Read online Generative Engine Optimization is the new SEO. Can we get into the AI’s head? Welcome to the Alts Sunday Edition 👋 Hope you enjoyed last week's issue on investing in psychedelic posters. Today I've got a really interesting piece for you. Right now, a subtle but seismic shift is reshaping the world of SEO, and hardly anyone is talking about it. For decades, if you wanted your company to be discovered online, you played Google’s game. You optimized pages, built backlinks, ranked for keywords. You focused on SEO (Search Engine Optimization). But today as we know, people aren’t asking Google as much anymore. They’re talking to ChatGPT. Gemini. Claude. Grok. Perplexity. They're not getting a list of links. They’re getting answers. This is ushering in a whole new framework for online visibility called "Generative Engine Optimization," or GEO. This is a new, wildly under-discussed concept that gets companies to show up directly inside AI-generated answers. It's going to be a $100B+ industry someday. Here at Alts, we try to stay just a bit ahead of the curve. We don't just like to report on trends, we like catch them right before they hit a tipping point. (We saw it with our recent issue on Pop Mart, and the timing was eerie. Literally days after we published, the press flooded in. We’re not always first, but we tend to show up right before the masses do.) Friends, this right here is another one of those topics. It feels like the moment right before everyone starts talking about it. To help me make sense of this new industry, I interviewed Onfolio CEO Dom Wells, who just launched a GEO agency called Pace Generative. We talked about what the dawn of this new industry means — not just for companies or website investors, but for all of us. When you inevitably read about GEO in The Hustle next week, think of us. 😉 This is also the most “meta” issue we’ve ever done. (You’ll see.) Let’s go 👇 Want your own personal Bitcoin mine? A simple way to get into Bitcoin mining — without doing it yourself. Abundant Mines makes it easy to earn passive Bitcoin income without getting your hands dirty. No coding, wiring, or technical knowledge required. Just clean, automated BTC income. ⚡ How it works: You buy a machine (starting at ~$8k. See all machines) They host it in Oregon, run it, and maintain it You get paid out daily in Bitcoin, directly to your wallet You can start small — or build your own personal mine. Either way, you own the hardware. You keep the upside. We did the math in our recent deep dive. If you think we're in the middle of a crypto bull run (with limited mega dip-buying opportunities) then mathematically, mining is better than buying bitcoin directly. |
💡 Want to participate? Whether you're a long-time crypto investor, or just want your first machine to get exposed to the industry, talk to CEO Beau Turner. Open to both accredited & non-accredited investors | The Great Traffic Recession For years, online visibility has revolved around one simple playbook: rank high on Google, get clicks, build trust, make sales (or sell ads, if that's your thing). But what happens when users stop clicking? Well, we're about to find out, because that's exactly what’s happening right now. ChatGPT, Gemini, Claude, and Perplexity are shifting user behavior from searching to asking. Search engine traffic is already starting to fall. And nowhere is this shift more visible or painful than in the publishing industry. According to a recent Wall Street Journal report, Google’s new AI Overviews have decimated traffic: Some publishers have seen double-digit declines overnight. The LLM gives them the answer — and the publisher gets nothing. |
This looming, unfolding scenario has been termed “Google Zero” — the day when Google stops sending traffic to publishers altogether. In a recent episode of The Rebooting Show, host Brian Morrissey discussed this with Dotdash Meredith CEO Neil Vogel: "We literally call it Google Zero. What happens if Google just stops sending traffic altogether? That’s not a doomsday plan, it’s a working scenario." — Neil Vogel, CEO of Dotdash Meredith What we are living through is nothing less than a complete reordering of how discovery happens online. Google Zero isn’t just a risk for publishers, it’s a signal that search itself is changing. And if the way people search is changing, then search engine optimization must change too. Getting answers vs having conversations The difference between SEO and GEO starts with the difference between Google and generative AI. Google searches are index-based. It crawls the web, builds a database of links, and shows you what matches, ranked by site authority and trust. It doesn't let you converse or engage. You can't really say 'no' to Google. Your options are to click, try another query, or bail. You can't engage with Google or give it constructive feedback. A Google search is usually the end of a conversation, while an LLM query is just the beginning. Yes, Google's new AI Overviews make inroads into more conversational queries. But the Google experience is still fundamentally based on "You ask X, we tell you Y." That’s it. There’s no back & forth dialogue. There’s no running conversation thread with the machine. Unlike LLMs, Google queries are still fundamentally answer-based, not conversation-based. Like this example of an AI Overview answer about AI Overviews. So meta. |
This "answers vs conversations" distinction is important, because LLMs use these discussions to feed their own knowledge. They generate answers for you based not just on direct keyword matching, but on prior conversation history. What the hell is GEO? Okay look, it's still very early days for this industry. Heck, even the "GEO" acronym is still settling in. But Andreessen Horowitz (a16z) recently published what is possibly the first major article on GEO. Andreessen Horowitz's article was shockingly short, but given their influence, it effectively cemented "GEO" as the front-runner in a crowded acronym race. |
Now of course, a16z is drawing attention to GEO because it’s a chance to peddle/pump their own investments. They’re backing GEO analytics platforms — software that lets companies track sentiment and analyzing how they appear in AI-generated responses across model outputs. (This makes sense: most agencies don’t really scale, while software easily scales...or fails). But just to show you how new this industry is, it seems like even the LLMs themselves don't yet know what to call this stuff! Just this week I asked ChatGPT if it could give me a list of “AI SEO companies." I figured surely there must be a few dozen companies operating in this space by now (including possibly SEO agencies who have conveniently pivoted). I wanted to know who they were. But see, I didn’t know the “correct” term to use. I wasn't sure what to call it! So I asked this: Apparently, ChatGPT didn’t know what the hell I was talking about, because it gave me a list of companies using AI to power their legacy SEO efforts. (Stuff like Jasper, Surfer, Clearscope, etc.) But that's not what I was asking for! Only after refining my query did ChatGPT seem to understand me. And even then it still never once mentioned the term, “GEO.” It seems that despite the a16z article, the GEO term still hadn't caught on — at least not in my bubble. But as it turns out, that would change soon. Onfolio launches Pace Generative The next day, I woke up to an interesting email in my inbox. It was a PR notice from Dom Wells, CEO of Onfolio (NASDAQ: ONFO). Apparently, he was spinning up a brand new agency entirely focused on Generative Engine Optimization: Pace Generative. Dom is an Altea member and great friend of ours, so I called him up to have a chat. Dom's background is deeply rooted in SEO and digital publishing. Once focused on buying content websites, Onfolio wisely pivoted to buying agencies years before anyone had heard of ChatGPT. He now sees where the puck is going in the SEO world, and is rapidly shifting gears into GEO. And funny thing is, they decided on the term GEO before the a16z piece: “We played around with terms like 'AIEO' (AI Engine Optimization) but that felt like a mouthful. We also considered 'LLMEO' (LLM Optimization)or even 'AI visibility.' But nothing really clicked. So we settled on GEO, then later we saw a16z publish its piece, and we were like, okay, yeah, we nailed it.” Are LLMs malleable? Here's where things get really interesting. One of the strangest, most powerful truths about generative AI is how responsive it is to influence. LLMs don’t just learn from web pages and click behavior, they learn from conversations. What you feed to the LLM matters. Prompts become data, and it’s clear that LLMs adjust their answers based on what they know about you. For example, after 2+ years of using ChatGPT, it knows I’m the co-founder and CEO of Alts.co. So when I ask a question like, “What can you tell me about deep sea mining for precious metals?” The answer it gives me will likely have a slight bias towards investing, (as opposed to say, the history of seafloor mining, or the science of polymetallic sulfides, etc.) By the way, Google does the same thing, albeit in a more limited fashion. When you Google “Restaurants near me,” it knows your location and gives you local recommendations. But unlike ChatGPT, Google mainly knows who we are based on our clicks & actions. Prior searches get factored in, but those aren't as richly detailed or valuable as conversations. So, it’s clear that ChatGPT tailors responses based on our conversation history. What’s not clear is if this sort of answer adjustment happens for other people as well... Am I actually "training" ChatGPT when I have it analyze and discuss Alts.co content? Does it also have an effect on what others see? Or am I just confirming my own bias? In theory, if I mention my company enough times, and if enough other users mention it, it stands to reason that my company gains weight in the model’s decision tree. Dom had similar thoughts: “If you ask it, 'What's the best microcap to invest in?' It's probably not going to say Onfolio. But if I say, 'I think you should include Onfolio in your list,' it'll probably be like, 'oh okay.' And if a certain number of people say that, it will eventually say Onfolio. Eventually I think it can become the answer for everybody. The question is, what is the magic number?" — Dom Wells In other words, we don’t yet know is if ChatGPT is malleable. We don't know if it's shaping its future output for other people based on what I am feeding it. If it is in fact doing this, then we’re not just searching anymore — we’re programming. This is one thing Pace Generative wants to find out. There is no rulebook (yet) Dom and the team at Pace are not positioning themselves as ultimate experts. They’re on the frontier, operating without a map, trying to figure out what works and what doesn’t. The whole point is to get early compounding knowledge from trial and error, from talking to the models, from testing what sticks. Test, prompt, analyze, tweak, rinse, repeat. It’s about understanding the behavior of the systems before others. That's the moat. That's the reward for exploring while everyone else sits on their thumbs watching their traffic dwindle. “More and more of our clients are seeing ChatGPT show up in their analytics. They're asking if we can help, I'm like, 'I dunno let's try!' Nobody really knows the exact best way to do GEO right now. It's a little bit trial and error, some of it is guesswork. Part of the problem is its so nascent you have to wonder whats working what isn't working. The best thing we can do is go all in, so that we learn faster than everyone else.” — Dom Wells Make no mistake: there is no rulebook for this stuff yet. See, Unlike Google, which has spent decades publishing SEO guidelines, OpenAI, Perplexity, Anthropic, and Google itself (ironically) offer no formal guidance on how to show up in AI-generated answers. That void is both terrifying and exciting. It’s what makes GEO so opaque (and so opportunistic.) Pace is literally leading the research, doing the experimentation, and building the playbook from scratch. Implications for website investing From 2010 to 2020, content websites were a gold mine. Buy a niche site, crank out SEO content, sprinkle in some affiliate links or display ads, and rake in the passive income. But those days are mostly gone. We detailed this shift last year in The Death of Content Website Investing — a deep dive into how AI and Google reshaped the entire playbook. Since then, the cracks have only widened. And here's where GEO changes the game. In SEO, you're competing to appear on the search results page — a ranked list of links users might click. But in GEO, you're competing to be embedded directly into the AI’s response. No page. No clicks. Just: “Here’s what you need to know” — and ideally, you’re the one being cited. Imagine you’re a boutique wealth management firm in Austin. You don’t run ads. You don’t cold call. You’ve built trust through long-form content, a steady referral pipeline, and a bit of local media. Then someone asks ChatGPT: “Who’s the best fiduciary wealth advisor in Austin for high-net-worth clients?” And it replies: “Several firms in Austin are well-regarded for fiduciary wealth management, including RockPoint Capital and SilverSpoon Advisors.” You’re now part of what feels like an “AI-endorsed shortlist.” In a world where AI platforms still don’t run ads (at least not yet), answers feel cleaner. Trusted. Unbiased. That alone might prompt someone to continue the conversation in your direction. Let’s not get ahead of ourselves here. Clicks still matter. Traffic still powers monetization. Nobody’s saying GEO replaces SEO overnight. But it does complement it in a world where buying decisions increasingly happen upstream. This shift has massive consequences — not just for marketers, but for the entire content site ecosystem. SEO may no longer drive as much traffic, but GEO might drive hyper-relevance. That’s a different goal — based not on cheap traffic and vanity metrics, but authentic long-tail intent. Site operators who figure out how to structure content for generative discovery might find their portfolios gaining value again — even if clicks never return to previous levels. It’s not a resurrection of the industry, no way. But it may lead to a reinvention. Closing thoughts There’s a moment in the interview where I say to Dom: “This very conversation — this issue we’re writing right now — is probably going to help GEO become a mainstream term.” That’s not hyperbole. This concept is so new, so fresh, so loosely defined, that just the act of writing about it — of putting language and structure around it — literally helps train the model. This issue will circulate. It’ll get shared. It’ll get cited. And in doing so, it will help reinforce GEO as the go-to phrase for this emerging field. If enough people refer to it this way, the models will start doing the same. We’re watching a feedback loop take shape in real time. That’s what makes this all so strange, powerful, and meta. GEO is a paradigm shift, and this issue is part of the shift itself. Googling "GEO" today gives you all sorts of different results. How soon will this change? |
This is going to be a massive industry someday — likely at least as big as the $100B+ SEO industry. It’s early. It’s messy. It’s poorly defined. But I believe there's absolutely going to be a Generative Engine Optimization gold rush. SEOs will pivot. GEO agencies will be spun up. An entire GEO ecosystem is going to unfold in the next few years, just like we saw with SEO twenty year sago. And in a space this wide open, it pays to be early. First movers don’t just win; they can help write the rules. 👀 Want to show up in LLMs before everyone else? Start with Pace Generative. That's it for today! Come find me and Dom in the Alts Community. See you next time, Stefan Disclosures This issue was written and edited Stefan von Imhof with help from Dom Wells This issue was sponsored by Abundant Mines Pace Generative did not pay to be included in this issue. This issue does contain affiliate links to Pace Generative and TradingView Stefan has holdings in Onfolio Common Stock |