Transcription
So, right now, your potential customers are having conversations with AI about which local business they should choose, and you're not a part of that conversation. If your business isn't showing up when someone asks ChatGPT, "Who should I hire for this?" You're invisible.
For context, I've been running my SEO agency since 2017, and I've built it to seven figures in under 3 years. I have seen every single SEO and AI trend since then. So, in this video, you're going to learn the complete AI search optimization system that works in 2025. You're going to discover the five AI search ranking factors that matter most, how to structure your content so that AI recommends your business first, and the exact prompts and tools you need to dominate AI search results in any market.
So, I'm going to start by making sure we understand how AI and ChatGPT in particular decides which businesses to actually recommend. At its core, ChatGPT is pulling from three main sources. The first is an index of the internet. Now, ChatGPT uses Bing, not Google for this. Remember that Microsoft was one of the largest early founders for OpenAI. So, ChatGPT uses Bing. Second, ChatGPT relies a lot on high trust databases like Reddit, Wikipedia, Crunchbase, other ones like that. And third, it reads and understands public reviews. So the core truth of this is that ChatGPT is fundamentally doing the same thing that Google does. Its goal is to decide if a business is trustworthy based only on the information that ChatGPT can find about that business online.
Now, way back in 1999, Larry Page wrote a Stanford PhD dissertation that put forth the idea of using other websites as a way to determine this trust. The theory being if a trusted website links to you, you must also be a trusted website because trusted websites don't link to spam. That fundamental idea has made Alphabet, Google's parent company, into a multi-trillion dollar business. So of course ChatGPT is going to come up with similar results. Google has been perfecting this for 25 years now.
Of course, the critical difference here is that Google will show you exactly three results, maybe a fourth one with ads, and those three results are listed as ABC with a map and the business name and not much else. ChatGPT is usually going to recommend around three local businesses, but it will give you an explanation of each of those, not just a list. It will tell you why it's recommending those three businesses in particular. And we also know that most of the local queries for Google are a handful of words, two, three, four words, plumber Houston or something similar. Whereas with ChatGPT, users are giving it much more context for what they're asking for. Hey, can you find an emergency plumber who's good at dealing with a leaky water heater and available to come now? Much more context for ChatGPT to go out, read reviews, and find the right company to recommend for that very specific question. Quite different from Google where the winners are ranking high for plumber Houston.
Now, from a data processing standpoint, we know that Google processes around 16.4 billion searches a day. Now, ChatGPT is significantly less than that. It handles around two and a half billion prompts a day. Still a massive number, but about 1/5th of what Google does. What's interesting about this, and I hear from people all the time, that ChatGPT is killing Google. And it's simply not true. Yes, ChatGPT prompt volume is growing all the time, but so is Google search volume. Yes, Google search volume is still growing. What we see users doing is they're going to ask ChatGPT for recommendations for their very specific use case. Then they're going to go to Google and double-check those recommendations, do verification on Google search.
Now, we have known for years that if you show up on top of Google search, that's a massive signal of trust. Searchers just trust your company more when Google puts you on top of the list. The same thing happens for ChatGPT. When ChatGPT recommends your business, that's a massive signal of trust for the person asking for a recommendation. The magic happens when you show up in both. If your business is recommended by ChatGPT and shows up on top of Google search, man, when that customer calls your business, it's going to be a very easy conversation to turn them into a paying client.
Now, when we talk about adapting ChatGPT prompting and searching, the younger generations, Gen Z, millennials, they're moving much faster toward ChatGPT than the older demographics. Kind of funny because ChatGPT uses Bing's index and not Google's index. My joke about Bing's average user is that they are typically a little bit more fluent and a little bit older than Google's average user. This is because most of the time people who are using Bing just bought a new Windows computer and don't know how to change the default search engine. So, humorously, we now have the older demographics using Bing and the younger demographics using Bing because ChatGPT uses Bing. It's pretty common to see that ChatGPT's recommendations are highly correlated with Bing search results. If you're ranked higher on Bing, then typically it's more likely that you're going to get that recommendation from ChatGPT.
Now, when we talk about testing your current visibility, we want to see which businesses ChatGPT recommends. But if we just go to ChatGPT and ask it who's the best plumber in Houston, if it knows that you're a plumber or that you serve plumbing companies, it's going to want to recommend the business associated with you. ChatGPT is programmed to try to make its users happy. Google does this also. If you frequently visit a plumbing website in Houston and then search for plumber Houston on Google, it will likely rank the website that you visit a lot highly. The way we get around this in Google is we go into incognito mode or something similar so that Google doesn't see what we've been doing on its platform recently and it's just going to show us what its algorithm sees as the best option. So we need to basically go into incognito mode for ChatGPT. Of course, ChatGPT doesn't have an incognito mode. So we came up with a prompt that emulates an incognito mode as closely as possible. I'm going to show that prompt on screen. Feel free to pause it. Grab the prompt. You can also grab it in my school community. I'll put a link to that in the description below.
We also want to make sure that we can also track how much traffic you're actually getting from ChatGPT with Google Analytics. Just go into your Google Analytics account, click on acquisition on the left rail, then hit traffic acquisition, and in the center there should be a small search box, type ChatGPT, and Google will give you exactly how many clicks you receive from ChatGPT over the time period that you've selected.
So, let me quickly touch on why I keep talking about ChatGPT in particular. ChatGPT has significantly higher prompt volume than any of its competitors. So if we were talking about optimizing for Perplexity or for Grok, that would be the equivalent of talking about optimizing for Bing. Why would we do that when Google dominates the search space? So I keep talking about ChatGPT because they're the player that we need to rank for first. Once ChatGPT is recommending your business, then we can start to look at some of the other AI systems and optimizing for them as long as it doesn't jeopardize what we've got with ChatGPT.
So now that we have a pretty good understanding of how ChatGPT is deciding which businesses to recommend, I want to shift and talk about how to structure content that ChatGPT can read. If ChatGPT can't read or can't understand your content, it's not going to recommend you. And we know that the AI prefers conversational content that sounds like a human expert explaining things. There's a study done in 2024 that well-organized authoritative content increased inclusion in AI answers by 37%. So with that in mind, we know that AI systems gravitate toward comprehensive content coverage that have clear sections. They have structured content formats, lists, tables, FAQs. These are shown to improve the chances of being featured in ChatGPT's answers. So really when I talk about wanting to write well-structured easy to understand content so ChatGPT understands what you're saying, well-structured easy to understand content is also what you should be writing for Google. So we're not talking about having to learn a new discipline here. It's just doing good SEO that serves the dual purpose of helping Google's algorithm understand what you're talking about and helping ChatGPT understand what you're talking about.
You may be familiar with the acronym EAT, Experience, Expertise, Authoritativeness, and Trustworthiness. This was a term coined by Google and it appears many times in Google's quality search rater guidelines. And basically what this means is that Google wants to see that the businesses that the websites it's recommending demonstrate EAT. In recent algorithm updates, Google has been adding things like watching brand search volume, tracking entities. Basically, in a way to make sure it understands that this website is experienced, has expertise, is authoritative, and is trustworthy. In a very similar way, ChatGPT also cares a lot about EAT. And we talk about this when we talk about mentions. When your business name or your personal name is mentioned in high trust authoritative sources, ChatGPT will be more likely to recommend you. These are these external trust signals. We also want to build EAT into the content that we write. Use real names of the writers. Make sure those names match mentions on third-party websites.
So when we look at subheadings, we want to frame these subheadings typically as natural questions. So a good subheading might be, "What are the best running shoes for marathon training?" We would not want to use "best running shoes rank." Now when we do frame these headings as questions, we want to answer conversationally and directly in the first sentence and we're going to repeat the question in that answer. "The best shoes for marathon training are XYZ." Beyond that, share firsthand experience in the content. Share case studies and include statistics and cite reputable sources. One of the key things that we do at my agency for every article we publish, even in local SEO is we're always going to link to a third-party trusted source. Whether that's Wikipedia, harvard.edu, whatever source you like. Showcase your credentials, your industry expertise, and this is very important. Use a natural variety of related terms, not just repeating the same keyword. We want to create what's called a semantic web around your topic.
So, let me explain what that means just a little bit more. I'm going to use the example of Star Wars because I'm a giant dork. So, if I were trying to get recommended by ChatGPT about something Star Wars related, I wouldn't just write an article about Star Wars and write Star Wars in that post over and over and over again. Instead, I would make sure to use semantic keywords that ChatGPT knows are related to Star Wars. That helps it see my content as authoritative and trustworthy. So examples would be lightsaber, Darth Vader, Luke Skywalker, Princess Leia, General Organa, and I know those last two are the same person. So that's what we mean when I say creating a semantic web around the topic. And honestly, that's the exact same thing we do when we're writing for Google's algorithm.
So, let me share with you the single content writing prompt that we use pretty frequently that helps write content optimized both for Google's algorithm and for ChatGPT. As before, go ahead, pause the video, uh, screenshot it, give it to AI, and AI can transcribe it for you or you can grab the same prompt in my school community.
So now that we have the right type of content, this content structure is giving AI something to work with. But ChatGPT and Google is also going to rely on external trust signals for ChatGPT. That's going to be reviews, citations, and mentions. Let's talk about how to get those.
So I want to start by talking about reviews. This is very interesting because reviews are not a significant rank factor for Google search itself. I cannot tell you how many times I have seen businesses with thousands of reviews being outranked by businesses with a few dozen. They are important for Google search because if you have 50 reviews and the other two in the top three have a thousand, you're not going to get any calls, but you can still rank with many fewer reviews. ChatGPT is a little bit different. ChatGPT will read reviews for context. Your Google reviews, yes, but also Facebook reviews, Yelp reviews, any reviews it can find, including what people are saying about you on Reddit. And it uses those reviews to decide what specific services your business is good at. For example, if somebody leaves a review and says, "Hey, I use this business to replace a lead main drain pipe in Chicago and they did a great job." Then someone else goes to ChatGPT and says, "Hey, I need to replace a lead main drain pipe in Chicago. Who should I call?" You're very likely to get recommended because that review listed that very specific service.
On top of reviews, we want to make sure that the name, address, and phone number are consistent across the entire internet for your business. AI struggles to understand that that entity is the same one if the name, address, and phone number start to be different. So to build up this third-party trust, we need to make sure we have consistency. So let me show you something. We call it the citation hunter prompt. It's going to get ChatGPT in deep research mode to crawl the internet and looking for all of the mentions of name, address, phone number similar to the one that you're using on your Google Business Profile. Any of them are different, you'll be able to go in and remove it or update it to make sure that you're getting the full amount of external trust that you should be getting.
Beyond this, there are a lot of easy to get, high trust, high authority third-party websites that come a long way toward trust. Your local chamber of commerce and not just the closest one. You can get listed at half a dozen chambers of commerce if they're near you. The Better Business Bureau, Yelp, LinkedIn, Wikidata. Wikidata is the structured database behind Wikipedia. Very easy to get listed there and quite hard to get listed on Wikipedia. There's also industry directories and Reddit. A lot of these are free. The ones that aren't, like the Chamber of Commerce, are typically fairly low cost. Getting listed on these is beneficial both for ChatGPT and for Google search itself.
So we talk about reviews. We also want to make sure that we're responding to every review left that we can respond to. This is of course showing active business management, but I understand that it can be very difficult to get people who are reviewing you to leave the name of the specific service they hired you for. Well, you can work that into your review responses. For example, you could leave a response that's something like, "Hey, thank you for calling us. We're happy to have been able to provide emergency water heater service. We pride ourselves on same-day response." Something like that. It allows you to fit these specific services or expertise in your review responses that ChatGPT will then pick up on.
So, the last thing I want to show you is a prompt that we use almost every day at my agency. This helps us find local sources of trust. This will get ChatGPT in deep research mode to look for local organizations, uh, charities, uh, events that are looking for sponsorship. And in exchange for that sponsorship, they're going to mention your business. Hopefully, they'll also provide a link back to it. This will help establish that external validation both with ChatGPT and with Google search. A lot of these aren't going to be free, but they can typically be had for a couple hundred, and they are well worth it in terms of how much external trust and validation they're going to build.
So now we have our arms wrapped around external validation signals. Let's shift and talk about something a little bit more technical. This is code that explicitly tells AI about your business. So this code is called schema. The way I think about schema is if you were moving from one house to a new house, you would pack all of your worldly belongings into moving boxes. Schema is the label on that box. What's in the box? Now for ChatGPT, if you don't have schema, it has to open the box and ruffle through everything to figure out what room in the house it should go to. Yeah, it works, but it's much harder than having the box label. Schema is also helpful for Google but not nearly as much as it is for ChatGPT. Remember that Google basically built itself on crawling the internet and understanding everything immediately. So in contrast to ChatGPT, think of Google as having X-ray vision. Again, the schema can be helpful, but with X-ray vision, it's less helpful to know where that box should go straight away. But schema is code that humans don't see when they visit your website. It's your only opportunity to talk directly to Google's algorithm and directly to ChatGPT. Google itself has said that schema has become especially important in the age of AI. So our goal with schema is to make every important piece of information on your website as unambiguous as possible to machines.
So let's talk about schema types. Most local businesses that I look at don't have any schema. So, it's an easy advantage for you to start putting some on your site. The schema types to prioritize are organization schema, article schema, author schema. They should be on almost every single URL across your domain just to establish who you are and what your organization and what your website is. Local business schema is also incredibly important. Business type, hours, location, service area, and all the services you offer. That local business schema should only go on your Google Business Profile landing page. That's going to be very helpful to ChatGPT and to Google itself to make sure Google understands that business website is the same one that the GBP is about. We also have things like geographic schema if you're trying to establish geographic relevance and FAQ schema if you have an FAQ on that article. Schema should be written in JSON-LD format. If you don't know what that means, no problem. It's recommended by Google. And really, you can find any number of tools online that will automatically generate schema for you. Don't stop there, though. Give that schema that's been automatically generated over to Claude. It's typically much better at writing code than ChatGPT. So, give it to Claude and ask it to improve it. Ask it to add more details, more information. And every time you get schema back, always, always test it with the structured data testing tool. Both Google's test and schema.org's own test. They will give different results. Make sure your schema has zero errors, zero warnings on both tests. If there are any errors or any warnings, give that back to Claude. Tell it what the errors and warnings are and Claude will rewrite it eventually and fix it.
Now, I briefly mentioned FAQ schema. This is really important. I'm going to give you a prompt that we use to find the actual language and questions that your customers are asking about your service in your city. When you use this prompt, take the exact answers you're getting and build them into FAQs on the pages. Then build the FAQ schema that mimics that same language. If you build FAQ schema and use a question very similar to what somebody prompts ChatGPT with, ChatGPT's answer is almost going to be word for word what you have in the answer block.
Okay. On top of schema, we also want to just do a quick double-check of the robots.txt file. You can find this by going to yourdomain.com/robots.txt. Just make sure you're not blocking OpenAI's GPT bot, Google's Googlebot, Microsoft's Bingbot, or any other crawling bots like that. I know it might be tempting to want to block GPT bot, try to protect your content from AI, but if you do, you will never get recommended. You can't be recommended if you're blocking the bot.
So, that gives us the technical basis set on the website to make sure that ChatGPT knows what your website is about and who you are. Now, let's talk about figuring out what questions customers are actually asking AI. This is essentially a form of prompt engineering. I mentioned this very briefly when I talked about FAQ schema, but what we're trying to do here is understand the exact prompts that your customers are typing into ChatGPT so we can optimize the content on the website for those prompts. When we think about prompting Google, the prompt is really just two or three words, plumber Houston. Whereas an AI, somebody's going to type in, "I have a leaking water heater in Houston. Who should I call?" That's much more conversational, much more context-rich, and a lot more details than what people are typing into Google. Businesses and SEO experts have been trained for 20 years to optimize for these short keywords that people type into Google, and that causes them to miss conversational search behavior that's much more common with ChatGPT.
So go back and use the incognito ChatGPT prompt I gave you at the beginning of this video. And when ChatGPT provides a recommendation, ask it to explain why it's recommending those. Ask it to explain what questions are people asking about that service in that city and it will give you answers. It will tell you what people are prompting it. Use the other prompt that I gave you to check Reddit, Quora, Facebook groups to find what people are searching for, how people are commenting, what language they're using about local businesses in your service area in your niche. Same thing. Our goal is to figure out the language that people are using to talk about your business. We can also use one of the most famous sources for this that exists. That is the "People Also Ask" section on Google itself. Those also typically convert very well to AI prompts.
Once you have all of this information, create a list of 10 to 20 variations of the customer queries. For example, if you're a plumber, you might have, "Who's the best emergency plumber in Chicago?" "I need a plumber who can come today in Chicago." "I'm looking for a reliable plumber for water heater replacement in Chicago." Test each one of those in ChatGPT using a version of the incognito prompt I gave you before. Is your business mentioned? What context? What position? This will help you start to identify the gaps in your presence when people are searching for your service in your city. Then go back to your content and add that FAQ section that answers all of these questions in natural language. Implement FAQ schema just like we talked about before and use a conversational tone both in the content itself and in the schema.
So, this is the exact same AI search system that my agency uses to dominate local markets and help hundreds of businesses become the AI's favorite recommendation in their industry. But if you find yourself thinking, "Well, this is great, Caleb, but where do I find local clients to implement this AI search system for?" Because having the most advanced AI optimization strategy in the world doesn't help if you don't have clients to use it on. So, that's exactly why I created this next video where I break down the exact five ways that I would use to get SEO clients if I was starting over today. The same methods I use to build multiple seven-figure agencies. I'll show you how to choose the client acquisition method that fits your budget and fits your personality and can get you your first paying customer within 30 days. When you can deliver AI-powered results this fast, clients will happily pay premium rates. So, check out that video now and let's get you some clients.