My Approach to GEO, and How It’s Different (Or Not) from SEO

I recently wrote an article about Google’s “Optimizing your website for generative AI features on Google Search” document, in which I described what I think about the guidance in relation to my approach to SEO/GEO.

In this article, I’d like to talk more about my GEO philosophy, in general. More specifically, I want to explain why I think GEO differs from SEO, and also why I don’t think that.

If that sounds confusing, it’s because it is (and isn’t).

Let’s start!

How is GEO different from SEO, and how is it not?

A standard GEO practice would be to use a question heading (like the above) and then immediately answer that question, likely using semantic triples (subject – predicate – object). But that’s not how I want to word this section, and since I believe in a people-first approach to SEO/GEO, let’s start with some background information about my philosophy, as I think that’s the best way to structure this article for readers.

I was initially one of those “GEO is just SEO” people — after all, that’s what Google’s stance is, as well — but I’ve since softened my position on that.

In other words, I do believe that GEO involves some, let’s call them areas of focus, that are different from traditional SEO strategies. Now, some SEO strategies have always encapsulated what today are known as GEO practices, so that’s why it’s not always fair to call GEO a separate discipline. But I do draw that distinction because, again, I believe GEO has a few areas of focus that are slightly new to many (not all) SEO strategies.

For starters, GEO has a bigger focus on natural language processing and semantic search. We’re less concerned about keyword targeting (lexical meanings) and more focused on entity clarity in multidimensional spaces (vectors). Again, traditional search has been hybrid for many years, so this emphasis isn’t new to SEO, but I do think GEO strategies need to focus on semantic meaning more than keyword targeting.

GEO strategies also have a bigger focus on earned media than most traditional SEO strategies would. Of course, backlinks have always been a focus in SEO, but now in GEO we’re more concerned with authoritative and relevant mentions from third parties. This means more emphasis on digital PR, subreddits, and similar sources of third-party validation.

I also think that GEO strategies are more multimodal than traditional SEO strategies were. This means more emphasis on video content, especially from YouTube and organic social platforms like TikTok and Instagram, as well as unique imagery or even tools. Again, this isn’t new to SEO, but I do believe the amount of emphasis on non-textual content slightly distinguishes GEO strategies.

Even the types of written content we focus on can be different. For example, AI assistants often cite listicles and comparison posts. While these aren’t new to SEO strategies, again, the heightened emphasis on them is slightly new.

Then we get to the technical hygiene of websites, where making sites agent ready is sort of a new philosophy that distinguishes GEO from SEO. For example, AI agents take screenshots, interpret rendered HTML, and read the accessibility tree (ARIA tags) of a website. Again, not new for SEO, but the emphasis on, say ARIA tags, is somewhat renewed due to GEO strategies.

Also, websites need to load fast (core web vitals and overall page experience) if they’re going to be fetchable and chosen by AI assistants or agents. Again, not new, but we do get factors like 499 error codes, which weren’t really considered in traditional SEO strategies. Say nothing of LLMs.txt files or markdown versions of websites, two tactics of debatable efficacy for GEO strategies, but again, new tactics for SEO.

In summary, GEO has some areas of focus that are slightly different from traditional SEO strategies, so that’s why I see them as separate disciplines (even if some SEO strategies have always incorporated those same areas of focus, which is why it’s not a definitive distinction between SEO and GEO).

How would I do a GEO strategy?

This part is hard to answer, because, like with SEO, a GEO strategy really depends on the brand, its audience, its competitive landscape, and its business goals.

Having said that, if I’m working with a hypothetical brand, this is how I might approach a GEO strategy:

1. Ensure the technical hygiene of the website.
2. Identify target prompts (usually of commercial intent).
3. Find competitor gaps.
4. Create owned and earned media to cover those gaps.
5. Measure the results.

Let’s dig into each of these a bit more.

1. Ensure the technical hygiene of the website.

If you want your pages eligible to appear in AI assistant answers, you shouldn’t block any of the relevant user agents with your robots.txt file. This includes user agents from ChatGPT, Gemini/Google, Perplexity, Bing (Copilot), Claude, and others.

Your pages also should be indexable for search engines, so no no-index meta robots tags.

You also want your site to load quickly, so passing core web vitals, especially on mobile, is ideal.

As for interpretability, you want to mark up your pages with relevant structured data (preferably JSON-LD). Common types include product, review, FAQ, how-to, person, and organization. In my opinion, structured data is most useful for building entity clarity in knowledge graphs, which AI assistants could reference. In other words, it’s a long-term play.

I don’t see a downside to adding an LLMs.txt file to your website, other than the time it takes to create and maintain it. While it’s likely not applicable to AI assistants today, it could be used to help AI agents in the future (or even now).

As for markdown versions of webpages, I haven’t used this tactic myself and question its efficacy, but I could see it being helpful if you have a heavy-loading page that isn’t currently being cited by AI assistants.

You also want your site to be agent ready, and the easiest way to help ensure that is to use semantic HTML and ARIA tags for accessibility. This is a best practice anyway, but you want your site fully accessible to screen readers, as they’re a good proxy for AI agents.

2. Identify target prompts (usually of commercial intent).

Once your website is technically sound for AI assistants to fetch content from, the next step is understanding which prompts you want to target.

This starts with understanding who your audience is. Doing customer calls, analyzing CX tickets, reviews, and Reddit threads, and talking with your sales team are all great ways to build customer personas.

Once you have an idea of who you’re targeting, the next step is to map out their buyer’s journey and identify which questions they’d have along the way, focusing on commercial intent.

Personally, I still use keyword research for this, at least to get started. While the interface may have changed with users moving from traditional search to AI assistants, their search intent is still largely the same. Keywords tell you what your audience is looking for, then you just have to convert them into prompts.

To convert keywords into prompts, start with commercial intent keywords as a base, then find qualifiers. The qualifiers can be who’s searching (ICP), what they evaluate (criteria), limits they have (constraints), or alternatives they’re considering (comparisons).

You also want to pay attention to query fan-out, and understand how the AI assistant is grounding its answer for any given prompt. You can find query fan-outs in tools like Profound, custom tools like Qforia, or you can look at grounding queries in Bing Webmaster Tools.

Once you have your prompts broken down by topic or persona, you can either manually track them or use a tool like Profound. Just remember that AI assistant answers are probabilistic, so just because you appear one time, doesn’t mean that’s guaranteed to happen again. You’re looking for larger trends of visibility.

Remember also that the best prompts align with your business lines. You’re looking to achieve visibility where it matters to your bottom line.

3. Find competitor gaps.

Now that you’re tracking your prompts, you want to pay attention to mentions, citations, and sentiment.

Identify which brands are being mentioned in the answers, which sources are being cited, and how those brands are being described (positive, negative, or neutral).

If you’re not mentioned or cited, take note of who is. What type of content did they create?

Your goal is to understand where the AI assistants are sourcing their information, so that you can supply the content necessary to be part of those conversations.

4. Create owned and earned media to cover those gaps.

Now that you know what content is being cited, your job is to produce better alternatives — what Google calls helpful, reliable, people-first content — to entice AI assistants to choose you as part of the conversation.

This is where strategy comes into play, because you need to not only identify the type of content (the format) that’s desired for that prompt, but you also need to supply it from the type of source that’s favored. Maybe that’s your own website (especially if you see competitors being cited), or maybe it’s from a third-party site, like social media, an industry forum, or a publication.

5. Measure the results.

This is the hardest part of any GEO strategy, because measurement is not a solved problem.

Some of the tools available that I use include Profound, Microsoft Clarity’s AI Visibility tool, and Bing Webmaster Tools AI Performance tabs. I also track AI assistant referrals in GA4 manually, but GA4 recently introduced a channel for AI Assistants, though I haven’t seen it in my GA4 platform yet.

Ultimately, you want to track share of voice, citations, mentions, and sentiment, but your goal is conversions or leads. If you spot an increase in branded traffic to your website, or an increase in qualified traffic and conversions, those are positive indicators.

I’ll also mention log file analysis. If you’re like me, you have a tough time getting your hands on log files, but these can show you which bots (user agents) are hitting which pages. If you’re pages aren’t being visited, that’s an indicator you have work to do to get cited or into training data.

Miscellaneous Long-Form Questions about GEO Strategies

I recently spent my weekend having on-and-off conversations with ChatGPT about GEO strategies. I asked ChatGPT to ask me questions and then I answered them verbally. I thought that content would make a good FAQs section for this blog post. I’ve retained the original questions but adjusted my answers to be more readable for you (and extractable for AI assistants).

Let’s dive into some Q&A!

Can you walk me through how you would develop a strategy for improving an organization’s visibility in generative AI search results?

First it comes down to your business goals and objectives, namely identifying the lines of business that are most important, as well as the personas in your audience that correspond to those lines of business.

Once you have your personas and business goals outlined, which you can get from different sources, such as CX tickets, your CRM, or even reddit threads, you want to map the buyer’s journey for those personas and identify the questions and answers they’d ask, especially ones involving AI assistants.

Next you want to identify the prompts or topics that are relevant to your personas’ buyer’s journeys. Then you can start tracking those prompts/topics in AI assistants (or using a tool like Profound), identifying whether you’re mentioned or cited and what the brand sentiment is like.

You can then compare your visibility against competitors using share of voice or looking at competitor gaps in mentions or citations. That’s followed up by a content strategy for both owned media (the focus of traditional SEO strategies) and earned media (a bigger focus in GEO strategies).

Once you have your content strategy in place, you can measure your progress, again using a tool like Profound, so you can see how your brand’s visibility is evolving with respect to mentions and citations relevant to your target audience.

Ultimately, you want to find ways to automate all of this or have repeatable processes in place. Your goal isn’t to do one-off practices every time but rather have a repeatable approach for analysis, content creation, and measurement.

For the above question, getting more specific, how would you tackle the strategy step-by-step and also measure the results?

I’d use a tool like Profound to track visibility of our primary topics and prompts along the buyer’s journey for our different personas. For example, let’s say we were selling used cars. I might have a topic for used trucks, used minivans, used SUVs, and then have individual prompts within each of those topics. Then I’d create personas and tie those personas to those prompts so that we can filter the data later. That’s one of the hardest parts about tracking AI visibility is filtering your data to find what’s most relevant.

Then I would compare those prompts against the major AI assistants. Rather than looking at visibility scores across multiple AI assistants at once, I’d focus on ChatGPT, Gemini, AI Mode, and AI Overviews, because those are the big four where most visibility and referral traffic come from.

I would look at mentions and citations for individual prompts or topics within different AI assistants because each AI assistant operates differently and is going to cite different sources. One might have a bigger focus on YouTube while another might have a bigger focus on Wikipedia or news sites or whatever it may be.

You need to understand those dynamics based on individual AI assistants you’re targeting for individual prompts.

Then I would understand where our competitors are at, whether we’re being mentioned or they’re being mentioned. And if they’re being mentioned or cited and we’re not, then those present opportunities for us to create content that fills in those gaps, whether that’s owned media or earned media. So next I’d create a content strategy, and probably a digital PR outreach strategy and a social media strategy or a video strategy.

Once we create that content, we’d track our mentions and citations and our brand sentiment to see how things are evolving with that content strategy. Based on that data, we can make judgments about what content is working or what’s not working, and we can lean more into the areas bringing visibility.

I’d also look at referral traffic from the different AI assistants to determine what platforms are sending us the most qualified leads so we could tie that back to conversions. I’d also measure branded traffic to the website, maybe using Google Search Console or Bing Webmaster Tools, to see if we are getting an increase in branded traffic that correlates with more AI visibility.

But ultimately, we’re trying to tie everything back to the business goals, in this case selling used cars. While AI visibility is important, and citations and mentions are important, what ultimately matters is getting the right visibility with the right car buyers to lead to profit for the business.

How would you handle a situation where your content is not being surfaced or cited by AI models, even though it’s high quality?

First, I would check to make sure the content is eligible to be fetched on the technical side, making sure there’s not a no-index tag, the folder in the URL structure isn’t being blocked by robots.txt, and we aren’t relying on client-side rendered JavaScript.

We can check indexability in Google Search Console, for example, and confirm that the page is indexable. And if it is, then the next focus would be on the content itself.

There are two aspects to content: there’s the structure of the content, and then there’s the quality of the information we’re mentioning.

On the quality side, if the content doesn’t contribute any new information, if it’s just commodity content that can be found anywhere, then it’s unlikely to be of value to AI assistants when they’re synthesizing their answers. So we’d probably need to introduce original data, quotes, statistics, more external sources, whatever it may be to make our content more valuable and more unique for the user and ultimately for the AI assistant.

Then it comes to the structure of the content. We want to make sure that we’re using a logical heading hierarchy, that we’re creating atomic sections of content that can be retrievable as chunks, because as we know, AI assistants take chunks of content and then convert that to vectors. To make our content be atomic, we’d use tactics like bottom line up front (BLUF), which is a writing style where you put the information directly under the heading so the reader can get to the answer as quickly as possible.

We’d also want to use semantic triples, meaning subject, predicate, and object, which is just a cleaner way of writing that AI assistants can easily understand and extract.

We also want to add as much detail and related entities in the content as possible. So rather than saying, “I have a used truck for sale,” you’d say something like, “I have a used 1997 F-150 in black and white color with a four-wheel drive” and then maybe list the price. So you’re giving the models more detail than just vague statements.

How would you evaluate the success of a GEO campaign, and how would you adjust if the results weren’t meeting expectations?

First of all, you have to identify what your measurement for success looks like. It could be referral traffic from AI assistants; it could be mentions and citation share relative to competitors, like your share of voice.

In general, with the tools available to us, I think you’d need to use a tool like Profound or Peec AI to help measure your success. You could also use tools like Bing Webmaster Tools’ AI performance section or Microsoft Clarity’s AI visibility section, but those are limited to Copilot, so they’d be more directional than holistic.

So I think you’d want to use a tool like Profound that gives you more coverage across different prompts and AI assistants. Say, for example, we wanted to get more visibility for our used car dealership. In that case, we’d identify who our target audience is: maybe we have truck buyers, car buyers, SUV buyers, and then we create topics and related prompts around those. Then we’d identify visibility gaps with our competitors.

So say, for example, the target prompt was “I want a used F-150 in Denver, Colorado.” If we’re not showing up in that answer, we’d want to identify which sources are. Let’s say we saw a competitor getting cited for their sales page focusing on F-150s, then in that case, we’d want to create (or optimize) similar content for our website to try and fill in those gaps.

In terms of success, we’d measure that ultimately by our visibility’s contribution to business goals. So we’re looking for qualified traffic or qualified leads that equal conversions. Ultimately, referral traffic from AI assistants that contributes to conversions would be most important. But then we can also look at an increase in branded traffic that correlates with more AI visibility. So it really kind of depends.

For instance, say we’re using Profound’s visibility metrics as our baseline, then we would look at our AI visibility score over time. We would look at our mentions and citations over time. We’d look at our brand sentiment over time and see if we have improvement there. And if we do, then that’s confirmation that our strategy is working. But if we don’t, then we have to reevaluate which of our content contributed to improved metrics and which didn’t, and then understand why. Maybe it wasn’t the right type of content. Maybe it was the way it was structured or the quality of it.

So ultimately, to determine success, we need to identify what success means up front and then correlate that with our tools that we have available and then measure it over a time period compared to competitors.

How do you approach aligning GEO strategies with a team that may not be familiar with generative AI or AI-driven search?

I think you have to break things down to their simplest form. In GEO, there’s a lot of jargon that’s been developing around certain tactics, but ultimately, our goal is to create high-quality, helpful content that aligns with the buyer’s journey and that gives the brand more visibility in AI assistants. If you break things down like that, you can explain them simpler to team members.

Since SEO is so foundational to GEO, I think that understanding the SEO basics, and this is Google’s advice as well, is the first step of a GEO strategy. There’s a lot of information out there about SEO best practices, but I think we’d start with Google’s guidance around SEO best practices and around helpful content if I were to start educating someone about GEO. Then we can get into more advanced topics.

So, in general, I think it helps to break GEO down to its simplest form, which is that we’re creating helpful content for users that gets us AI visibility and contributes to brand awareness and ultimately sales.

How would you integrate GEO strategies into an existing content team’s workflow, and what challenges do you anticipate?

If the content team already has a successful SEO strategy in place, then I think integrating a GEO strategy is just a matter of identifying the topics or prompts that need to be covered because the team is likely already using traditional SEO keyword data and creating topic clusters based on that. But with GEO, the user journey can be different given query fan out, where there are more queries that could potentially be covered. We’d also want to focus on topics, not necessarily individual queries, but the idea is that the user journey could be different for GEO than SEO, so the types of content that we need to create might be different.

We can also use tools like Profound to identify content gaps with competitors. But, to summarize, if a content team has an existing SEO process in place, then it’s pretty simple to integrate GEO strategies because the best practices of SEO content kind of translate into GEO best practices: using logical heading structures, having atomic paragraphs that can stand in isolation and be chunked, using bottom line upfront writing (BLUF) style, using semantic triples, naming related entities, using internal and external linking, all these kind of things that are SEO best practices apply to GEO strategies as well.

There is also a bigger focus on earned media in a GEO strategy, so we’d probably need to focus more on digital PR or social media or video content to integrate that into our overall content strategy. But in general, if there’s an existing SEO framework in place, that’ll contribute to a good GEO strategy.

On the flip side, if an SEO strategy doesn’t exist already for a content team, then we’d really need to go back to GEO/SEO basics, and we’d need to do a full content audit to understand which of the content on the website is valuable for the user journey and which isn’t, and probably prune some of that older or off-topic content. We’d also probably need to reevaluate the structure of the content to make sure that it’s more extractable for AI assistants. So there’d be more fundamental work to be done there.

I also think that there are ways to automate content production for GEO, but you’d need to be very careful with how you do that because you could easily end up generating commodity content, which is ultimately not going to help you long-term and will actually hurt your SEO and GEO prospects. So if you’re going to be automating content, you need to do so in a smart way that has a lot of human involvement — human in the loop — and that probably uses RAG, retrieval augmented generation, to contribute sources and original data to the content.

How would you ensure that your GEO efforts remain adaptable as AI-driven search evolves?

The important thing is staying up to date with best practices, so I’d devote a lot of my time to being on X and LinkedIn and staying up with industry blog posts. I’d also do my best to stay up-to-date with the scientific literature and patents around AI, but a lot of that can be read through analysis of other SEOs or GEO practitioners who write about it.

Having said all of that, it also comes back to the user experience. Fundamentally, if you’re optimizing for what’s best for your users, then you’re going to position yourself well in GEO/SEO or just marketing in general, as platforms like Gemini, AI Mode, AI Overviews, or ChatGPT ultimately want to do right by their users first and foremost.

How would you measure the impact of GEO on brand authority or trustworthiness, beyond just traffic or conversions?

The first thing that comes to mind is looking at branded search volume and whether that’s increasing or not. This involves using tools like Google Search Console or Bing Webmaster Tools to see if visibility gains from your GEO strategy are having an impact on branded search, because a lot of times people will go to an AI assistant for research, but then ultimately go back to a search engine to search for the brand.

Also mentions and citation rate can correlate to brand authority and trustworthiness. If we’re being mentioned more in AI responses and we’re being cited more, that’s probably an indication that the brand is being trusted more. Ultimately, we want to be part of the training data as well. This is more of a long-term strategy, but I would look at how we’re appearing in those prompts that don’t start a web search — the ones that only rely on training data. On May 7th, ChatGPT started linking to the brand name more in its responses, while its proportion of non-web-search answers has purportedly gone up. So ultimately your goal is to be part of the training data.

We could also focus on knowledge graph presence, so how the entities of our brand and related to our brand are positioned in the knowledge graph, whether they’re there to begin with. And if not, then we can use tactics like Wikidata, structured data, and maybe just the content strategy, in general, including third-party mentions, to try and contribute to the entity clarity in the knowledge graph to contribute to the brand’s overall authority and trustworthiness.

Can you describe a time you adjusted a content strategy due to changes in how AI models or generative search were evolving?

I updated the entire content strategy to refocus the topics from a sort of a keyword-based approach, using tools like SEMrush and Ahrefs for content ideation, to using tools like Profound to look for gaps and competitors relative to citations and mentions.

The first step was updating the topics. The second step was reframing how content is written. We know that AI models don’t necessarily look at the entire page. They look at chunks or passages and convert those into vectors. So I had us write content that was atomic, where we followed an organized heading structure throughout the page, used techniques like bottom line up front (BLUF) where you’re answering the question right away, used question headings, used more FAQs and FAQ schema, basically more of a question-answer type strategy.

A good example of this approach was done for a vegan dog food company, where initially we were writing a lot of keyword-based topics, like “can a dog eat this or that ingredient.” These articles were historically driving a lot of clicks and traffic from that, but their traffic slowly started declining as AIOs rose. So we started focusing more on AI visibility and readjusted the content strategy around citations and mentions.

How would you approach aligning a GEO strategy with a broader marketing team that might not yet be familiar with AI search concepts?

GEO strategies often go far beyond what traditional SEO strategies did in terms of how they cooperate with other marketing or business departments. For a content strategy, for example, you might need to work with video production, social media, or digital PR teams. Pretty much every type of content that you can think of can contribute to the GEO ecosystem because AI assistants feed on all of it. So it’s important to get aligned.

The biggest factor is teaching other teams to understand how AI visibility works, at least at a high level. This involves relaying to them the importance of what we would call “entity clarity” and getting brand alignment — similar messaging across different channels, making sure that the story being told is cohesive and that we’re being rich in offering original information.

How would you handle a situation where AI models consistently misunderstand or misrepresent your content?

The first part is to understand what information is being misrepresented and then tie it back to the original sources being cited. Are those sources owned media that we control, or are they earned media (third parties) that we have to reach out to and correct?

We also want to contribute new content in order to help fight disinformation. I worked on a situation where a brand had a government recall years ago, but whenever they’re mentioned in AI assistant answers, it always mentions this recall. What we did was created written and video content specifically around the recall, but from our point of view, to help influence how the LLMs understand the brand and would talk about it in a more balanced way.

How do you prioritize which parts of a website or content library to optimize first for GEO?

It starts with identifying the business goals of the website. Say there are multiple services or products offered, then we’d want to identify those areas that are most important for the business. Then we’d correlate those lines of business to prompts, and then we would track those prompts, usually with a commercial intent, to identify which sources of content are feeding the mentions and citations for them. It could be blog articles from the website, it could be product pages, or it could be from a product feed.

But that’s how you identify the most important parts of the website or content is you start with the business goals, then you look for the gaps in your strategy, and then you try to fill in those gaps with existing content or new content, both owned and earned media.

Can you explain how you measure whether your GEO efforts are improving user engagement or satisfaction, beyond just rankings?

You would look at referral traffic from AI assistants in GA4, and then you would look at the engagement rate to your cited pages and understand if those engagement signals are going up or down.

You could also look at an increase in branded traffic as an engagement signal. So, for example, if you’re getting visibility in AI assistants, then more people are seeing your brand and will likely go to a search engine and search for your brand name. So an increase in branded traffic also might be a contributor to engagement signals.

But ultimately, you’re looking for conversions. So you could use referral traffic and then correlate that to conversions to see if you’re converting your AI visibility into conversions for the bottom line.

How would you handle a situation where your content is performing well in AI-driven results but not seeing an increase in actual conversions?

The first thing you have to look at is what does success mean? Are we getting visibility, mentions, and citations? And if so, then we need to understand if those are going to our own website or they are going to another website. Because it’s possible that we may not be getting conversions on our own website, but we could be getting third-party conversions, which ultimately contributes to the bottom line as well.

I think we also need to look at the CRO of the website pages to make sure that the pages are optimized for people to convert on. CRO strategies involve items like having CTAs in prominent places, maybe running a sale or promotion, or bundling products together.

You’d also want to look at assisted conversions. Are you seeing conversions come from other channels, like an increase in paid media or email conversions that could tie back to AI visibility gains? So, you have to look at the situation holistically and across channels, not just focus on your AI visibility metrics related to conversions.

Can you describe how you balance optimizing for AI-driven search while still ensuring that your content remains valuable for human readers?

I take what I call a people-first approach to SEO and GEO. So ultimately, I approach things by thinking about the human reader first, because a lot of the optimizations we make for GEO in terms of how we structure pages, use headings, use semantically related entities, add internal and external linking, use a bottom line up front writing style, incorporate semantic triples, and use specificity over vague language, that also converts to better writing for users. For example, instead of saying “used F-150,” we’d write, “1997 F-150 Super Cab with 100,000 miles.”

But ultimately, if you’re optimizing for your users, then you’re going to be doing well for GEO and SEO because, especially on Google’s side with Gemini, AI Mode, and AI Overviews, that’s what their guidance has always been — to create helpful, reliable, people-first content for users. So I follow that philosophy and before I make an optimization, I ask, first and foremost, is this going to be beneficial to users? If it is, then it’s probably going to be good for GEO.

But then we also have to consider that users are likely enjoying getting their answers synthesized in AI assistants. So by making content extractable for AI assistants, we’re not necessarily going against what users want. So it kind of works both ways.

How do you handle situations where different AI platforms, like different chat-based systems, interpret or display your content differently?

This question strikes at a fundamental aspect of GEO strategies and a mistake that I think people make, which is they log into a tool like Profound or Peec AI and they just look at their visibility score as a whole, and then make judgments based on that. In reality, that score is an average of all these different AI assistants that they’re looking at. And these AI assistants work differently, and they prioritize different sources. So in a strategy, you need to really understand what your visibility is like in individual chatbots, and then understand what sources those chatbots are favoring.

For example, Gemini, AI Mode, and AI Overviews are going to favor YouTube a lot. ChatGPT might favor Wikipedia, so it depends. That’s why you’d need to use a prompt-tracking tool to analyze differences at the prompt and platform level to find out what sources are being favored and referenced.

Ultimately, you’ll want to have a holistic strategy that focuses on your AI visibility as a whole, but then also target those individual AI assistants that you think are going to bring you the most value in terms of benefits for your business. Maybe you look at your referral traffic and conversions and you see that ChatGPT brings you the most qualified users. Well, then you probably want to lean into ChatGPT visibility as part of your strategy. So you really want to have a nuanced approach, as opposed to trying to go after all AI assistants at once.

How would you handle a scenario where a competitor is consistently ranking higher in AI-generated responses than your content?

The first step is to understand what sources their visibility is stemming from. Is it from websites that they control, or is it from third-party websites they’ve been mentioned on? And then we have to target a similar strategy. If citations are coming from their website, then we target our owned media and focus on blog pages, product pages, or whatever type is feeding the answer. If it’s third parties, then we’d want to try and get listed on those same sites or other sites of equal value and relevance.

But, you also don’t want to mimic another competitor too closely, because you want to still be true to our own brand voice and strategy. So we want to have unique content that offers more value than what competitors are providing. For example, if they have a listicle blog post with 10 reasons to buy a certain product, we want to have a similar blog post, but ours should be more in depth, feature more expert insights, have more external links to authoritative sources, have more unique multimedia, or whatever it is that gives us an advantage in terms of value to users.

Fundamentally, though, we need to understand where a competitor’s visibility is stemming from and then create a similar strategy of our own.

How do you stay current with evolving AI search algorithms and ensure your GEO strategy remains effective over time?

I’m an avid user of social media. X is my go-to, although I also use LinkedIn and BlueSky. I was a big part of SEO Twitter back when Twitter was Twitter, so I have a pretty good list of people that I follow to stay up to date on industry trends.

I also pay attention to a lot of new articles that get written in the GEO and SEO space. I also try to stay up to date with academic papers and patents, although I often rely on an LLM to analyze them, or I read analysis from others in the industry.

Testing is also a good way to stay ahead of the game. Trying new tactics, ideally with users in mind first, is a great way to explore the limits of what’s working for AI visibility.

Can you walk me through how you’d measure the success of a new piece of content in a GEO strategy, from launch to evaluation?

We’d first need to identify what the purpose of that piece of content is, so we’d map it back to our particular prompts or topics that we’re trying to get visibility for. Next, we would see if that piece of content gets indexed in search engines, and second of all, whether it gets cited in AI platforms. From there, we’d map it back to the prompt or the initial topic and then try to determine if there was a correlation there in terms of AI visibility gains. But ultimately, our pages need to be indexable and citable for them to have value, and then they need to relate back to a prompt, and ideally that prompt aligns with a persona and business case.

How do you approach balancing short-term wins—like quickly improving visibility—with long-term, sustainable GEO growth?

Shorter-term wins might be like identifying a particular prompt or topic and then getting cited or mentioned for that topic, something that has to do with web search, whereas a longer-term goal might be getting your brand part of the training data set so that it’s showing up long-term and particularly for prompts that aren’t doing web search. So there are differences there between short-term and long-term, but ultimately you’re trying to increase your brand’s visibility in the training data set, which is the long-term goal.

When you’re creating a GEO strategy, how do you ensure that the content you’re optimizing remains adaptable as AI technologies evolve?

First of all, you need to focus on the user experience above all else. If the content satisfies a search intent that you know is going to be there for a while, then it’s going to have long-lasting value. In terms of optimizations, you can follow best practices as they’re known today, which would include a logical heading hierarchy, bottom line up front writing style, using semantic triples, and using related entities.

But ultimately, for your content to be optimized long term, you’re going to have to stay adjustable and keep if fresh. So as tactics change or information evolves, you may re-optimize the content. But as long as you’re focusing on the user experience first and foremost, you should be in good shape long term, because that’s ultimately what these AI assistants are trying to reward, especially Google-based platforms, which is user experience.

How do you balance automation tools—like AI-driven content creation or analysis—with human creativity and judgment in your GEO strategies?

You have to think about creating unique, valuable content from the start because ultimately, you don’t want to be automating commodity content, which is very easy to do just with generic prompts. So you need to think about how can you contribute uniqueness and add value from the start. Often that comes by following a RAG model (retrieval augmented generation). So you’re setting up your sources to be pulled into and synthesized as part of the answer, feeding like a brand knowledge portfolio basically, which has your brand voice, writing guidelines, previous writing samples, and any unique data you have available, like independent studies you’ve done.

You also need to find areas to have humans in the loop. You don’t want to automate blindly. Ideally, someone will editorialize the content, adding unique perspective, quotes, or data. Automation can be powerful, but you have to be careful with it because you can easily create a lot of commodity content of low value.

How would you handle a scenario where your GEO efforts boost visibility but the AI assistant is summarizing your content inaccurately?

That’s a hard question to answer because we don’t necessarily have a clear concept on how content gets synthesized into AI answers after it goes into the retrieval set as candidate passages. I think the first step is to identify what information is being returned incorrectly and then tie that back to the sources we believe it’s coming from, whether it’s our own or whether it’s a third party. Then we’d have to adjust those sources accordingly to trace back where the inaccuracy is coming from and then try and cure it at the source.

But again, nothing’s going to be perfect. There’s always some risk of hallucination. So maybe in addition to trying to fix that original source, we also create additional pieces of content that compound and add additional context around the topic to try and influence the AI assistance answer overall. Of course, the more clarity we can get for our core and related entities in the knowledge graph, the more accurate information is likely to appear in AI assistant answers.

How do you determine which AI platforms are most valuable for your business’s GEO strategy, and how do you prioritize efforts accordingly?

You firstly have to use different tools to determine the value of AI visibility for your business. You can use a tool like Profound to measure your visibility in terms of citations and mentions by platform, according to certain prompts. Ideally, these prompts will have the most business impact for you, typically commercial prompts that align with your main lines of business. You also want to look at referral traffic in GA4 for the different AI assistants and look at your traffic, your engagement rate, but also your conversions from that. Kind of combining those together, you can get an idea of which AI assistants you’re showing up in the most and which are contributing conversions for you the most.

Equipped with the knowledge of which AI assistants are most valuable for you, you can then either go one of two ways. You can lean into the valuable AI assistants and focus on trying to optimize for them further, or you can focus on some of the other AI assistants where your visibility is weaker. My suggestion would be to double down on the platforms where you’re strongest and try and build topical authority in those AI assistants as much as possible first, especially if they’re giving you conversions from referral traffic, and then focus on the kind of secondary AI assistants where your visibility is a little bit lower afterward.

When working cross-functionally—for example, with product or PR teams—how do you ensure that GEO goals align with broader business initiatives?

The first step is to have alignment between the teams. So you’d probably have to have a weekly or monthly sync where you talk about priorities for the business. You want to get aligned on what the business priorities are in terms of products or services of priority. From there, you want to translate that back into how that works for AI visibility. So you’ll want to identify which of those lines of business correspond to specific topics or prompts to try and get AI visibility in. Then you need to identify the sources of the citations or mentions providing that visibility. And then you need to make sure that business critical information is uniform between the different departments. Ultimately, GEO success isn’t just about your owned media. It’s about your owned media, your earned media, the entire ecosystem, from video to images to documents. So you want to have alignment on goals from the beginning, and usually that requires some sort of regular meetings.

If you had to advocate for additional resources—whether that’s tools, budget, or team members—to enhance your GEO efforts, how would you present that case to leadership?

It needs to start with the business case. We firstly need to identify what impact AI visibility is having on the bottom line. Some of that is hard to track because visibility translates to awareness, and awareness doesn’t necessarily translate to conversions. But we can look at metrics like referral traffic from AI systems in GA4 and see how the conversions are from there, and then we can extrapolate what kind of impact AI visibility is having.

For example, if referral traffic from AI assistants has higher engagement rates, lower bounce rates, and is converting at a higher rate than, say, traditional organic search or even paid search or some other channel, then that can justify an investment in GEO.

Outro

GEO is a big topic, and I’ve only scratched the surface with this article, but I hope it’s provided some clarity as to why I think GEO is separate from SEO (and why I think it’s not).

In the spirit of two things can be true at the same time, let’s end with a classic TOOL song about identity.

Until next time, enjoy the vibes:

Thanks for reading. Happy optimizing! 🤗

Related posts