstream-accept it before it was useful. I was in a convo with an air monitor manufacturer, and made references to AI capabilities and they were out of the loop on what was realistic or not. What’s real capability and what’s a lie.

The lie I’m talking about is that AI can do anything — that you can achieve any goal if you just feed in the right prompt and get the right output. I’ve talked to GEO and generative search engine optimization experts, and there’s always a big gap between what’s promised and what you can really deliver. What are some of the dilemmas you’ve solved, and what dilemmas remain after you analyzed the tools currently available?

Joel Miller

A story that immediately comes to mind is employing a custom GPT for sales enablement for one of our clients. I listened to 30 or 40 hours of incoming lead phone calls in their entirety, taking notes. I could have had an AI analyze transcripts and tell me what it thinks the weaknesses are — and some of those things would be true — but it’s not going to understand the tone of the conversation. It’s not going to understand how a particular pause came across. There are all these human elements I need to parse through.

So I took all those observations and put them into a custom GPT. Then using CallRail, we automatically grade incoming calls. The human work informs the AI’s categories; then the AI does what computers are good at: ‘Here are the six categories that need help — how did each call perform?’

Jeremy Rivera

The AI can only analyze the actual text. You might end up with a call where if you looked at the transcript, you have no idea where it went wrong. It looks like you hit A, B, C — said this, this, and this — and you can’t identify why they didn’t convert.

But if you listen to the call, you could hear the hesitation after friction point B. Maybe it’s ‘too expensive’ or ‘doesn’t have enough seats’ — but they didn’t say that. It was just an uncomfortable silence that you can feel. If you’ve done sales, you get that moment of ‘I just missed it.’ There’s definitely subtext that humans are programmed to understand.

Joel Miller

Exactly. I’ve always been fascinated by words that mean different things depending on tone. Someone might say, ‘Well, that’s great.’ Or they might say, ‘Well, that’s just great.’ Those are two very different things — but if you’re just analyzing text, the AI might say, ‘That was an excellent moment in the call where the person seemed pleased.’

Jeremy Rivera

My friend Matt Brooks at SEOteric says that ChatGPT is your least trained but most popular customer support representative. I’ve definitely hit that moment where it’s giving answers beyond its capabilities. And if you’re asking for answers, you immediately bias an LLM — because it cannot be objective in its response out of the box. It is an answer engine, and therefore the wave particle: if you observe it, it changes. You have to give it the input, which then influences the output.

Joel Miller

Right. And that’s one of the ways the human component has to stay part of it. We’ll likely have an advantage for some time because we’ve used it so much that we innately start to understand how to get better results — we feel the tension of ‘I know when I phrase it like this, it’ll give me a certain output because it thinks that’s what I want, and I actually don’t want that.’ I don’t just need a yes-man.

Jeremy Rivera

Yeah. Until GPT is programmed to grow a spine.

Joel Miller

Exactly. Yes, for sure.

Classic SEO Meets AI Overviews: Navigating the New SERP Reality

Jeremy Rivera

Tell me about your current interplay between classic — and I feel so old saying this — traditional SEO marketing and what’s emerging. We have this trifold system within LLM tools: it’s got its training database, which you can influence with SEO; and when it doesn’t or can’t find the answer, it does live searches. In some ways, GPTs and Claude are just fancy wrappers for Google results.

Joel Miller

It’s been changing rapidly for a lot of our clients in the last four months. We’re still at the edge of figuring out how to handle the drop in organic search results because of AI Overviews. We’re seeing — in a lot of cases — an increase in impressions, but clicks are down 40%. We’re live in the middle of figuring this out. But the key components of SEO are still the same; in a way, it almost just means you have to sharpen up. It might not be as easy as ‘you just need more interesting content more frequently’ — now we have to get a little more strategic.

Jeremy Rivera

What I’ve been seeing comes down to impacts by vertical. Publishers are getting hit harder and harder. Google is progressively more willing to poison its own well and disincentivize content creators. There’s a current lawsuit revealing more of Google’s monopolistic tendencies — their absolute willingness to rig pay-per-click bidding systems to maximize profitability, and to stack SERPs so many categories deep that organic results can’t even get on the first page.

Joel Miller

Yeah — clients are running searches and seeing: AI Overview first, then local pack results below that, then another group of sponsored results, and sometimes organic shows up with maybe two results at the bottom. Maybe not at all.

We’re seeing it in ad buying too: a lot of keywords that were working are no longer showing because that’s where the overview content goes. We’re having to shift more to high-intent combinations — ‘near me’ type queries — buying those more than we ever used to because otherwise it doesn’t even show up.

Jeremy Rivera

A strategy analysis piece is just dissecting and understanding SERP space more deeply. You can’t just rely on search volume numbers anymore — you have to understand the different ways it’s slicing. What type of results are showing up in your boring little niche. Concrete fences in Florida? Is it showing images? Guide videos? AI overviews where you can be cited? Gotta do your homework!

And it’s driving home: I need to work better with my email teams to drive value out of secondary captured audience systems. Use social media to capture audience, then direct them to pages and get conversions. It’s like fishing in ponds you created versus going out to the ocean. Those demand generation systems are more necessary now — you need that interdisciplinary capability as a marketer rather than being solely a Google reverse engineer.

Joel Miller

Exactly. The one thing — especially if you’re collecting emails — is that an audience can’t just be taken away if it’s your audience, whatever Google does next. If I’m capturing that information over time, I still control it. We’re helping clients understand: I know you were viewing this in a traditional lens — you put money in, you get leads out. But what if we actually started building you an audience you could take with you no matter what’s going on?

But some of that requires personality. Someone has to kind of be out there. On LinkedIn, something like eight times more likely if you post on your personal page versus a business page to get viewed and clicked. So we’re having to talk some clients into being out there more than they want to be.

The Human Renaissance: Authenticity in an Age of AI Slop

Jeremy Rivera

It’s a strange time. The more AI gets pushed to the floor, the more humans want to pass the sniff test: are you human? My wife just picked up an apprenticeship as a tattoo apprentice, and she was talking about how content in that space started creating AI processes — and now those types of videos that were super popular aren’t performing because people aren’t interacting: ‘so-and-so is just an AI avatar.’

Two years ago you had one person talking to the camera. Now AI can swap in anybody and these mass marketing tools realized the value of that niche. But the new Gen Alpha slang is — ‘that’s so AI’ — meaning slop, untrustworthy.

Those proponents of new technology who pushed AI content as the default — to get people to just accept it — have pushed such poor quality out there that now AI is synonymous with junk, with low quality, with no effort. That’s just the kind of outcome that’s not going to fly in your financial planning business like the Kalinka Group — you can’t AFFORD to be associated with low quality output, people are looking to TRUST you. Sadly, it’s happening SO often now. Why? The most advanced technology in the world is producing the lowest quality junk. It’s the Jeff Goldblum moment: you kept asking if you could and never stopped to ask if you should.

Joel Miller

It’s true. And scrolling through Facebook right now is kind of a nightmare — you’re asking yourself when the last time you saw a real clip was. Mostly it’s stuff that’s too outrageous to be true, and they hide it in ‘see more’: ‘these are created with AI for your entertainment.’ And you already knew that, because none of this actually happens.

What’s going to happen? Does Facebook start to monitor and stop it? Do they even care as long as we’re engaged and angry and commenting?

Adding GEO to the Stack: What Clients Are Asking in 2025

Jeremy Rivera

There are people in a lower acceptance quadrant who are going to be looking to us for answers to map out how they play in these playgrounds. What are some of the things you’ve had to add to your stack of analysis or processes to cope with our current reality over the past six months?

Joel Miller

A big part of it is GEO concepts — everyone wants to know how you’re going to help them show up in an AI search. They were saying that before they saw the organic dips, but now that conversation is getting louder. Another piece is just navigating the AI landscape with intelligence.

A lot of people are curious or use it at a very cursory level, and they know from our personalities that we spend a lot of time with it — they’re looking for guidance: how should we be viewing this? What should and shouldn’t we be using it for? We’ve been able to step in and say: here’s what we think, here’s how we see it working, here’s the stuff it’s good at, here’s the areas you probably shouldn’t be using it. Cautious optimism. I don’t want to be the person saying the world is falling apart because of it. We see more that clients are nervous about AI than excited about it — so we get to come in and say, ‘yeah, but here are some things you might want to consider utilizing.’

Case Study: Academic Tutoring and Resisting the AI Shortcut

Jeremy Rivera

What are some of the niches and industries you’ve been working with, and what answers have you been giving clients for their marketing issues in this age?

Joel Miller

One of our biggest long-term clients is a tutoring company — they do ACT, SAT test prep and academic tutoring. They’re called Academic Approach — highly recommend them. It’s interesting because a lot of their competitors are using AI tools for tutoring to replace human tutors: subscriptions, ‘use our AI tutor,’ $35 a month versus an hourly rate.

That’s something we’re advising them not to do. Their business model is high-touch, highly customized service. Even though AI can probably teach the basics, it’s not going to understand exactly what a student is going through — how they struggled through the first two or three practice tests, how to improve their executive function, how to hone parts of their skills that will actually apply to their academic and college careers. It’s about avoiding the shortcuts for the personalized service. It’s very close to how we run our own business: we don’t want to cookie-cut anything.

Jeremy Rivera

The irony of an education program turning around and offering AI tutoring is curious — but you’re right that it’s also a business model proposition: you’re offering a SaaS for a human custom element. The problem with SaaS is it iterates and evolves, you have to keep them addicted to the platform, give more value, and you need a lot more subscribers to keep up with that business model.

I would also be recommending finding ways to highlight the value of human participation. I think that’s going to be the ultimate response in the next couple of years: a human renaissance — these new tools can do all this cool stuff, but what can we silly humans do? We can do so much more.

Joel Miller

I think so too. If AI levels the playing field so everyone can bring any idea to market pretty fast — and let’s say it’s decently good, getting better and better, Claude Code can one-shot any idea you have and you can launch it — we’re still going to end up with the same problem: why does anyone care? And how do they even know that you made this thing? That’s always been the issue. That’s why marketing matters. That’s why an audience matters.

The human element is going to be the most important thing. It always was, but it gets highlighted when all these other barriers disappear — like ‘do you have enough money to make the thing?’ or ‘I need a video for this.’ If Sora or whatever it’ll be can make a whole Hollywood-grade movie in three minutes, the question becomes: is the story good? Do people care? And if they care, they probably care because of the person who came up with the story. There has to be some human element. It’s not just type in prompts and press go and we all subscribe. We still want to know who’s making it and why — all those underlying things that machines don’t innately have on their own.

Where to Find Joel Miller

Jeremy Rivera

Let’s wrap up here. If people are interested in the human element that is Joel Miller — where can they find you? Substack? LinkedIn?

Joel Miller

On our website The Sky Floor, we have a blog that I write frequently — you can subscribe there. I’m also on LinkedIn as Joel Miller, and I post some content to Medium — look up The Sky Floor or Joel Miller on Medium and you’ll find us for sure.

Jeremy Rivera

Fantastic. Thanks so much for your time and unique human insight.

Joel Miller

Thank you for having me.

---

Key Takeaways

• The highest tier of value any marketer or agency can deliver is not executing tactics but enabling clients to become who they want to be — that human partnership is the one thing AI cannot fully replicate.
• AI tools have real, practical uses in sales enablement and call analytics, but human listening remains essential for interpreting tone, subtext, and silence that transcripts can’t capture — as the CallRail custom GPT case study illustrates.
• AI Overviews are driving a significant drop in organic click-through rates (down ~40% for some clients) even as impressions rise, forcing marketers to get more strategic about SERP space, high-intent keywords, and owned audience channels like email.
• Building an owned audience — especially email lists — is now more critical than ever. Unlike organic or paid traffic, an audience you own cannot be disrupted by algorithm changes or Google’s evolving monetization strategy.
• For industries with high-touch service models (like academic tutoring), AI-based shortcuts threaten the core value proposition. Doubling down on personalized, human-led service is the right strategic counter to commodity AI tools.
• The flood of low-quality AI-generated content has made authenticity and real human presence the scarcest and most valued commodity online — a ‘human renaissance’ is already underway, and marketers who lean into it early will have a significant advantage.