SEO, GSO, GEO, AEO Explained: A 30-Year Catch-Up for the Rest of Us
AI-driven search is rewriting the playbook. But if you never learned the SEO 101 playbook, this post is for you. Also, I used an AI agent to shop for birthday cakes.
This week’s issue of Social Signals was written to Ozzmosis by Ozzy Osbourne, RIP.
Everyone is talking about the new SEO being powered by AI, but what if you never learned about the old SEO? This is your post.
Alphabet Soup! SEM, SEM, GSO, GEO, AEO. Whatever, right?
Everyone is suddenly asking for the AI-powered SEO handbook, but many marcomm leaders never learned how plain old SEO works. Maybe that’s you! And that’s okay.
When you start getting into the details of the new realities of SEO in an AI era, we tend to spend half the conversation translating vocabulary (e.g., “citations”), undoing panic (“our traffic might drop even if we are winning”), and explaining that this is not a brand new discipline so much as the next chapter of the same book.
We all want quick levers and guaranteed tactics, yet the real work is still slow, foundational, and shared across web, content, and search teams. Add in the fact that LLM algorithms change faster than Google ever did, and your beautiful charts or the latest study can be outdated by next week, and you get a perfect recipe for confusion or paralysis.
The hardest part is organizational, not technical.
We need the comms folks, the SEO crew, the site owners, and the analytics people rowing together, but the people in the room may not even control those levers.
You also need to reset KPIs: brand mentions inside AI answers can matter more than clicks. All of that requires careful framing so you do not overwhelm or sound like a doomsayer.
Short of going back 30 years and paying more attention to the SEO smarties, we have some catching up to do.
The move is to know (and teach) just enough SEO 101 to ground ourselves, show where we stand today, and position a POV as “true for now, evolving fast” instead of “we cracked the code.”
So let’s go back to the basics…
SEO & SEM 101
Think of SEO as the art of earning visibility by making your content irresistible to both humans and search algorithms, while SEM is the science of buying visibility through paid placements like Google Ads.
SEO focuses on organic results, so you invest time in on-page optimization, site health, and authoritative content that compounds value over months or years. There is a lot of guessing and test-and-learn involved. And the platforms like Google and Bing are always changing how things are indexed, so you have to stay on top of those trends so you can adjust (at best) and not be de-indexed (at worst).
SEM gives you instant shelf space at the top of the results page, but you pay each time someone clicks, and your presence disappears the moment the budget stops. It’s guaranteed placement, and it’s worth the cost. But you have to be vigilant of what people are searching for versus what you have to offer, what competitors are bidding on, and not to bid against yourself (especially if your company has multiple lines of business or multiple stakeholders buying search ads).
Great strategies treat them as complementary: SEO builds durable equity and brand trust, and SEM delivers targeted reach when you need speed, specific audience segments, or a test bed for messaging.
A (Brief) History Lesson
SEO pre-dates SEM. Marketers were tweaking titles, meta tags, and site architecture for AltaVista, Lycos, and early Yahoo in the mid-1990s as search engines became mainstream. And this was years before pay-per-click listings appeared on GoTo.com in 1998 and Google AdWords in 2000.
So these SEM paid ads were layered on top of an already-existing SEO practice, not the other way around. An understanding of “old SEO” foundations still mattered even as new monetization came about back then, and the same is true now as today’s AI layers emerge.
Do We Need A Separate Framework For GEO/AEO? No? But We Got One
Do we really need to add to the alphabet soup? Probably not, but we’re going to get served a bowl of it.
Even though at Google Search Central Live Deep Dive Asia Pacific this week, Cherry Prommawin and Gary Illyes led a session on how AI fits into Search and said we shouldn’t.
They were asked whether we need separate frameworks for Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO), and their presentation suggests that GEO and AEO do not require wholly new disciplines.
I ran into some of this last May and August when I wrote a series of Social Signals issues (reminder, paid Social Signals subscribers can read the full archives) including some of the emerging terms — SearchGPT, SOM, MTO — and was called out for inventing new terms when we don’t need new terms! That’s one of my favorite parts of this newsletter: hearing your feedback, pushback, and ideas.
I think y’all were right. And yet here we are in July 2025 talking about new terms like GEO and AEO. To me, the challenge isn’t learning the terms.
The challenge is that marketing and comms leaders who were never search experts to begin with are now expected to understand search and how it's changing. Or worse, they are pretending to know the GEO and AEO parts, but don't understand any of the foundations of SEO.
So beyond the terms themselves, we probably need a bridge framework to get everyone caught up to what they need to know, for the (real) experts to stand out, and for our brands to match pace.
So let’s dig into that a bit… and to be clear, I haven’t spent three decades leading SEO search, so others are way smarter than me on this and could probably do this much better!!
5 Legacy SEO Things You Should Know
Evaluate Your Content via E-E-A-T: Experience, Expertise, Authoritativeness, and Trustworthiness. Google Quality Raters lean on this framework to judge content quality.
Technical Foundations: Site speed, mobile friendliness, clean code, and crawlability determine whether your pages are indexed and ranked high.
Search Intent and Topic Clusters: Keyword queries of what people are searching tell us what kinds of content to create (to meet what they’re searching for!)
Backlinks and Digital PR: Quality links show a vote of confidence. Author mentions, social proof, and digital press hits all feed authority signals.
Structured Data / Schema Markup: Markup helps search engines understand entities, events, and products at a glance, increasing the odds of rich results.
6 New SEO Things You Should Know
GSO: Generative Search Optimization is the umbrella strategy. It’s about optimizing your content for the entire AI-powered discovery ecosystem—search engines, AI agents, voice assistants, copilots, etc.
GEO: Generative Engine Optimization is about feeding AI overviews the clearest, most verifiable facts so that your brand becomes the cited source. It’s a subset of GSO and focuses specifically on how your content shows up in generative search results like ChatGPT, Perplexity, or Gemini giving an AI answer.
AEO: Answer Engine Optimization focuses on voice, chat, and zero-click results. The goal is to earn the answer box and keep users from bouncing to a competitor. It is also a subset of GSO.
Chunkable, Promptable Content: Break long posts into concise sections with descriptive sub-heads. AI models love content that can be lifted into quotes or bullet lists.
Multimodal Signals: Images, video transcripts, and even podcast show notes now seed large models. Optimize your alt text and captions.
Source and Provenance Tags: Author bios, date stamps, and content watermarks reduce hallucination risk and boost trust in AI summaries.
AEO in practice —> Google’s AI Overviews Are Changing Search Behavior
Google’s AI Overviews are quietly but immediately changing how people interact with search.
According to new Pew Research, when these summaries appear at the top of results pages, users are significantly less likely to click on traditional links, with click-through rates dropping from 15% to just 8%.
Here’s an example of a logged out user searching for trending birthday cake recipes for 11 year old girls that gives an answer without any click at all. Note that this is a LOGGED OUT user, because Google rolled this functionality out through its entire product last month.
See those links in the summaries and on the right side? Even the links within the AI summaries get minimal engagement (1%).
And notably, more users are ending their search sessions entirely after seeing an AI Overview, signaling a growing reliance on AI as the final answer.
I do this. I bet you do, too.
PR is (Probably) a Very Good Investment for GEO (For Now)
Generative AI appears to rely heavily on earned media and journalism today. New research from Muck Rack shows that more than 95% of citations come from unpaid media sources, 85% from earned sources, and another quarter from journalistic sources. Half of total AI responses included at least one earned media citation.
These numbers highlight the role that earned media strategies play in GEO and AEO visibility today. Today.
I keep saying “today,” because we know from the last 30 years of SEO, things are gonna change. But I can tell you from immediate recent experience that earned media hits, press releases on a wire, blog and website content, and social media content are all populating for clients in launches.
Example: Cetaphil Turns to AEO to Top ChatGPT’s Skincare Answers Gen Z
Gen Z is literally standing in the Target skincare aisle and asking ChatGPT, “What cleanser should I buy?”and Cetaphil wants to be the first answer they see. Of course.
According to Ad Age this week, the brand is rewriting every product description, tuning its GEO and AEO, launching its own chatbot, and weaving AI into paid, earned, and social media to make that happen.
Early wins from this case: Amazon Prime Day traffic from AI chat tripled 33× YOY, and AI-driven Google campaigns are up 30%. Hmmm… maybe there’s something to this, huh?
The Agentic Web Is Here, and It’s Already Reshaping AEO
Just as we’re getting all fluent in AI-powered SEO and wrapping our heads around AEO strategy above, the game is already moving forward again with AI agents that can browse the web on our behalf. This is a “RIGHT NOW” thing.
As of this week, OpenAI’s ChatGPT Agents officially rolled out to all paid users, and they aren’t just summarizing content anymore. They’re clicking, scrolling, filling out forms, shopping for birthday gifts, and even playing chess (albeit poorly).
Here’s my example using ChatGPT Agent yesterday to complete SKIP ALL THE TRADITIONAL MARKETING FUNCTIONS…
My agent went away and researched trendy cakes. It recommended a unicorn cookies‑and‑cream layer cake, gave me the baking instructions, and put all of the ingredients immediately into my Instacart. Note that it chose the brands (e.g., Essential Everyday, Kemps, McCormick, Betty Crocker) itself!
From here, it’s auto-loaded and ready for me to checkout (I had to click in and go find this to show you), and I didn’t have to do ANYTHING.
It’s not perfect, but it was seamless and intuitive. And this is literally week one. This is the worst this technology will ever be.
And again, think of how many traditional marketing and communications functions this skipped over.
Social media inspiring a trending cake idea
Google searches leading to articles and posts about trending cakes and/or recipes
Recipe sources, including publishers, recipe sites, and CPG food recipe landing pages
All marketing efforts from CPG food companies angling you to be aware and consider buying their brand over others
The shopping experience of Add to Cart, suggested items, and Point of Sale (POS)
While the current experiences others are having are also glitchy, the implications are serious: we’re moving from an internet optimized for humans browsing to one where bots roam the web, mimicking human behavior but skipping past ads, skimming content, and never meaningfully engaging.
It’s an early but important signal that AEO is about more than surfacing the right answer: it’s about preparing your content for a world where agents, not people, are doing the “searching.” In fact, I think agents may have just destroyed AEO before we even got to a best practice stage.
That means the old SEO mainstays listed above still matter, but we now also have to write for invisible visitors with no emotions, no patience, and no brand loyalty.
You’re not just optimizing for humans asking questions anymore; you’re now optimizing for bots answering them.
And in a world where agents skip your display ads, skim your CTAs, and auto-generate replays of your site without clicks, being the answer source is the win. AEO won’t be about being cited in summaries very soon, and instead it will be about being machine-legible and bot-actionable.
Generative Answers, Paid Placements Are (Probably) Coming
Just as we had a short period of SEO focus before SEM products were created 30 years ago, we can expect the current LLM engines will add some kind of advertising product.
Today (July 2025):
Perplexity is testing “sponsored follow‑up questions” and courting big-brand partners like Indeed and Whole Foods, although answers are still generated by Perplexity, not the advertiser.
Google/Gemini has gone further: Search and Shopping ads now render directly inside AI Overviews on desktop (mobile rolled out earlier), essentially turning generative summaries into new ad inventory.
OpenAI hasn’t launched classic ads in ChatGPT yet, but it quietly added product recommendation modules and is signaling “free-user monetization” (read: ads or commerce rev-share) as soon as 2026. I was quoted in The Drum about Open AI’s paid placement plans back in December.
Anthropic’s Claude is monetizing via enterprise licenses and cloud marketplaces—not ads—at least for now. That’s a human em dash by the way. I typed that out.
So we're in the in-between today, mirroring the 5 year gap between SEO and the birth of SEM in the 1990s, and when those products launch it will revamp everything we think we know about SEO, GEO, and AEO.
Will we all need to be SEM experts at that point, too? Maybe.
What’s Next in the EO’s of SEO, GEO, AEO?
What’s next is a moving target, and that isn’t going to change.
Large language models will keep reshuffling ranking signals, voice and visual search will mature, and shoppers will consult agents we have not built for yet.
The momentum of new innovations like AI companies launching their browsers, paid placements, and introducing agents to paid users means that a lot of what we say are “SEO 2025 Best Practices” today could change quickly tomorrow.
In fact, based on my birthday cake example above, I think agents are assuredly to absolutely destroy any best practices you’re reading about on LinkedIn (or Substack!) today. So my advice is work on learning the fundamentals while holding loosely the mandates.
The play is to get comfortable iterating in public: learning about the next thing as it comes, auditing content every quarter, expanding schema to new data types, harvesting credible citations, keeping the gas on your PR and social efforts, and pressure-testing answers with your own research before the market does. Huh, sounds like another whole post right?
Next is not a milestone; it is a moving sidewalk that never slows down. Aren’t you glad we’re out for a walk together? -Greg
📲 Relevant Social Signals issues about AI search from the archives:
Is SearchGPT the new Google? OpenAI's new AI-powered search engine wants to be your default search choice (August 2024)
The AI Everything Issue: Search, Ads, School Policies, and… Water Bottles? (Nov 2024)
(reminder, paid Social Signals subscribers can access the full archives. Thanks for your support!)
🧠 Recap of After Now by Andus Labs
This week I had an opportunity to attend After Now by Andus Labs, a day filled with big thinkers asking the biggest questions as facilitated by
and his team.The day toggled between abstract questions (“utopia or doom?”), exploring how science fiction has helped shape our perspective on all of this AI change (for better and worse), and calls for concrete action.
A few of my takeaways from the sessions:
AI as a “non-specific amplifier: Douglas Rushkoff's concept that whatever mindset or intention you bring to AI will be amplified. This means that passivity or imagination will be reflected back and magnified by these systems.
AI as an instrument: Brian Eno suggested we should treat AI not just as a tool but as an instrument that requires time, depth, and practice. And to embrace non-linear and chaotic results!
Rethinking media: The Atlantic's Nicholas Thompson urged us to persuade both people and machines in this new landscape.
Early research will shape longterm public perception: Remember that study out of MIT that had everyone sharing headlines about how ChatGPT makes you dumb? The study’s architect, Nataliya Kosmyna, presented and demonstrated how a tiny sample size of 59 people can create a dataset that can be wildly misinterpreted and potentially damaging to the advancements we want to make (here’s my take on it from last month: Is ChatGPT Making Us Dumb or Are the Headlines? via The Cave Project).
Beyond automation: The real opportunity isn’t just automating existing processes but creating entirely new structures that operate at the speed of AI. Focusing too much on "making a faster horse" is missing the point.
Collective intelligence (“scenius”): Progress will come from collaborative experimentation (aka a lab mentality) rather than waiting for easy answers or magic switches. Share more! Learn more! Then share more!
The consensus of the event aligns well with my POV: it’s not enough to speculate about what might happen; we must actively shape how AI integrates into our lives and work. It's both an opportunity and a mandate.
Cheers to the future we'll shape together! 🥂✨ -Greg
📢 See Greg Speak
Digital Summit Minneapolis: Who Knew the Best Brand Advocates Already Work Here?, Minneapolis, MN - August 6-7
University of St. Thomas: Marketing, Insights, and Communications Team Onsite, St. Paul, MN - August 12
Minnesota Blogger Conference: The State of Blogging in 2025: The Blog Is Dead, Long Live the Blog, Minneapolis, MN - September 13 (GET TICKETS)
PRSA Iowa's OctoPRfest Conference: Reboot Required: Rewiring Strategy for Today’s Cyborg Audiences, Des Moines, IA - October 9
Interested in Greg speaking at your event? Hit me up.
🐝 Bee Acquired by Amazon and the Future of AI Wearables!!?!?
In a paid-subscriber-only issue of Social Signals last month I told you about my relationship with the Bee Pioneer AI wristband assistant.
Key quotes:
Bee has essentially become an extension of my memory. A wearable second brain. A small language model of me.
It’s a life feed, with meaning.
If mass adoption happens, what does "privacy" even look like? Are we heading toward a world where the right to be unrecorded has to be explicitly requested?
You can read the whole piece here ⤵️
My Bee, Myself, and I: Living with an AI That Remembers Everything
Imagine if every time you said "I’ll take care of that tomorrow" or "We should really follow up on this," someone actually remembered. Not just remembered it, but turned it into a to-do list, wrapped it in context, and reminded you with kindness. That’s what it feels like to wear Bee, an AI listening assistant that captures, processes, and reflects back the shape of your day.
I’m bringing this up because just this week it was announced that Amazon has acquired the Bee, according to a LinkedIn post by Bee co-founder Maria de Lourdes Zollo.
Per TechCrunch:
This acquisition signals Amazon’s interest in developing wearable AI devices, a different avenue from its voice-controlled home assistant products like its line of Echo speakers. ChatGPT maker OpenAI is working on its own AI hardware, while Meta is integrating its AI into its smart glasses. Apple is rumored to be working on AI-powered smart glasses as well.
HUGE SIGNAL. As I wrote for you last month, we talk about AI and ChatGPT as productivity tools, but this is something different. An AI personal listening device doesn’t just help me get more done; it helps me see what I already did. Process it. Understand it. That alone has been transformative.
As we consider Harvard Business Review’s Top Use Cases of AI in 2025, I think it’s fascinating how use cases have already evolved from generating ideas and making content to helping us organize our lives, find purpose, and pursue more knowledge about ourselves. In particular, because that is what Bee unlocked for me…
I also wrote extensively in that issue about the privacy implications, and I don’t have the answers yet. What I do know is this: as these tools become more powerful, bolstered by companies like Amazon, and even cheaper (although $50 is CHEAP), so does the need for etiquette, transparency, and maybe even legislation.
We’re used to thinking about data privacy in terms of platforms. Now we have to think about it in terms of people.
The future of AI wearables isn't just about productivity. It’s also about permission. And the sooner we have that conversation, the better prepared we’ll be. -Greg
⚡️ Social Signals
Why are you seeing so many old posts on LinkedIn? LinkedIn’s algorithm now favors relevance over recency, intentionally surfacing older posts from meaningful connections to prioritize content that drives real professional value, like job changes or expertise. Are you noticing this, too?
Roblox launched a new licensing system that allows creators to build official experiences using popular franchises like Stranger Things, Twilight, and Like a Dragon, giving smaller teams access to IP that was once limited to major studios.
I’ve been talking about neural wristbands since CES 2023! Meta is actively discussing bringing this tech to your wrists soon: "Imagine controlling your devices with a subtle hand or finger gesture. Our cutting-edge research turns intent and muscle signals into seamless computer control. This breakthrough wrist technology is redefining how we interact with computers—intuitive, precise, and ready for the future.”
DuckDuckGo introduced a new filter that lets users hide AI-generated images from search results, reflecting some signals of demand for control over AI content.
MIMA (the Minnesota Interactive Marketing Association) is looking for new board members. Check it out here.
Good Read of the Week: Can The Washington Post’s TikTok Guy Make It Without The Post? Dave is a great guy and once responded to my DM to go speak to a class at Drake University. I’m pulling for him!
Video of the Week: The Last Advertising Agency on Earth, circa 2010 by Saatchi & Saatchi Canada, is a fantastic re-watch.
Term of the Week: “high-dimensional co-occurrence token karaoke.”
Quote of the Week: “There’s a new programming language. It’s called English.” by Nvidia CEO Jensen Huang.
Insta of the Week: HearForTheSpeakers documents all of the hidden speakers around us at places like Disney, shopping malls, and more. I love this account so much.
Drone Adoption Signal of the Week: Drones being used to indicate exit for an event in Osaka, Japan.
Reel of the Week: Welcome to Amy Sedaris’ dollhouse.
Thread of the Week: The view of the Pyramids of Giza from a nearby Pizza Hut and KFC located a quarter mile away.
Social Video of the Week: My “Oops Wrong Party” Party notched more than half a billion views across TikTok and Instagram last week. So much fun!
Keep going! 🚀✨
Greg