Found in AI: AI Search Visibility, SEO, & GEO

Start Here: The 5-Step AI Search Visibility Strategy from 80+ Episodes

• Cassie Clark • Episode 84

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 26:33

Send us Fan Mail

It's the official one-year anniversary of Found in AI! 

In this solo episode, I go back through the entire Found in AI archive and pull out the five-step strategy for AI search visibility — built from a year of guest conversations, prompt tests, and a few things I got wrong along the way.

Whether you've been listening since episode one or you just found the show, this is your starting point.

What's covered:

  • Why AI search visibility is a cross-functional problem that shows up looking like a content problem
  • The five-step strategy: Describe → Structure → Refresh → Corroborate → Measure
  • How the FSA Framework (Freshness, Structure, Authority) fits into a broader visibility strategy
  • Why fixing your brand description across surfaces is step one — before you touch your content
  • The weekly assignment you can do in an afternoon to find your actual roadmap

Episodes mentioned:

Step 1 — Description

Step 2 — Structure (FSA)

Step 3 — Freshness (FSA)

Step 4 — Authority (FSA)

Step 5 — Measurement

If you're new here:

New to AI search:

Trying to measure it:

Selling this internally:

Want the arguments:

I'm an AI search visibility consultant and a fractional content strategist for startups and enterprise brands. If that sounds like the kind of help you're looking for, email me at cassie@cassieclarkmarketing.com. 

Or request your AI Search Visibility Audit: https://cassieclarkmarketing.com/ai-search-visibility-audit/

Let’s connect:

LinkedIn → Cassie Clark | AI Search Visibility Consultant
Website → https://cassieclarkmarketing.com

SPEAKER_00

Over the weekend there was a discussion on Reddit. You all know I love Reddit for stuff like this about what makes an AEO GEO strategy actually work. Now the poster had mentioned they had great SEO, they were trying a few things, but none of it really seemed like it was working. I commented and mentioned the show like, hey, here's a resource you can use, but when they asked me where to start, I said start with the first three episodes and then skip to the ones about entity authority. I have been thinking about my answer since then because that's not really a great answer. There are over 80 episodes of Found in AI as of this recording this morning, and that's a lot of information to sort through if you are building your strategy from scratch. Now, because it's also the official one-year anniversary of a show today, August 11th, I've gone back through a year of the episodes to pull out the actual strategy. Hey, welcome back to Found in AI. I'm Cassie Clark, an AI Search Visibility Consultant. And if you're new here, this is the show where we talk about GEO, AEO, and what it actually takes to show up inside AI generated answers. Today the show turned to one. It is also my birthday. Yeah, I launched last year on my birthday, partly on a whim, mostly on a whim, if I'm honest, but also because I knew I would remember that anniversary date. And at the time there were exactly zero shows dedicated to AI surge optimization. One year and 80 some episodes later, here we are. Okay, let's get into that strategy. Okay, so I want to start with what changed in my thinking over the last year because it kind of reframes everything else I'm about to say in this episode. When I started the show, I was thinking about this from the content problem. I really thought that AI search visibility was just a content issue. And that kind of felt obvious, right? AI engines read your content, better content means more citations, make the content clearer, make it easier to extract, get cited, done. And that's not wrong exactly, it's just not the whole problem anymore. And it's not even the part that actually blocks most teams. Somewhere around month four of the show, I started noticing something. The brands that were struggling the most, they weren't publishing badly. Some of them are the best publishers on the internet. They have real writers, real editors, real budgets, but they were still invisible. And what a lot of them had in common wasn't a quality content quality issue at all. It was just that nobody owned the thing. PR wasn't talking to content, legal owned the boilerplate, the product marketing team wrote the positioning, and then five other teams kind of accidentally rewrote it. And nobody, like literally nobody, owned how the brain got described on someone else's website. Which, as we're going to get into, is a lot of where this is actually decided. Over the year, here's the through line of it all. And if you turn this episode off in three minutes, I want this to I want you to take this with you. AI search visibility is a cross-functional problem that shows up looking like a content problem. That's why the strategy I'm about to talk through doesn't just start with the content, it starts a step earlier. Okay, so for your AI search visibility strategy or your AEO or GEO strategy, whatever you are calling it, there are really five steps that's describe, structure, refresh, corroborate, and measure. If you've heard me talk about the FSA framework before that's fresh a structured authority, three of those are FSA. It sits right in the middle of all of them, but there's a step in the front and a step in the back, and if you skip either one, FSA doesn't do much for you. Let's start with that first step. The first step is the one that everyone kind of skips. I admittedly also skipped it, but it's the one that I would put first now. So before an AI engine can cite you, it has to know what you are. And it builds that understanding by looking at how you're described. Not just on your website, but absolutely everywhere. And then it checks whether those descriptions agree with each other. David Kirkdorfer said something in his episode that I've thought about since he said it. He described what LLMs do as word math. Certain words add up to certain meanings. You can change the words a little, you're still in the same neighborhood. But if you change them enough, then you've jumped categories entirely. And here's the thing: no one on your team like actually decides to, hey, we're moving categories. It just happens with those tiny little word tweaks over time. And it happens quite often. Like you test new hero copy, your sales team writes their own one-liner because the official one feels kind of clunky. Regional offices describe the company as slightly different because the corporate version doesn't quite fit the actual market that they're sitting in. And then every single person is making a totally reasonable local decision, but the model doesn't see one brand described five ways, it sees five weak, half-corroborated entities. And that is a big difference when you put them all together. Tommy Landry made the same point from the local SEO side. He works with businesses that already had a name, address, phone number, consistency drilled into them for years. And he said the principle pretty much just carries right on over. Same name, same category language, same description, absolutely everywhere. And he was super blunt about it. If you call a service one thing, call it that thing everywhere. Don't flip between names unless they are two genuinely different services. And I tested this on myself, which is really the only reason I believe in it as strongly as I do. When I changed my descriptor when I first started the show from content writer to fractional content strategist, and then started using that same phrasing, which by the way, I've changed it again, so we're back to testing. But when I started using that same phrasing and not just similar phrasing, the same words on my website, on LinkedIn, on Reddit, on YouTube, in the podcast intro and outro, in the show notes, those engines picked it up noticeably faster than when I was just describing myself in slightly different ways. So the fix, I call this a strategic source of truth. It's one document and it answers what we are we, what category are we in, who is it for, what do we do, and in what exact words? Then every single team pulls from it. That's your PR, your content, demand, sales enablement, social support. Everyone uses the same sentences and no one changes it. David's version of this is that product marketing is the hub and everyone else is a spoke. And that message travels in whatever container each team built, whether that's your press release, your landing page, ads, decks, whatever. It's a different container but the same contents inside. If you want to go deeper on all of that with David, go listen to his episode. It's Is Your AI Visibility Problem Actually a Messaging Problem? And then also Tommy's episode, how does local SEO translate to AI search visibility? Both of those are in the show notes. Okay, so moving on to step two, and now we're getting into the content and it's where the FSA framework really starts. Now, quick refresher if you are absolutely new here. FSA stands for freshness, structure, and authority. I came up with this after running a lot of controlled prompt tests across Chat GPT, Perplexity, Gemini, and Google AI overviews. And what I kept noticing was the content getting cited wasn't really what I expected. It was not the highest domain authorities like you would expect, and it wasn't the best ranked pages, but it was content that was recently updated, cleanly structured, and then corroborated across multiple places. And once I saw that pattern repeat across all the models, it kind of became predictable. That's the whole point of the framework. It takes what feels random and then makes it into something that you can actually work with. So structure. That's the most immediately fixable of the three, which is why I'm putting it first today. But here's the thing about these AI engines. They don't just reproduce your blog post, they pull a chunk out of it. So the question stops being, well, is this a good article? And it now becomes, can the machine lift a clean, correct, self-contained answer out of this blog post without having to interpret anything else? And that's really a different question. That means answer first and then give the context. So that means using those clear h2s and h3s that read like actual questions that someone would ask. It also means that you're stating your definitions in plain English. You're using short paragraphs, lists where it makes sense, and nothing is buried into the fourth paragraph of a narrative intro that you've spent 45 minutes writing. Now, there is a shortcut that I use on all of this, and I mentioned this in the original FSA episode. Write the answer before you write the intro. Just write the actual answer to the prompt first. The intro has to exist, it can exist after that. There are two things from guests that really put this into focus for me. Shane Tupper made the point that buyers don't prompt the way that they search on Google. Google is three to seven words, and an LLM prompt is often 30 words with the person's whole situation loaded into it. They might include their team size, their current tech stack, their budget constraint, their use case. They might include all of that. So the content that actually gets pulled is the content that answers a contextualized question, not a keyword. That means comparison language, trade-offs, explicit this is for you and this is not for you if. Roseanne worked inside of a large organization where content went through four or five approval layers before it could go live. She told me that she streamlined that process and she is really, really proud of the work that she did. But even with streamlining it, getting a page from brief to published took a considerable amount of time. So if we do the math on that, if it takes you six weeks to get something out the door, six weeks is your ceiling. That's not the goal, that's the actual ceiling. Every approval layer is a tax on a signal that you're already literally being scored on. Now, here's why this is an organization problem and not a content problem. No one in the approval chain thinks of themselves as a person hurting AI visibility. Legal thinks that they're just managing risk and they are. Brand thinks that they're protecting consistency, and like I said in that episode, they are. Each layer in this approval process on its own is defensible, but when you add it all together, that becomes a problem. So this is what I mean when I say that AI search visibility is a whole organization problem. It's not that your content team is doing bad work, it's that the operating system around them was built for a world where being three weeks late didn't really cost you anything. But now it does. So there's a practical version of what to do here. You do not need to publish more, you just need to demonstrate that the site is alive. Update your timestamps that reflect real updates. Revise your explanations to reflect current information. Use new examples, new data, maybe add a what changed or what's next section when the substance is really hasn't moved any. And because I'm also constantly tweaking my own strategy, my one rule that I'm implementing in the near future is to just make one update per day. Just across the entire site, just one update. If you are sitting inside of a big organization listening to this, the highest leverage thing that you can do this quarter might not be writing anything at all. It might just be getting one page type pre-approved so that it can ship in the same week. That alone changes your freshest ceiling. If you want to go deeper on all of this, go listen to how approval layers slow down AI Search Visibility with Roseanne Mullet. Alright, so step four is authority. Now I want to be clear about this one because it's also the single most common misread that I get when I talk to people about this. Authority in AI Search is not your domain authority. It's really just not. A domain rating of 90 does not get you cited. I have watched sites with a DA of four get pulled into AI answers over industry giants. I ran a whole experiment on this. It happens. What authority actually means here is the corroboration part. Does the rest of the internet confirm what you say about yourself? Jonathan Bins framed this better than anyone on the show. He said teams treat AI search like a content production problem and it's actually a validation problem. He laid out three disciplines that have to work together. That's your SEO, digital PR, and reputation management. Not just one of them, but all three working together. And he made this distinction that I think is super important. There's a difference between being an entity and being a cited entity and then being a recommended entity. Those are three different things. And the gap between the second one and the third one is where the most brands are stuck right now. So here's where corroboration actually gets built. This comes from a full year of guests, all talking about really the same thing. And these are the three things that keep coming up over and over. Third-party mentions, whether they are linked or not, really help. This is the biggest mental shift from a traditional SEO. Unlinked mentions now do work that the links used to do. The model doesn't need a backlink to notice that six unconnected sources describe you in the same way. That's corroboration. Then we have those review platforms. Jonathan made the point. He said that engines love sources where they can get a feel of the opinion pretty efficiently. So in SAS, that's G2, Captera, Trust Radius, whatever the equivalent is in your category is the thing to go look at. Your reviews aren't just a sales asset anymore, they're really trading data, really trading input. Then we have community services. Carl Peterson's episode is the one to go listen to here. He is seeing community content get picked up within 24 hours in some cases, which is way faster than what I had been seeing in my own test. But he was also equally clear about the failure mode, and I think this is really important. Communities, they sniff out those marketers instantly. Well, you hear this about Reddit constantly, accounts get deleted, posts get removed, you get shadow banned. And the fastest way to lose on those platforms is to show up and just drop links without really any context or any help provided at all. The trick here is to speak the platform's language first, be helpful, be a person, or just don't go at all if you're gonna do the opposite. He was really firm on that one. Next we have original research. Paul Rowe, again, back to Paul Rowe, called this knowledge graph enrichment, and his method is worth stealing. He finds a gap, he literally asks the engines whether the thing he's about to publish already exists, and if it doesn't, he's contributing new information. And that's the kind of thing that the engines come back for. He does live screen recorded prompt tests, he publishes the video plus the transcript plus the write-up, and it's multi-modal. It's original, it's timestamped, and all of it points back to a primary source on this website. That last part really matters here. He was really empathetic that whatever you publish off-site needs to link back to the canonical page that you own. Don't strain that mention somewhere. And then finally we have local and trade press. This is Basha Coleman's underrated pick, and I agree with her. Smaller outlets, local news, niche trade publications, those are the things to target. It's usually cheaper than advertising and maybe even better regarded by the engines. She had a caveat though, I think it's a good one to call out. Lead with data or research, not with really a personal announcement. No one really cares to cover your new VP of whatever. I mean some do, but somebody might actually cover your data and then share that across the internet. So if you want to go deeper on all of that, we have Jonathan Bent's episode on the validation problem, Carl Peterson on mention tracking, Paul Rowe on Knowledge Graph Enrichment, and SEO and PR are finally married with Basha Coleman. And again, all of those will be in the show notes. Okay, last step, step five. This is the one that is kind of what's keeping teams from getting budget for the first four steps. There are two distinctions that I want to call out that took me a while to get right, and I say that openly because early on I was conflating the two together. One is those citations are not mentioned. The brand can be mentioned in an AI answer without a single piece of its own content being cited as a source. That is a completely different situation with a completely different fix. And if you're just tracking your citations and then reporting it as the other, well you're gonna make bad calls when it comes to how to spend your budget or what to do with your strategy. Citation analysis is a diagnostic infrastructure. It tells you where authority is concentrated and where it isn't. So think of it as the map, not the actual destination. And the second one is that top of funnel measurement has kind of broken down and it's really kind of structural. Bash's advice was just go bottom of the funnel and stay there. Think about your leads, your signups, your pipeline, anything that's gonna increase your revenue. Not really so much recessions. And the number that made this really concrete came from Vlad. He told me roughly a quarter of his agency's leads were coming from Chat GPT on something like a hundred clicks. A hundred clicks under the old dashboard that traffic is a rounding error. You'd never really look at it twice. But it was driving 25% of your leads. That's really kind of wild when you think about it. So, where do you actually look for this information? Well, being Webmaster Tools has an AI performance report now. It gives you citation counts, page level citation data, and grounding queries. And those grounding queries are the phrases the model used when it went and retrieved your content. That is really the most useful metric in the whole dashboard because it tells you how the model has categorized you. Now, if you pair that with Google Search Console, use that to tell you what to create next. But Bing tells you how you're being understood. Those are two different jobs with these tools here. And then my favorite thing is to just go run manual prompt test. It's the same prompt, same day of the month, logged data, logged out data, just write it all down. It's kind of boring, it's not really glamorous, but it's really the only longitudinal data that you'll actually own. Now, there's one caution here. I am hedging on purpose. Some tooling in this space just measures cold logged out API calls. So real users are logged in, they have all of that personalization on, they have that search history, and that's a difference between what the API brings back and then what those users are actually seeing. We have some information on this coming out pretty soon. So really just treat your tool output as a signal, but don't treat it as the ground truth just yet. These tools are still working these things out. If you want to go deeper on all of this, the episode to listen to is what do Bing's AI performance and chat GPT ads mean for search. Stop with the news update, but it has all the information about this new Bing report in it. Go listen to what is an AI visibility audit. And finally, the AI competitive intelligence episode with Vlad. Okay, so everything that I just described is really just a quarter of the work. And I love to leave you with something that you can do today to get a like a kickstart on your strategy. Every episode has something like that. So here it is for today. First, open five surfaces where your brand is described. Look at your homepage hero, your LinkedIn company page, your latest press release boilerplate, a directory or review profile, and then go look for a page written by somebody outside of marketing. Like a job posting is really good for this. Put all five descriptions next to each other in one document. The second thing you're gonna do is look at whether you're one entity or five. You'll know within a few seconds of just looking at this. Like you might actually already probably know this, but it's always good to have it documented. The next thing is like pick the correct one, write it down, put it into your strategic source of truth document that's gonna be version one. This doesn't need to be a whole project, it's just one paragraph. The one that's closest to what you actually do, and then everyone's gonna reference that thing. Now, four, go open chat DPT, perplexity, and Google AI mode, pick any of the Google one till there's three of them, pick any one. Ask each one the same question. What is your brand and who is it for? Don't ask if you're the best, don't ask for a recommendation, just ask what you are. It's gonna give you an idea of how these models see what your brand is and who you actually serve, and then compare that answer to the paragraph that you just put into your strategic source of truth. The gap between those two things is your roadmap, and it'll tell you which of the five steps from the previous conversation, the five steps we just covered. It'll tell you what you which one that you actually need to do first. If the engines described you as something that you're not, then you have a step one problem. That's your description. If they describe you correctly but they don't cite you, you've got a step two or three problem, the structure and freshness. And if they just don't know you at all, that's step four, and that's the longest one. That's the authority building, but at least now you know where to actually start. Okay, last thing that I want to leave you with. If you're brand new to all of this and you're still looking for episodes to kickstart your AEO GEO strategy, start with what's the difference between SEO, AEO, and GEO. Then listen to how do AI engines decide what to cite, the FSA framework explained, and then listen to should you skip SEO and GEO and go straight to AI search? That's the 2026 update, not the original one. The original one is fine, but the 2026 update is just it's more current. If you're trying to measure this, listen to what is an AI visibility audit, then go listen to how should brands measure visibility in AI search, and then go look for that being AI performance episode. If you need to sell all of this internally, and this might be the most important bucket, go listen to David Kirkdorfer's episode on messaging, listen to Roseanne Mullet's episode on approval layers, and then go listen to you can't SEO your way into AI Search Visibility. Those three together are the argument that you need to take to your leadership team to get approval for all of this. And then if you want the argument, it's not their consensus, go listen to SEO Agencies Have Two Years Left with GLAD. Go listen to good SEO is good geo, but is that true? And then listen to the three-part AJ Gergage series on dashboards and agents. I'll put all this in the show notes grouped exactly like that so you don't have to write anything down and you don't have to go looking for it. Okay, that feels like a good recap of where to start with your strategy. That's really 80-some episodes condensed into one. And honestly, the thing that has held up best is the least technical thing on the list. Be described the same way everywhere. Be structured so a machine can lift you, stay current, and get other people to say the same things about you too. Everything else is really just implementation. Okay, thanks for being here. Really, thank you to the people who have messaged me about this stuff, the people who said, Hey, I'll listen to your episode, the ones that have left a review, the ones who have shared the episode with your team, the ones who have messaged me to say, hey, I want to be on your show. You are the reason, all of you collectively, that this show has made it to a year's worth of research without me running out of things to be curious about. And I'm really actually very glad that you've been here. If you want to know where you actually stand before you start changing things, an AI Search Visibility Audit is where I'd start. You can find that at CassieClarkmarketing.com or in the show notes. Year two starts on Thursday. Until then, stay visible.