GEO explained: how to monitor and manage what AI says about your business

Warm-up and intros

Hello, and welcome to The SEO Party. I'm Timmy, we have Naveed producing, and joining us is Alex. Alex, how's it going?

Good. How are you guys doing? Happy New Year, and ready for some SEO action.

Oh yeah, absolutely. I think it's going to be a good one. A little on the chilly side, and I'm not used to it, but I'll power through. So, Timmy, what are we talking about today?

Artificial intelligence. AI. You might have heard of it.

Please do tell.

It's becoming a part of everybody's lives. We're finding it creeping in, and it's become a buzzword. People are using it to study, to make their lives a bit more interesting, maybe to write weird stories, even to make some money. And it creates a lot of questions that need answering. Joining us today to help answer them, we have Umer and Salman. Welcome to the podcast.

Hey guys, great to be here. Thanks for having us.

Glad to have you. First and foremost, who are you and what do you do? Salman, how about you start?

I have about 15 to 16 years of experience in the SEO world. I've been running an agency here in Dubai for about 12 years, and we've worked with all kinds of clients across the globe. It's been quite a roller coaster, and it looks like the industry is up for some changes with AI, which we're all excited to talk about today. In summary: SEO agency owner for more than a decade.

To add to that, Umer and Salman are from a company called SearchEdge. Umer, what do you do there?

Like Salman said, we were tracking the industry and were a bit ahead of the curve, especially when we saw voice-based search coming up and asked how answers would be rendered and what that meant for SEO agencies and brands. We started with a few pilots and a few clients, and I've been consulting on some of those projects. What we've observed is very different from the world of SEO. It's a genuinely new customer dimension in terms of how customers ask and how answers are given.

I think agencies and businesses still have a big misconception that AI search is SEO 2.0, which it really isn't. The success metrics you look for and how customers engage are so different that it requires a different strategy. It's been a fun roller coaster so far, and the market is moving at breakneck speed. This is also going to be the foundation for agentic commerce, so you can imagine how important it is. CMOs at the clients we consult are telling us there's a significant drop in organic SEO traffic, and they want to know where that ghost traffic is going. I guess that's what we're here to talk about.

GEO 101: from blue links to recommendations

I'm interested in knowing more about GEO. What is it? Could you explain it for us?

Absolutely. The wrong way to think about it is as an extension of SEO. GEO stands for generative engine optimization, and the word "generative" tells you what it really is. In SEO, even with a long-tail keyword, Google gives you ten blue links and you go and do your own research. Generative engines generate the answer. They have their own opinion.

So as a brand or a company, the questions become: when a customer asks for a product or service you offer, how are you being mentioned? Are you mentioned favorably? Are you losing to your competitors? That's the new game. It's about the content that's out there. How do ChatGPT or Perplexity put together an answer? They take in all the relevant content, synthesize it and reason it out against the intent in your prompt. That's where "generative" comes in. "Engine optimization" means optimizing that content so the answers come out favorably for your brand, your service or your product. That's the essence of GEO.

To piggyback on what Umer said: in the old days of SEO, the search engine results pages were where intent lived. A customer searched for something, your blue link showed up, they clicked and maybe bought. With GEO, what we're seeing is that the LLMs are where demand actually lives, not just intent. We can see entire user journeys go from the very early awareness stage right down to being ready to purchase. That's where your demand is housed now: on ChatGPT and Perplexity. And that's where the smarter companies are moving.

That's interesting. From your understanding of GEO, what's the real shift people are missing? Umer said people confuse it with SEO and how search is done. If you had to put it briefly, what's the missing link?

Naveed, the best analogy I often give is this. Think of SEO as a billboard you drive past, hoping it catches your attention. GEO is being a consultant sitting in the back seat, influencing the direction. As a brand, think about influencing the answer so that your brand, product or service gets recommended. Think influence and getting recommended, not visibility. That's the most foundational difference in thinking. Once you grasp that, the prompting framework and everything else comes second. First, think about intent and influencing it, rather than just visibility.

Umer, if you don't mind, I'll add a few words. With GEO, we're talking about a very nuanced response from the LLM. It doesn't give you a one-dimensional results page where you click a blue link. It gives you a nuanced answer tailored to your prompt and to who you are as a person. That adds more dimensions to the equation, and the biggest one is sentiment. If you're suggested as the best option but with a slightly negative sentiment attached, that's about as good as not showing up at number one in the first place.

So it's important not only to think about rankings, but to think holistically: what are the LLMs saying about our brand, what's the sentiment, what's the emotion behind it, and where do we stack up against competitors? That's a multi-dimensional approach compared with ten blue links, which is pretty one-dimensional if you ask me. That's why it isn't an incremental step from SEO to GEO. It's orders of magnitude different in how LLMs work, how they suggest responses and how they refer companies. And you'll notice that traffic or leads coming from ChatGPT and similar tools tend to convert better, because people treat it as a referral, almost as good as a recommendation from a friend. That's exactly why the way LLMs portray your business, and the sentiment around it, matters so much. It's an entirely different ball game, not just an evolution from SEO to GEO.

Build for humans and machines

So it's not about rankings anymore?

Exactly. It's about the recommendation, winning the answer. And to drive home what KPI to measure, Naveed, let's take an example. Say you're a SaaS company, a CRM. If you're getting brand mentions, but in the context of "it's a good beginner-level CRM" or "a generic CRM", that's bad for you if you're in a specific niche. You don't want to be branded as a beginner-level CRM. In SEO we worked with nouns, like "CRM". Now we're working with adjectives, which qualify the intent and how we want the brand placed in the market. What do we want customers to think? What are the customer decision points, and at those moments, what is AI telling customers? First, you need a finger on the pulse of that. Then you need to measure it, and then influence it through your content strategy. We can get into that later, but that's how I'd describe it.

I have a question about that. For years we've structured sites around architecture: knowing exactly which keyword to focus on, then building topical authority. We also have Google saying don't follow the suggestions people make, like putting a short description at the top and bullet points about this and that. But honestly, from a user experience perspective, and we build a lot of pages, when you break content up like that, people stay longer and find what they need faster. So it works for users and for AI. How do we need to start now? We used to start with the architecture and put in the keyword; now we have a topic. Is that technique going to go away, or do we still need a foundation for structuring pages and a website around topical authority?

Great question, and I know Salman will have his two cents as well. First, we definitely need to think of our marketing and our website as being for two audiences: one human, one machine. Absolutely. Already, there's an AI layer in the middle acting on our behalf. If it doesn't know about a capability, it's as good as you not having that USP, one your existing customers know you have but your prospective customers don't.

Previously, the way we structured a website, say a typical portfolio or catalog website, was: these are our lines of business, these are our services, and any human can navigate that. The ideal for LLMs is to make it intent-driven. Whatever the customer's intent, can the AI navigate your website and work out the answer? Whether you're e-commerce or a service business, you want the context of what the product is, its use case, its benefit and its USP, in a relevant, contextual, long-form format. If you just list four categories and how you've split your business, AI doesn't care. It needs to know what, why, how, and what's different, because it's recommending on behalf of the customer who asked. You need to give it every reason to choose you over your competitor, so the website has to be organized that way.

Some tools have come up, such as llms.txt, which is a kind of road map that tells LLMs: this section of my website is about this; for customer service questions, go to this page. That helps LLMs find the right pieces to answer the customer's question. Frankly, it's an evolving field. Some people advocate creating a knowledge layer, which is more for enterprise-level websites, but for small and medium catalog and portfolio websites, you can get away with an llms.txt file that gives that context, a bit like the robots.txt file we already have. The one thing I'd finish with is: have contextualized information, whether you're a window cleaning company or anything else, for the LLMs to pick up.

To piggyback on Umer's comments, let me take a step back. GEO is forcing businesses to move closer to the customer and understand them better than we did in SEO. In the SEO days, we just cared about the keywords customers searched for. The keyword tool gave us the highest-volume keywords, we optimized for them, and that was it. GEO is forcing businesses to put their ear to the customer's heart, so to speak, and figure out the conversations, pain points, desires and fears their customers have. Once you understand that, you can design a prompt framework aligned with your business, because you now understand your customer much better, and then optimize for the conversations happening on the LLMs. There's a lot of talk about bullet points and tables, and it's all true, but the objective of all of it is to make sure your content embeds itself in the conversations happening online. The better you do that, the more you'll succeed with GEO.

I want to add something from experience working on multiple websites. Even before AI, if you wanted a site to rank and have strong topical authority, you kind of had to do this: answer your audience and write for the correct user intent. So for people already doing that, not much changes; maybe you expand the content. Give me an example. Say I have a website that sells web services. Do I answer this intent within the existing content, or create support content like blogs or subpages to expand on it? I want to give a clearer picture to people who already have existing sites.

If you don't mind, I'll take that first. With SEO, we look at keywords and traffic and optimize the user experience and content to hit the customer's pain points and lead them to a call to action. With GEO, we're tapping into conversations. Say you have a blog post with an H1, H2, H3 structure and a thousand words of content. Everything is great. What we've noticed is that if we distill that post into a two-to-three-line summary that answers the question right away (without making the user read a thousand words, you're saying "if you're short on time, here's the gist"), followed by the full post, we see a big increase in visibility on LLMs from that one change, in our experiments and testing with clients.

The other thing I see a lot of companies do wrong is that they look at ChatGPT or GEO and go on a content creation spree, flooding their sites with dozens of blog posts, service pages and area pages. That doesn't work, because those pages aren't aligned with the conversations. You need to do the thinking and research to understand what each page is supposed to solve, and which conversation it should embed itself in and be cited for. Asking that question helps us develop useful content rather than pushing out blog posts, service pages and area pages at random. Umer, please go ahead.

Beautifully put. I can show you a quick example. Naveed, am I good to share my screen?

Yes, go ahead.

Prompt portfolios and citation maps

I'll make that point concrete. When Salman talks about being customer-centric, having your ear to the customer's heart, we mean going really granular. First, know your customer types, because not all customers are the same for any business. In the awareness stage, what questions are they asking where you want to appear, and if you're not appearing, you're losing? You draw up your prompt sets for that. Then in the consideration stage: pricing, packages, what you offer and don't offer versus your competitor. There's a prompt set for that. Then the decision stage. And you do this for each of your key personas.

Then you run those prompts. Essentially, I'm talking about a prompt portfolio that covers, and this is a key point, the semantic neighborhood of your customers' buying decisions: the decision points and stages for your brand. Understand those questions, what AI is telling customers about them, and how you're being framed. Is it just a mention, or the right mention? Like the example I gave: "beginner CRM" is bad for me; I need to be specific to my industry. You run the prompts on the top LLMs, ChatGPT and Perplexity, and then you analyze the citation map.

Remember, an LLM doesn't have a mind of its own. It builds its answer from the content available out there. So the key point is: where does it go to build that answer, and are your sources, including your website, being cited or not? Connecting back to Salman's point, that should inform what content, and what chunks of content, you have on your website. Remember, you'll only have a 20 to 30% direct impact on citations. The majority will be third-party, and that informs your third-party content strategy.

Say you're falling behind on visibility for one aspect of your business: customers think the product is too pricey, or whatever it may be. You have to act through the levers you have, and the way to uncover those levers is the citation sources that ChatGPT, Perplexity or Google's AI cited to produce the answer you didn't like. Or maybe you like the answer and want to keep being cited. That's what's under the hood.

At the heart of it is whether you can build a prompt set that truly reflects your customer's journey and decision points. We were talking about cruises. For one person, the key decision point is whether there's a daycare, so they can hand off the kids and enjoy time with their partner. For someone else, it's a particular food option. When you're at the heart of those questions, how is AI answering them, and how is your business showing up? That's what you want a handle on, and that's the core of the strategy we use. Sorry for the long answer, but I thought an example would show what's really under the hood.

That's great. Let me clarify what I saw. You took a resort and broke it down by what they have, like the different water parks, and identified the search queries people might put into LLMs, whether informational, navigational or transactional. So if we have these questions from our clients' data, do we add them to existing pages, or create additional content like blogs? What's the easy strategy? If I run a business and have about 200 questions people ask, do I add them to my existing content, or build long-form content for LLMs so I get found more?

The questions I was showing are just prompt sets. Their purpose is to uncover the sources and citations AI is using to answer. If I see I'm not doing well in an answer, then I get to the stage you're talking about: an FAQ section that addresses it for my brand in a favorable way, provided I can deliver on it. So those questions don't go on your website. They go into an AI visibility tracking platform. Semrush offers one, Profound is another, and there are plenty of tools out there. In the old days you tracked your keywords; now you need to track prompts. How am I doing on these questions? They're at the heart of my market share, and not just for what you offer, but for where you want to be. If you want to beat a bigger competitor, which areas are they winning, and how do you establish in the customer's mind that you have better goods, better delivery or a better product?

So the questions go into the AI visibility platform, and where you're not doing well, you build that narrative into long-form content or an FAQ on your website. That still won't be enough. You also need content outreach and link building to the right sources that give you relevance, authority and credibility; all three have to be in sync. That goes back to Salman's point: PR matters too. You may be the cheapest, or have the most relevant solution, but if there's negative brand sentiment, ChatGPT may drop you from consideration. It all connects together.

From prompts to pages (and PR)

So from what I gather, it's not just throwing a bunch of prompts into a tool and seeing what happens, which is what I see people doing. There's a lot more strategy to factor in. Alex, sorry, you were saying something?

When I'm on this podcast, I like to give people an idea of how to structure things. How does someone come up with these terms? How do you know what people are searching? You mentioned Profound and Semrush. Do you start with the prompts, see whether you're included, and then work out a strategy to get included? How do you do it?

I want to drive one point home, because if I can, it will clear up a lot of questions. In the standard SEO world, you'd have a page of content with keywords, H1s, H2s and an FAQ section. The purpose of that FAQ section was often to stuff keywords into the content. If a dentist's page asked "How much does a dental implant cost?", we'd write a one-paragraph answer that really said "get in touch and we'll let you know". There was no actual answer. LLMs are forcing businesses to actually answer the questions people ask: a dental implant will cost you between $500 and $1,000. That's a direct answer.

When we look at most businesses' websites, they already have a decent amount of content. In most cases it just needs enrichment so it becomes parsable by LLMs. In some cases we do want to create more pages or more citations. But all of that comes from the strategy, and the prompts are the second step. Step one is sitting down with the business owner and asking: what do you care about in your business? Which aspects are critical to you? What kind of sentiment matters? Are there competitors you want to be stacked up against? Back in the day, we'd ask clients for the ten keywords they wanted to rank for and then spin off a hundred more. Today we don't ask what keywords they want to rank for. We ask what they care about in their business and industry, and then we develop the prompts as phase two. Without phase one, the prompts won't work and won't uncover actionable insights that move the needle. Sorry, please go ahead.

No, that's perfect. Beautifully put. Those prompts are there to uncover the customer's decision moments. If you win those answers, you win market share against your competitors. And you can't get to the heart of it without the prompt set: this is their top product, this is their lowest seller that they want to push, this has been their core offer and it's declining, so let's check the sentiment on it. Maybe the key USP of their main product isn't being mentioned in AI answers. You uncover all of that, and you can only do it well if you truly understand the customer personas, their stages and their points of comparison against competitors, then see which citation sources influence the answers and build your strategy. So the prompting step is important, but the step before it, truly understanding the business, the customer, the sector, the use cases and the competition, is even more important.

Alex, I'll throw in a nugget most businesses will find very useful. Put a prompt into ChatGPT like "give me the top 10 resorts in this area". If your business isn't number one or isn't cited, follow up by asking: "Tell me why this website wasn't suggested as the number one result." Do that across three, four or five platforms and distill the answers. The LLMs will tell you exactly why: you don't have enough customer reviews, or you're weak in this or that aspect of the business. They'll literally tell you where you're dropping the ball. Take those responses, distill them, and build a content strategy that, for most businesses, takes only a few weeks to implement. It works like a silver bullet.

I want to add something. Before AI, Google gave us featured snippets at the top with pricing and so on. I've seen people do all the research and then put pricing in image graphics instead of an actual HTML table. And many businesses don't want to give out their price at all, which in my opinion is crazy. We work with a local limo company, and their SEO team suggested a page for people arriving in Orlando who don't know the area: your options for getting from the airport to the hotel and moving around. The SEO company did an amazing piece of content, and the client put it aside because they didn't want pricing on the website, even though it was given as ranges. I see the same old-school model with dealerships hiding vehicle prices. People shop around. When I'm searching, I need to know how much I'll pay. If you hide the price, sometimes I won't even call, because it isn't clear and I feel there's a game behind it. Transparent pricing, structured correctly on the page, helps with Google search and with LLMs.

100%. A quick reflection on that, Alex. We consult for many B2B enterprises, often channel-driven businesses: say a distributor that also has an e-commerce website. Many of their customers are bulk or seasonal buyers, and for the right reasons they don't show prices, because it's by quotation. But in agentic commerce, if you don't have a price, you won't be considered in the data set. That puts them in an even trickier situation. Again, it's about how detailed your prompt set is. Take your example: I want a hotel, a good restaurant on the way for the layover, and so on, and this is my budget. How could ChatGPT factor in a business that doesn't mention its price? So "price behind a login" is going to be more and more of an issue for channel-driven businesses.

It's such a fast-evolving field that businesses are adjusting and adopting, and it's not giving anyone much time. I was checking yesterday: if you use ChatGPT's research assistant and ask a small question, it builds a full comparison table covering price, USP, ease of use, technical details and reviews. Nothing is hidden anymore. You have to be on top of everything: user-generated reviews, Wikipedia, what you say on your own website and what third-party sources say. And about 80% of your citations will come from third parties. So link building just got a whole lot more important, because nobody cares what you say about your own business; it's what others say that gets factored in. But before link building, you need to understand your strategy: what sentiment you're trying to change, and which qualifying criteria or USP you want considered in the answers. That's what the prompt set is for.

The ads question and the window to move

I want to ask you something. Naveed mentioned my trip. My wife plans all our trips, and with AI she planned our last one. We found a resort we wouldn't have found in Google results or on the sites that promote the biggest brands, plus the right flights and times. We had one of the best trips of my life for the best cost, and found places that are more for locals and were amazing. The experience is great. My question is: at some point advertising will have to come in. Right now the AI cares about giving us the best result so we keep using the platform, like Google did at first. What happens when someone can pay $10 to appear in the list? It feels like a time bomb before that experience gets ruined.

Good question. I'd call this the pre-monetization stage. Monetization will kick in; it's just a matter of when, and the model is likely to be different. With Google Ads or Shopping ads, the ad was matched to the relevance of the keyword. But if an organic recommendation contains an ad, that's a conflict of interest. Would people trust ChatGPT if its answer wasn't based on your intent, the right resort, restaurant or vacation package, but on the fact that a hotel paid a certain amount of money? That defeats the whole purpose. So it won't be the same; maybe it'll be something like in-app ads. If they start trading relevance to your prompt for payment, that's a downward spiral, and from what I see thought leaders saying, it will take a different path, because otherwise it defeats the purpose.

100%. What I'm afraid of is that they start pushing, within the organic listings, businesses that have ad budgets. When you don't understand how they do things, they could prioritize those businesses in the algorithm without us knowing, while we still believe it's organic. It's going to be a difficult game for a while.

Alex, that raises probably the most important point, and something we tell clients: get started. We're not at full agentic commerce yet, and we don't know how some of it will pan out, but the earlier you get a handle on how AI views your business, the better. For an e-commerce business, that may mean fixing your shopping feed. For a portfolio or services business, it may be whether your USPs are coming through and how you're doing against competitors. Get started. If you're in the game, you can pivot. We don't know what the monetization phase will look like, but companies that are ahead in their AI visibility and brand sentiment across customer decision moments will be able to pivot and grab the opportunity. Even in agentic commerce, the transaction is just the bottom of the funnel. At the top, AI is still analyzing whether this is the right business to recommend and whether it meets the criteria the customer asked for. Once you meet those, whether the transaction happens organically or through monetization is just the mechanism. So my general advice to businesses is: get started. It won't be time lost, and you'll be able to pivot faster once it's monetized. Salman, let me know if you want to add to that.

That's a great point, and I'll add another dimension by rolling the clock back about ten years. When we started running Google Ads and Meta ads back then, our ROI was exceptional: low cost per lead, good-quality leads, good-quality traffic. Now the market has become very messy. Google is gaming the system too much, there's a lot of AI and spam, and the ROI on campaigns isn't as good. So even when LLMs get monetized, you'd still have about a year or two before it gets really ugly the way Google did. That's an exceptional opportunity for businesses, even after monetization, to jump in and build something of value.

Look at Meta. Back in the day there were thousands of targeting options. If you wanted to target people who love golf, you could drill down to the type of clubs they use, the shoes they wear and the clothes they buy. Over time, they've made it more and more general, because they make more money showing your ads to a wider audience. That will happen with AI as well, but there will be a honeymoon period during monetization, and businesses that ride that honeymoon period will have an advantage for the next ten years.

Fundamentals, made urgent

The one negative I'm seeing with many businesses, including mine and others I'm involved in, is that cost per acquisition has roughly quadrupled over the past couple of years. As a business, you can't simply quadruple the price of your service, so a lot of businesses absorb it. It's becoming insane. And as you said, even when you target the right keywords on Google, there's a lot of spam. It's hard to be a business entering this world now unless you think ahead about AI, so you can benefit from it before everybody else does.

One thing I'll add, Alex: the fundamentals don't change. Understand your customer. Your digital marketing manager can come up with keywords or prompts, but go and talk to your salespeople. They'll tell you where sales are lost. Nobody will give you better insight into what customers are thinking, what their alternatives are and where you're losing than your salespeople and account managers. Speak to them, understand those things, and reflect them in your prompt set.

This is something Salman emphasizes: churn in GEO is going to be invisible. In SEO, you could see you were ranking on page 10, clear as day, and that's why you didn't get the traffic. In GEO, you'll lose without knowing you're losing. It happens behind the session; you're just not winning the answers. That's why it's important to track the answers to those questions and how your business is doing on brand mentions and winning the answer in the first place. Otherwise, the churn is invisible.

That's two of the things we do for clients, and this isn't a plug: first, build a prompt set that reflects their performance, customer sentiment and decision points. Then, for optimization, we look at the website from a few angles. There's readability, or machine parsability, but even that comes second to answerability. Do they have relevant content that answers the question? Is it in a format, schema and header structure that makes it quick to skim? And is there an intent signal, connecting the context of the product or service to the customer's intent? This is where you link the product to its use case and to the customer's what and why within your content. Once you have those three things ready, that's still only 20% of the game. The other 80% is the third-party sources you need in your ecosystem.

Make GEO measurable

Let's talk about opportunity, because this is an amazing one, and I'll connect it to SEO with a specific example. SEO was a black box. All we had was a set of best practices: this is a good link building strategy, these are the keywords. What went on inside the algorithm, you never knew. You followed the strategy to the letter and hoped for results. That's not the case with GEO. Like Salman said, whatever answer the AI gives, you can ask it which sources it cited, and where exactly the negative sentiment that lost you the answer is coming from. Then you have a game plan. The information is all out there, which we never had in SEO. These are the under-the-hood technical things a business needs to understand, and if it does this well, it can dominate. A business might not actually be as good as it seems, but the AI will think it is if it does a great job on GEO. I know that edges toward a grey practice, which we wouldn't recommend. You need to be honest, and you need a presence in user-generated content and reviews, so you're approaching it holistically.

Alex, one quick example to drive the point home. I've always believed keywords were one-dimensional: you plug in a keyword and get semantically related keywords back. We recently had a client that installs double-glazed windows and doors. Through a marathon prompting exercise with that client, trying to figure out who their customers are, we realized that most of their customers looking for double-glazed windows live under flight paths. If your house is under a flight path, you'll be looking for double-glazed windows. A keyword analysis would never uncover that. It's a pain point we only found by constantly asking the right questions and working out the customer's pain points. We created content around flight paths and how double-glazed windows can solve the noise problem, and people were having mental health issues because of the noise. It took about six to eight weeks to start generating very high-quality LLM traffic from those prompts, because people were actually typing into LLMs, "I live under a flight path, what should I do?", without even knowing double-glazed windows were a solution.

That's what I mean: there are so many unknown unknowns that we'd never find in the keyword universe, but now we can figure them out with GEO.

And niching within your industry, because nobody is the same and everybody niches in different things. That's brilliant, actually. I need double-glazed doors for my snoring.

So, we were talking about what businesses care about. What if a business only cares about revenue? How do you make GEO relevant for business owners who are mostly eyeballing revenue?

If I take that question apart, at its heart is: which KPI should they care about that translates into revenue? In GEO, it's winning the answer. Is your brand being recommended? They'll see that in their brand mentions across their prompt portfolio, and in the positive, highly convertible leads they're getting. On the technical side, Google Analytics (GA4) is currently fairly basic at showing what kind of bot traffic comes to your website. There are tools we use for some enterprise clients that can tell you which pages are being crawled to produce which answers, and where you can change things.

But to your question: getting recommended is the most direct correlation you can build to revenue. Don't rely on visibility signals alone. In the CRM example, being mentioned in the wrong context, as a generalist, beginner-level or small-team CRM rather than a specialist tool, leads to customer churn. That's a recipe for disaster. So know the points where you win customers and whether you're winning there. To track it, you need the right prompts and the right KPI, which in this case is citation share: are you able to influence the answer, and are you winning the recommendation? The right tracking tool, and the one we use is Profound, one of the best out there, will tell you that across 200 prompts you won the answer in 40. Many of those 40 will translate into traffic or recommendations. But I wouldn't put too much focus on traffic. You're winning customers through influence. Think influence, not traffic. I know that doesn't directly answer the question, and customers ask it all the time, but this is what I tell them: win the recommendations across all the key customer questions, and you'll see it in your revenue.

Implementation playbook and wrap

That makes sense. Let me ask one more question, because we're about to run out of time. In hindsight, GEO looks like content optimization, if you will. Is that what it is? What are the meat and potatoes? Because when you describe it, it sounds like textbook content optimization, and we're all content marketers, so we can call that out. Can you elaborate?

I'll make it quick and share my screen, because I have a slide I presented to a client on this. Other than informational, commercial and transactional keywords, there's this generative piece. Alex gave the perfect example: the AI has to generate an answer, and this is where content chunks come in. LLMs take chunks of content to build their reasoning. So do you have the right content, with the right relevance and authority signals, to influence the answer in your favor? In a way, yes, it's still about content, but very strategic content. There are three phases: like Salman said, the research, understanding what the customer wants and where you're losing, and that should inform your content strategy. Content strategy itself has become multi-tier. A basic content strategy of "let's get these kinds of high-authority, relevant links" won't work if it isn't rooted in the questions your customers are truly asking and building you up as an authority, the default answer, or winning the sentiment. Otherwise, it's wasted effort, or wasted money if you're paying for fancy publications. So yes, but with the right strategy behind it.

Naveed, if I could chime in quickly. We did content strategy and optimization before LLMs, but what we optimized for was keywords: semantically relevant keywords, short-tail and long-tail. Now we're optimizing for customer pain. When you're clear that you're not just focusing on keywords but trying to solve a specific pain point, your content strategy changes drastically. You're not pumping out blogs targeting a keyword. You're solving a problem your customers actually have, and, as Umer said, thinking through which stage of the buying journey that customer is in and optimizing for that.

That makes sense. Timmy, do you have something?

Yes. As we're wrapping up, if there are any CMOs listening, what's the one thing you'd recommend they start doing in the next 90 days regarding GEO?

If it's a CMO, I'll assume an enterprise-level organization. Three things. First, get started. Start working with prompts and plugging them into the right platform to see where you stand as a brand and a business. But don't have just your digital marketing manager make up the prompts. With one client, to get to the heart of the right prompts, we uploaded all the data coming through their customer service department, the calls and queries, because it reflected how customers approach them, and that informed the prompt set. So go to the heart of understanding your customers and their touch points to build the right framework.

Second, content. Companies still have separate departments: product syndication, PIM, PR, third-party. They sit in silos because each has its own function. Now it's all one session. If you don't have the right content about one product, or your brand has bad PR around a certain consideration point, you lose consideration. Miss one, and you're out. So these siloed content teams need a cross-functional GEO orchestrator to bridge them. It's a content chain: use case, brand, authority, everything comes into play if you want to win the answer. You need to break those silos to win the recommendation.

And get started, because it's such a fast-evolving field. Things will change, but if you're on the journey, you'll have a handle on your key prompts, how you're doing against competitors and where you need to fill gaps, and you'll be all right. For bigger enterprises, it will require some organizational change. The line of business needs to communicate the right technical specs and USPs of each product and combine that with PR. Like Alex said, if the price isn't there, you lose the recommendation. The same goes for specs: customers use LLMs to compare product specs, and if yours are missing, you lose consideration and the recommendation. That connectivity needs to happen in bigger enterprises.

Could you translate that for small to mid-sized businesses? I think the same things apply, but that would help our listeners a lot.

Absolutely. If anything, smaller businesses have an easier job, as long as they think in terms of the awareness, consideration and decision stages. The content they need should be in one place, and not only on their website but in their third-party citation sources. You need to know the ecosystem of third-party sources you can influence. You have indirect influence over user-generated content like Reddit, and depending on your industry, there are relevant publications you should leverage. As I said, link building, the right content and content partners echoing your message have become far more important in GEO than they were in SEO. You can toot your own horn all you want, but how third-party citations view your brand matters even more. There's a lot of work ahead of us, and whoever picks this up fast has a lot of ground and dominance to enjoy in the short run, especially in the pre-monetization phase.

Build and dominate. I think you need a lot of intestinal fortitude to do that, and now is the time, because this year will set the ground for the future.

Wonderful. Umer, Salman, thank you for joining us.

This was a pleasure. Thank you for having us.

Happy to have you. Alex, always a pleasure.

Same here. And that's the end of the episode. Thank you all for watching. Like, share, subscribe, all that good stuff, and we'll see you next time.