The New Way to Track Rankings for AI Search (Nobody’s Talking About This)

In this solo presentation Phil Risher explains how home‑service companies must shift from traditional keyword‑based SEO to AI‑driven search visibility. He outlines a three‑phase framework: (1) building an AI prompt list using frontline data (call transcripts, sales calls), Google Search Console long‑tail regex, and competitor comparison audits; (2) tracking three core metrics – AI Share of Voice (SOV), Mention Rate, and Generative Engine Optimization (GEO) – that together indicate how often a brand appears in AI overviews, how often it is cited as a source, and how well its content is consumable by large language models; (3) applying a diagnostic SOV‑vs‑Mention matrix to decide whether to focus on expanding share of voice (entity associations, high‑volume niche content, local citations) or improving mention rate (quantitative data, definition‑first answers, FAQ/service schema). He demonstrates how BrightLocal’s AI visibility tool aggregates these signals into a score and visualizes the matrix, and he stresses the importance of real‑world content derived from calls, videos, and podcasts to feed AI models. The talk concludes with actionable recommendations for building authority, defending market‑leader status, and a call to action for businesses to partner with his marketing firm.
Chapters
Back in 2017, when I started in SEO for home service businesses, it felt like black magic and voodoo talking to these SEO companies. You pay them thousands of dollars and they would say, "Yeah, your rankings are going up," but no one knew anything that was going on. It feels kind of similar right now with AI search and looking at visibility. People are asking, "Am I ranking for AI search? Am I ranking for ChatGPT? How do I know if I'm ranking?" And most people are guessing. This is very new, you know, AI search in general, and being shown in AI search is new to most people, but but also tracking your visibility on AI search is new. So we use a tool called BrightLocal to track rankings right now in traditional Google search, and then in their labs, they have an AI visibility search tracker that they gave us access to as one of their top partners. So in this video, I'm going to go through 3 different phases of what you need to do to set up your business to help rank, and also what AI visibility and tracking is going to look like in the future. So in this video, we're covering 3 main things. First is The phase 1, which is building your prompt list. Phase 2, which is the big 3 metrics you need to know. And then phase 3 is this diagnostic matrix is going to tell you what you need to do based off of the information. So if you haven't heard, search is changing. And in case you didn't know about it, 58% of Google searches right now end with zero clicks. The searcher never goes to your website. They never go to your website. They go to your Google Business Profile. They click your phone call, they click websites. They're doing their due diligence. AI overviews now appear on 70% of searches. So if Neil Patel did a study on this, if a search is longer than 6 words, so it's not just HVAC repair near me, it's how much does HVAC repair near me cost, that is going to cite an AI overview 70% of the time. AI overviews is a little AI overview section at the bottom of search. Very important. When someone asks ChatGPT best HVAC company near me, they get a curated list based off of what ChatGPT believes not 10 blue links to go down in a Google Business Profile. And so search is fundamentally shifting how people are interacting with these search platforms. And if AI does not know you exist, then you have zero chance of being in that result, of course. And your competitors are figuring this out right now. This is the hot thing that most businesses are talking about, specifically $1 million to $10 million plus home service businesses. All right, so first you need to build your prompt list. And prompt list is changing from what used to happen, which is keywords, is now shifting to conversations. So before, when we tracked rankings, you would go to a platform like BrightLocal or SEMrush or Ahrefs, and you'd set up a list of a bunch of keywords. And this would be like HVAC repair Dallas, HVAC repair Arlington, right? All these different services with the cities. And then every single week it would pull your rankings and then you would see where you ranked on these keywords in a neutral search. So it would basically give you a roll-up of, yes, my keywords are going up, or they're not, or they're going down, or what, whatever. I'm not saying that this is not important. However, when we talk about AI search, it's not about keywords because people are not searching by keywords, they're searching in conversation. So it's a big shift of how we're going to prompt to see if we're ranking for certain things. So here's 3 ways to build your AI prompt list, and this is what we're doing for our clients, but how you can think through this. So the first thing is the frontline strategy. This is probably the best way to do it that I found, and that is pulling your call tracking software or call recording transcripts and finding the exact questions that your customers are asking, and then use those as your prompts. So essentially, a call comes in, and let's say it's gutter cleaning, and then they say, uh, how much does it cost to get my gutters cleaned? Usually people take the bait and just give them a price. What you want to do is ask them, well, why are you calling me right now? What's going on? That is the prompt that you want. They're calling you because of this long prompt. That's the same question that they would go to ChatGPT or go to Google AI Overviews, AI Mode, Google Cloud, whatever, and search that exact reason why on there. So you're going to get a lot of your questions that are happening inside of your interactions here. The other frontline strategy would be your sales calls. If you take your sales calls and you record them, which I would recommend that you do, they can produce a transcript. That transcript has all the objections that your customer is going to ask. Then again, you create content on it and you track in the conversation because after they go through an estimate process with you, they're going to go to ChatGPT and ask it a bunch of questions and you want to be the one that shows up there. Frontline strategy is the best way to do it. All right, the second thing is Google Search Console regex hack. So essentially Google Search Console allows you to pull all of the keywords that, that are being searched and then where you show up on. So in Google Search Console, you can say, I want any keyword that's over 5 words. So anything that's 6 words or more, and I want to see what those keywords are. And this is going to show you the more conversational things that are actually happening online that people are actually searching. So pull 6+ word queries that mirror how people talk to ChatGPT. Another really good one here. The reason that I don't like this one in particular is because sometimes you can't see the full scope of things. You can only see what is actually happening in your specific situation. Where on the calls, it's literally real questions from your real customers. It doesn't have any filters from Google or anything on it, so you can capture other keywords that you wouldn't necessarily know about. For example, let's say in HVAC, someone doesn't know their exact problem that they might have, so they might not go and search it on Google, but inside of a call transcript, it might come out what their actual problem is. And then this is a real back and forth with the business that you can understand. And then the third prompt is the comparison audit. So this is what BrightLocal is starting with, and most of these platforms are starting with, which is the best service in the city, the brand versus its competitor, more high-intent conversion conversations. So I'll show you some real examples of what they pulled from their comparison audit, and I think that they're decent. But this is essentially how you want to build out these AI prompt lists. It's not just keywords, rankings, it's building out these prompt lists so that way now you can prompt it neutrally and then see what happens. And this is going to lead us right to the next thing, which is the metrics that we want to track. But here's an example: is blah blah blah blah blah good for fixing overflowing gutters on an older home in a certain area? Uh, business name, reviews, and pricing for installing seamless gutters on a two-story house. I need local gutter repair service in that can come out this week for sagging gutters before they pull. These are— so you can see these are branded, those two buttons right there. And then you can also see right here, this is more of a non-branded search for just holistically looking for things. So this is what a prompt list would look like. Now, the big 3 metrics. So what actually moves the needle and how to measure it. This is the important part of the next piece. So if you notice here, you have score, SOV, MR, mentions, like, what does this mean and how do they get this score? And then what's, what's good? That's what we're going to talk about next. So metric 1 is AI share of voice. So that SOV that you just saw, what it means is the percentage of time that your brand appears in AI response for for your prompt list. So the percentage of time that your brand appears in AI responses for your prompt list. This is why prompting is so important, because if I would have put a prompt list in there, let's say they did gutter cleaning and I put a prompt for an electrician, they're going to show up 0 times and it's going to skew the score because of course it would never show for that. Why does this matter? If AI does not know you exist, you have zero chance of a zero-click conversion. So AI has to know that you exist. So the percentage of time your brand appears. All right, so here's 3 ways to increase your SOV, share of voice. 1, entity association. So get mentioned on Chambers of Commerce, local news, BNI, industry directories, basically building out a local backlinking strategy to get linked to things. I would caution you on this, that when you talk about backlinking, this is more of a local, hands, boots on the ground, hands in the dirt type of relationship. This is not go and hire someone on Fiverr or even your marketing agency finding you backlinks because the traditional backlinks are designed for domain authority, which helps you rank higher on keywords. This is specifically talking about entity associations with local things or things in your, in your specific market that you would know about. Like if you're a duct cleaning, it would be NADCA. If you're restoration, it'd be IICRC. Like making sure that you have these entities that are referring back to you that also have authority. Number 2, volume of content. I've been saying this for a long time that more high-quality niche pages, niche content gives you more votes in the model. The more content that you give it, the more that it's going to know about your business. A lot of times marketing companies are starting from content at ChatGPT when they're marketing your business. That's wrong. What they should be starting with is your actual calls, your actual content, your actual information from your business, and then building content from there. This is why when I just mentioned about pulling your prompts from your actual calls, imagine if you take those prompts, those calls, and you turn them into actual transcript. Now you have prompt engineering plus real content that's happening. Boom. This is the whole thing with AI search. This is why I've been preaching this from the rooftops, because this is what it's all about. Number 3, local platform mentions. Get your business mentioned on Yelp, Angie's List, Thumbtack, and local directories. This is more of a traditional local SEO strategy, which is going to be building out citations. And citations are local things that are very important. So like Yelp, Angie's List, Thumbtack, and all these other local directories, Brown Book, even heck, MapQuest, like all these other local things that pull local stuff, you want them on those platforms. This is why local SEO is not like dead, it's just shifting in how you use it. So that's going to be share of voice. You want to make sure that you have a percentage of time that your brand appears in AI responses. Next metric is mention rate. How many times are you actually mentioned in these things? So what it means, of the times that you appear, so we just talked about appearing, how many times does AI explicitly cite you as the source? So this is literally just saying, "Hey, by the way, you know, you might want to use this company." This is saying, Based off of this company's thing, here's exactly what you need to know. And this is why mention rate is, in my opinion, much better than just having a share of voice. 'Cause mention rate means that there's literally something they can cite you for and they trust you. So mention without citation equals brand awareness. Citation equals brand authority. Citation drives clicks. I wanna say that again. If you just get mentioned without a citation, it's awareness. This is awareness. If you get a citation, that means that you have brand authority, and that means that your citation will drive clicks to your business. This is why creating very good content is so important. So let's talk about 3 ways to increase your mention rate. One, have specific numbers, data that AI can pull from. AI is biased towards citing quantitative evidence. So if you can create something that is bona fide, your information, your data, your stats, it will crush because this is real information that large language models don't know about. Which again is why if you were to take your actual sales calls, in your sales calls you're talking about percentages. Hey, 30% of our customers have this problem over here. Those sales calls have so much juice in them. And then same thing with your actual CSR conversation and your call assurance scripts. If your CSR is saying really smart things and talking about real stuff that's happening with percentages, with data, with statistics based off that real stuff, This is why transcripts crush. Second, definition first format. Okay, this is talking more about how you structure your content. This is a big shift when we talk about content in general. Historically, blogging has been written for the human interaction. Blogging has been content that then someone can read and then maybe you sway them your way, blah, blah, blah. No one's reading your blogs, okay? Like if you have a website about HVAC, It might get cited and then someone goes there and reads a portion of it, but they're not just going through your website reading all of your blogs. That's, you know, 10 years ago. However, what this is talking about is the definition first format. Start paragraphs with a direct 40-60 word answer to make it copy-pasteable for AI. What this means, let's say that you create content and the content is why your HVAC unit is going to go out this summer. The first thing you should say is the reason why your HVAC unit might go out this summer is because you run out of Freon, you didn't get a spring inspection, your evaporator coil's broken or whatever. Like, you want to list out those things right away. This is the most important thing because this makes it citable for AI, AIOverviews, ChatGPT. It gives us a quick soundbite to answer the question, almost like an FAQ, frequently asked question type thing. And the last one is technical schema. Using service and FAQ schema, which is what I just mentioned, it gives AI a cheat sheet for your business, making sure that you have some FAQs on your website, making sure that you have something that can link out to real questions. Because again, the shift from keyword to conversation is in conversations, people are asking questions and AI wants to show them answers. It's not just go down a list and try to find the right person to work with. It's literally conversational search. So that's the difference between search, share of voice, and mention rate. I think that makes sense. All right, and then the last one is GEO, generative engine optimization. What it means, the process of making your content consumable for LLMs, not just crawlers. This kind of takes what we just talked about, and this is a holistic way that you should be thinking about content in general. Traditional SEO is optimized for Google. GEO is optimized for reasoning engines. Two ways to increase this information gap, add unique insights, proprietary data, boots on the ground experience AI cannot find on Wikipedia. This is what I'm talking about, about creating real content from real call transcripts, real videos, real stuff. It's all about real, real, real. Large language models are trained on real. They don't want things that they already know that they can just find online. That's how you help your business stand out. And the second thing is answer target structure. So making sure that you actually lay out your H1s, H2s, H3s as answers. AI extracts this format easily. If you're just throwing things up there and you're not doing the backend technical SEO piece correctly, this is going to hurt you from being cited because you're not going to have everything lined up for it when it goes and looks at your website. So this is more of a technical thing, but you want to make sure that you're doing stuff right. If you watch this and then you just go and throw stuff up on your website and you don't actually put the right H2s as questions, H3s as answers, and you're just starting from ChatGPT's slop, it's not going to work. All right, so let's look at this all together. When we talk about prompting, We talked about what prompts are. When we talked about share of voice, this is the amount of times that it is mentioning your business or, you know, talking about your business. The mention rate is how many times it actually cites your business. And that's why I believe mention rate is the most important because this gives you the clickable citation part of it. So in this example, when we look at this, the branded ones, okay, cool, that's good. The share of voice on this specific thing is 64% of the times they're talking about us. But then 100% of the time they're referencing us to use this. Same thing with this, which is a non-branded, "I need a local gutter repair service." We're mentioned 9% of the time and they share our voice over the different things that it recommends. However, they're citing us, mentioning us to use us almost 40% of the time, which is almost half the time. So I think that that's pretty good. So overall, The BrightLocal tool gives you a roll-up of like AI visibility score and this kind of thing. I think this is cool. It's going to show you like your local pack, your organic, your maps. This is not like a BrightLocal video or anything. I'm just showing you the tools that are out there. And this is a pretty cool thing. All right. The last thing is the diagnostic matrix. And this is what the data tells you. This is the piece that I think is going to separate a lot of marketing companies in the future, which is Cool, anyone can just set up some random dashboard and like, you know, track some things and prompts, but it's really about what the data tells you and what you do with it. That's the most important piece of it. So the SOV versus mention rate matrix. So let's say that you have a high share of voice, so you're shown a lot of times, but they're not mentioning you and citing you. So AI knows that you exist, but they don't trust your content enough to cite it. Very important piece. When we looked at this, The share of voice, they know that we exist, but the mention rate being higher and almost half of the time means relatively, I would say that they trust us for this for sure. Low share of voice, low mention, you're literally invisible. So you're not being— they're not showing your name anywhere, and they're also not citing you or mentioning you to get a click. You're gone. Low share of voice, high mention. So this is the hidden gem. If you don't have a share of voice you have great authoritative content, but AI only finds it for very narrow specific prompts. So with this, you might want to expand your reach with digital PR and entity signals so that you get more reviews on external platforms. This would be like, hey, you need to get more share of voice, so you need to get more local people linking back to you. Your content is great, and clearly they're citing it and they want to— they want you to get the clicks, but they also need more things to cite from you in different places. So another thing to consider on this is like, don't just have blog posts on your website, have videos that are on all the different platforms, YouTube, Instagram. That way they have other things that they can pull in as share of voice from different places. Go on podcast, you get it. And then high share of voice, high mention. You're the dominant authority. AI treats you as a primary source of truth. Defend the moat, add new case studies monthly so AI does not shift to a faster competitor. Basically, we're trying to build in our company these market leaders, and we are trying to push every single person to start creating content because as I've talked about through this whole thing, you can see how content and prompting has a through line with what are the people actually asking, how are they searching, give them the information, build the mention rate so that way you can get cited in these platforms, and then have it come over to click and, and actually work with you. Hopefully, this was helpful. If you do not have a marketing company that's talking about these things, this is exactly what we do for our clients. I'll put a link down below if you want a home service marketing company that actually cares about AI search and the shift that's happening and wants you to make sure that you're in this market leader tab. Click the link, set up some time to talk. I love the opportunity to see if we can help you out.
More from Phil

The companies that win the next five years won't be the ones gaming the system.









