I had a passive income project sitting comfortably at the top of Google. It owned all the important keywords for its industry. I had turned on AdSense, walked away, and let it run.
At the time, I had made a conscious choice to prioritise my life over monetising every waking second of my day. I was raising my children, self-building a home for my family—the big dreams that make you “time-poor” but life-rich. Then, the rug got pulled.
Without warning or notification, Google quietly changed a category I had selected on my Google Business Profile. Just like that, the profile vanished. It stripped my site of its localised authority in the SERPs, and all those hard-won gains evaporated as the site plummeted to page 4. I am still fighting to get that profile reinstated, armed with registered trademarks and documentation.
But that wasn’t the moment I realised the game of search had fundamentally changed. That moment came a few weeks later, courtesy of ChatGPT.
A customer of another one of my projects – a directory site, had completely changed their service offering. They asked to be removed from the directory because they were still getting inquiries. I promptly deleted their listing. Weeks later, I got an angry email. This customer—an older individual who was increasingly relying on ChatGPT as their primary search engine—was insisting I hadn’t removed them. ChatGPT was confidently telling them they were still actively advertised on my site. The customer thought I was lying.
That was my “Aha!” moment. Generative Engine Optimisation (GEO) and Artificial Intelligence Optimisation (AIO) aren’t just new ways to get traffic; if you aren’t careful, they can actively work against you. The AI was hallucinating outdated information, and the user trusted the machine over the human.
We are living in an incredibly exciting time for search, but it’s the Wild West. We haven’t yet had our “Google moment” in the AI space—that tipping point where a specific AI tool becomes a verb, wiping out the Altavistas and Yahoos of this generation. Until then, everything is moving at breakneck speed, and it is exposing a massive amount of industry bullshit.
The “AI Slop” Echo Chamber
If you want to understand the current dysfunction of the tech and marketing industries, just look at the job market.
Hiring managers are feeding simple prompts into ChatGPT to generate highly complex, abbreviation-heavy job specifications. They are over-specifying what they actually need because they don’t fully understand their own brief, hoping some “expert” will swoop in, knowing exactly what they want. On the other side, applicants are taking those over-produced job specs, feeding them into their own AI, and generating perfectly tailored, keyword-stuffed resumes and cover letters – only to have them filtered out by the job posting site’s AI detector.
Everyone is inflating their offering. It begs the question: why don’t we just let the two AIs talk to each other and figure out what they want? (although it may take a while)
This same “AI slop” is polluting the SEO industry. “Experts” are riding the wave of new acronyms (GEO, AIO, LLMs), pretending they have mastered a black box that even the AI developers themselves can’t fully explain. To truly understand how these models think, I recently installed a standalone, offline version of Google’s Gemma LLM on a laptop just to experiment with it in isolation.
When you strip away the internet connection, you realise that every version of these tools is a product of its training data. If that data is flawed (like the infamous “how many R’s in strawberry” debacle), the output is flawed. Where it’s a current research query, these chat interfaces are still heavily relying on traditional search engine results to source their answers—because, let’s face it, crawling an index is efficient.
Which brings us to the core truth: succeeding in SEO, GEO, and AIO all require the exact same foundation.
Building the Machine: A Case Study in Real Authority
Recently, we did a major website overhaul for a high-end construction industry firm. If I had to explain to a non-technical person what we spent the bulk of our time doing, I would say we spent it carefully pruning.
By default, platforms like WordPress generate a massive amount of “thin content.” If you create a custom taxonomy to tag a project’s location (e.g., “Dublin 4”), the CMS automatically creates a blank, public-facing archive page for it. Search engine crawlers (and AI scrapers) hit these pages, see nothing of value, and label your site as thin, unhelpful, or poorly structured.
To fix this, we didn’t just delete the pages; we orchestrated a precise, dynamic routing system. We redirected those empty taxonomy archives to custom, root-level “Hub” pages.
Visually, to the user, these hubs function as highly intuitive search filters. It’s not quite as simple as that, but under the hood, each one is a targeted, standalone SEO landing page. We used the client’s own project portfolio to dynamically populate these pages.
But we didn’t stop at the visual layout. We injected heavy, customised Schema markup (JSON-LD) directly into the code of these pages. We marked up genuine client feedback, exact job specifications, project sizes, completion dates, and energy ratings.
Why? Because Large Language Models do not have eyes. They read code. When ChatGPT or Google’s AI Overviews scrape that construction firm’s site, they don’t have to guess what the company does. The schema spoon-feeds the AI the exact entities, and as a design project for example: This is a Creative Work. It was built in 2024. Here is a 5-star aggregate rating from real humans. Here are the awards it won.
It is a microcosm of what modern search optimisation is all about: demonstrating undeniable, structured E-E-A-T (Experience, Expertise, Authoritativeness, and Trustworthiness).
The Ultimate Signal: Being “Real”
When an AI engine decides which firm to recommend to a user, it isn’t looking for the site with the most AI-generated blog posts.
I believe the single most important signal an AI looks for is proof that your business is real, and that’s going get more and more important as time passes.
It wants to see a well-structured, professional digital footprint. It cross-references the entities on your site. Do real humans work there? Can the AI find those humans participating in industry forums or actively posting on LinkedIn? Does the site have genuine, verifiable reviews? Is the portfolio of work clearly on display, accurately dated, and traceable to show an ongoing, historical development of a craft or business?
The game has changed, but the goal remains the same. The winners in the era of Generative Engine Optimisation won’t be the ones who figure out the best prompt. The winners will be the ones who do great work in the real world, and build a digital architecture clean enough for the machines to understand it.