How To Do an AI Search Optimisation Audit (Step-by-Step Guide)

Google has just added Gemini 3 to its AI overviews and AI mode. And I've just had notification that they're having to limit its usage because it is now so popular. These AI overviews and the more detailed AI mode are fantastic. It's so clear that they are the future of search. But have you ever noticed when you run a search sometimes your brand doesn't show up? I've just searched here for the best heat pump brands and there is one fantastic brand steelron that is just not mentioned here at all. Or maybe you've been inside chat GBT just having a look for the types of things that your customers might be searching for and again you've noticed that it just doesn't recommend you. Well, what you're doing here is basically a very basic manual version of an AI search audit. You're experimenting to see when and where your brand does and doesn't show up. But there are two problems with this approach. Firstly, it's very manual. So, it takes ages and if you wanted to track visibility over time, you'd have to record in a spreadsheet to see when and where you were visible and when you weren't. The second problem is much bigger, and that's in that you don't get any answers from this. All you see is the extent of the problem, i.e., we're not showing up. That's why in today's video, we're going to show you how to do a proper AI search optimization audit. In particular, we're going to focus in two areas. Firstly, how visible you are, and secondly, your AI brand sentiment, i.e., what these AI tools actually think of your brand. And at the end, I'm going to show you how to prioritize the fixes so that you can work on the stuff that is most going to improve your visibility in tools like AI overviews. Don't be steal. So at Exposure Ninja, we do AI search optimization audits for brands of all shapes and sizes from huge multinational global brands to much smaller challenger companies. And we're going to be sharing what we look at in these audits. Now, of course, in these audits, we're going into way more detail, and they typically take our team up to 25 hours to produce, but I'm going to show you a much more simplified version so that you don't need to spend the full 25 hours. We're going to be using one of our favorite tools for doing this, which is Semrush. And you can try Semrush for yourself at exposure.ninja/semrush 1. Thanks to Semrush for giving us free access for this video. Firstly, let's talk about the difference between an SEO audit and an AI search optimization audit. In a traditional audit, you're analyzing your visibility across traditional search engines like Google. This is the 10 blue links. Usually, you're looking at really three areas. You're looking at technical SEO. So, how fast does your website load and is it easy to crawl? You're usually looking at on-site factors like content and internal links. And you're looking at your backlink profile, i.e. how many other websites are linking to you. All of these things combine to impact your visibility in traditional search. But with an AI search optimization audit, things are slightly different because yes, traditional search visibility does impact your AI search visibility. But mainly what we're looking here is at your total visibility across AI tools and the sentiment. With an AI search optimization audit, we're looking at a slightly different set of things. We want to answer questions like when your potential customers ask AI for recommendations. Do these tools and the large language models underneath them really understand what you do, what you serve and your benefits and USPS. How do these AI tools understand where you fit against your competitors and do they put you forward or your competitors forward and what sources i.e. what websites are they pulling from in order to produce these responses. Now, you can do all of this inside the AI visibility toolkit inside Seamrush, which is what we're looking at here. So, the first step that you're going to do is put your URL in. We're looking at the Mayo Clinic today. That's an example of a brand that has pretty good visibility across both traditional search and AI search. When we stick the website in here, we get an initial dashboard. This is the overview section, which basically highlights where we might want to spend a bit more time digging in. We can see this visibility score, which is a score. how visible we are in the responses that Seamrush has indexed. Where does this data come from? Well, Seamrush analyzes responses in AI overviews in the special AI mode that Google runs as well as chat GBT and Gemini is also coming soon. It sees the prompts that people are putting in and it sees the responses that are generated. And this visibility score measures how visible this brand is for the relevant prompts that Samrush indexes. So 71 out of 100 is pretty good, but there is definite room for improvement. Next to that, we see relative visibility on different platforms and how this has changed over time. The orange line is AI overviews. This is Google's AI responses inside the search results. The green line is chat GPT. The pink line is AI mode, which is the dedicated AI response from Google. And then the blue line is the average. Next up, we've got some stats. The monthly audience is the number of prompts where your brand is seen. We've got the number of mentions. So, this is the number of mentions across all of the prompts and responses that Seamrush has in its database. And then we've got cited pages, which is how many pages on your website are being cited in AI answers. Straight away from this, you can get a picture of how visible you are against your competitors. And this is much more reliable than just running some searches yourself and seeing if you show up. And that's all well and good, but knowing that you're showing up is one thing. Knowing which topics you're showing up for is quite another because there will be topics that your brand is most keen on having visibility for. And in this AI visibility overview, we can dive into a bit more detail here. If we scroll down, you can see your performing topics. These are the ones that your brand is being mentioned most frequently. And in the Mayo Clinic's case, because it has such good brand recognition, actually, a lot of these topics are about the Mayo Clinic. So these are branded searches that we'd expect the Mayo Clinic to show up for anyway. So in their case and maybe in your case, what's more interesting is the topic opportunities. These are topics where your competitors are showing up but you are not. This can highlight content gaps, things that you might need to talk about on your website so that you've got content ranking for these topics and you've got a better chance of showing up. Let's say for example that Mayo Clinic wanted to improve visibility for antihistamine and histamine receptor type topics. At the moment they've got competitors being mentioned but they're not being mentioned so often. Well, here we can see some of the prompts that are showing competitors but not showing Mayo Clinic. Therefore, the opportunity is producing some content or being seen in other people's content that is getting featured for these topics and making sure that your brand is being mentioned. that increases the chance that the AI overviews or AI mode or chat GBT is going to pull through your brand and recommend it to customers. There is another way to win in AI overviews and AI mode. If we have a look at this example again, I've searched for best heat pump brands and you can see here that we've got some recommendations of brands and we've got these websites over on the right hand side. These are the websites that this answer is pulling from or using to back up its response. You can win by being one of those websites because they are going to get clicks. There are other sites being featured throughout this response as well. And these are known as sources. So sources are really important in AI answers for two reasons. Firstly, because you can get people clicking on them and coming to your website and then you've got a chance of converting them. But secondly, if you've got content on these pages and that content is mentioning your brand, there's a chance that you can influence the AI answer. if that content is being cited as a source. And this is something that we at Exposure Ninja have been doing for three years now. If you can find content or produce content that gets cited as a source, sometimes you can influence the response that is produced by seeding your brand in that content. So where does that show up inside the Semrush AI visibility overview? Well, it's this section here, cited sources. So, these are all of the websites that are being cited in the relevant AI answers which are featuring or mentioning your brand. There's really good news for the Mayo Clinic here because they've got an unbelievable source profile. Some of these sites are incredible high traffic, very high relevance. You can see the number of URLs that are being cited. You can see the number of prompts that this website is being seen in and the number of people visiting this website every month. So these are huge websites that are providing lots and lots of sources for these AI responses. So great to be featured in these types of sites. But on the other side of things, you have source opportunities. These are again websites that are being featured in the AI answers but that don't include your brand. So if you want to improve your brand's visibility further, getting featured in some of these sites can be your digital PR opportunity. Now, in Mayo Clinic's case, there's not actually that much actionable stuff here because a lot of these websites aren't really relevant for the types of topics that it's targeting. But if your visibility is lower than Mayo Clinics, this is likely to be one of the areas that is most important for you and gives you a bit of a hit list for your digital PR efforts to try and get yourself featured in these websites. And by the way, while you're doing your audit, you may end up finding yourself snooping around, see what your competitors are doing as well. You can see all of this data for your competitors. If you just go to the competitor research tab, you'll see exactly the same stuff. You'll see their visibility. You'll see uh their topics and prompts, the ones that they're showing up for. You'll see their sources as well. Sometimes this can give you a really good indication of a competitor's digital PR strategy. If you can see they've been featured on a lot of particular websites, you can then go and reverse engineer what they've done to get those features and maybe deploy that yourself. So, at this point, you'll know exactly where you stand in your market. You'll know how visible you are against your competitors, and you'll know the topics that you're doing well in, the topics that you're maybe doing less well in, the sources that you are being mentioned in, and the sources that you are not being mentioned in. If you want to help with your brand's visibility in AI search, that is exactly what the team here at Exposure Ninja does for our clients. We get them mentioned and mentioned more often and mentioned more favorably in tools like chat GBT perplexity and of course Google's AI overviews and AI mode. The first step if you're interested in considering working with us is to request a free review of your current visibility. We'll take a look at your visibility in all of these AI platforms as well as in other places like your paid search, your paid social. We'll also take a look at your website's conversion rate because it's no good being recommended and cited in these AI platforms if when someone lands on your website, they don't turn into a lead or customer. So, we'll see if there are any improvements that can be made there as well. All of this takes us a fair amount of time. We then turn all of this research into a video which we'll send to you usually within two to three working days. You might be thinking, Tim, this sounds crazy expensive. Hit me. I'm ready. Well, it's actually free. But the catch is not everybody is eligible. So, you do need to apply for this at exposurinja.com/re. That's exposioninja.com/re. But to take this to the next level, we need to start looking at sentiment because it's not just enough to be featured in AI answers. You actually want to be recommended and recommended for the right reasons. And this is where we start to think about brand sentiment, i.e. what the AI tools know about you and what they push forward to their customers. Let's take another look at the AI mode answer for this question about the heat pump brands. What's happened here is the AI tool has recommended certain brands and it's pulled out certain features and benefits of their ranges and it's presenting them to the audience. Now, if I'm in the market to buy a heat pump, I'm looking at this and it's starting to influence my selection choices immediately. Depending on what's most important to me, some of these features may resonate more than others. But the brands don't have direct control over what's being spoken about. And there are some issues here. Let me show you an example. Let's say for example that I'm DIN. Now this answer is reasonably good. But there is one problem. It's talking here about the R32 refrigerant. Now if you're really into heat pumps, you'll know actually you probably want an R290 refrigerant because it has lower global warming potential. So, if you're really into this, you'll see that and go, "Ah, maybe Dacin heat pumps aren't that advanced." And you might go for something else. If I'm DIN, I want to influence this. I want to make sure that that is correct because Dacin does have R290 heat pumps. As your customers have longer conversations with the AI tools, AI tools will bring out all sorts of information about your products and services. Some of it may be very true. Some of it might be sort of true. And some of it might not be true at all. But before we can start to correct any of this, the first step is to understand exactly what these AI tools think of your brand right now. And to do that, we need to look at brand performance. So here we are analyzing the brand performance of rocket mortgage.com, a mortgage site in the US that we've spent a lot of time looking at. We really like looking at this brand because it's doing some things really well and it's also leaving some money on the table. In the AI visibility tool, you're just going to head over to brand performance and the tool is going to compare your sentiment against your competitors. So, take a look at this. Rocket Mortgage has 75.56 positive sentiment versus some of its competitors. That is an amazing head start. We've got this big chart over here where we can see the sentiment score against the share of voice. And we can see that actually amongst its direct competitors, Rocket Mortgage has both the highest sentiment and the highest share of voice percentage. So that's really good news. Inside this brand performance overview, we get a bit more info about why your brand is being recommended, what the AI tools think is better about your brand against competitors, as well as some individual brand versus brand shootouts, which can be quite entertaining. If we want to dig in a level deeper and understand why the brand sentiment is the way it is, then we head down to perception. Okay, so we've got an overview of sentiment, but we can go two levels deeper. Firstly, as well as looking at general sentiment, we can start to look at particular sentiment strengths and weaknesses. And then beneath that, we can go into the drivers of that sentiment. So once we know what the sentiment is, we can see exactly where those positive and negative sentiment elements are. And then we can see why is all of this happening. So let's go to perception. Have a look at this sentiment breakdown in a bit more detail. If you want, we can split this out by platform. So rather than AI mode, we can choose perplexity for example. We covered in a recent video how these different AI platforms like to site from different types of websites. So if your visibility is better, for example, on userenerated content sites like Reddit, Quora, and YouTube, you can often find that your visibility is better on perplexity for example. Whereas if you have really strong presence on your own website, that may impact your sentiment on Google's products. So you need to make sure you're looking at different platforms if, for example, Chat GPT is a priority for you because it's going to pull from a different range of websites and therefore it's going to have a different understanding of your brand. My favorite piece though, and this is why I think AI search optimization is not just a marketing tool, but also potentially a business strategy tool, is all of this stuff down here. This is essentially customer feedback. This is understanding where people and AI tools and website content sees your brand as being really good versus things that you have to improve. Some of these things like aggressive or pushy marketing or follow-up, less transparent upfront about rates and fees, need to start application to see full pricing. These are things that you as the marketer can then feed back to the rest of the business to improve. We had a client in exactly this situation. I can't name them because we've signed an NDA, but consistently in their AI sentiment analysis, they were labeled as overpriced. Now, they actually weren't overpriced, but it was fair that their pricing could maybe have been a bit more transparent. We fed this back to the team and they restructured their pricing to be clearer about what people were getting at different stages of their price points. Within weeks, the AI responses updated. That negative label disappeared, not just because we'd changed some copy on the website, but because they'd addressed the underlying issue that was leading to that sentiment. And that's one of the great things about AI search. When you make changes, the changes are reflected in the AI results pretty quickly because all of these responses are being generated in real time. They're not being served up from some saved results. These AI tools are scanning the web and producing original answers each time. So, if you change the data, you will change the response. If you find all of this stuff overwhelming, there are some amazing sections at the bottom, the AI strategic opportunities. And these are broken into short, medium, and long-term time frames so you can work out which things you need to prioritize to generate the best improvement in perception. And by the way, if you want to get a demo of this tool, it is unbelievable. You can do that at exposure.ninja/i-envisisibility. That's exposure.ninja/i-envisibility. And if you want to go a level deeper than that, you can actually look at the narrative drivers. These are the particular things that you can do to shape the narrative in the AI responses about your brand. This is awesome. What you get in this section is specific recommendations of your positioning that you can change to improve perception. Take a look at some of these recommendations. So, converting FHA, e-closing, and pre-approval content into rule books that AI sites. This will position Rocket as a benchmark over cheap arrivals. You can then just take this to your content team and say, "Here we go. Let's just do this." And in Rocket's case, it even gives you recommendations of specific community programs that Rocket could leverage higher in order to improve their visibility on social impact queries. This is madness. If we go down to the bottom of the page again, we get these AI strategic opportunities about these sentiment drivers. And you can see we've got a bit more detail on this Detroit community work here. It's giving very specific recommendations because it's seen a sentiment gap that could be closed by taking this approach. So, by now you've got a bunch of suggestions coming out of the Semrush tool that you want to fix. There's loads and loads of items on your to-do list. How do you know which ones to prioritize? Well, here is the framework that we use at Exposure Ninja when we're prioritizing fixes for our clients. Firstly, we'll try and order things by impact and effort. Like if something is really high impact, low effort, then obviously that's a priority cuz that's a fix that we can make very quickly. But beyond that, if we had to rank common AI visibility issues, we usually start with any brand, product or service inaccuracies. If these AI platforms are misrepresenting who you are, what your product or service does, or who you serve, those are things that we would typically prioritize immediately because in all likelihood, you could be losing sales right now based on that. Second, if you see a lot of negative or neutral brand sentiment perhaps around pricing or customer service, again, we would typically treat that with quite a high priority because that could be suppressing recommendations in AI tools. After that, we get into the nice stuff. If there are missing mentions or missing citations, i.e. you're not showing up in some of the searches that you should be or your website isn't being cited when it really should be, then we start to look at that stuff. And that can be about improving your on-site and off-site authority signals. We'll sometimes find that a client's blog or knowledge base just isn't being picked up by AI tools at all. And in those cases, fixing that either weak topical authority or fixing those content gaps becomes the next priority. Another very common thing, particularly with some smaller brands, is that they just have very low visibility on third-party websites, i.e. they're just not being recommended in some of the publications that are being used as sources. And in those cases, a structured digital PR campaign can usually resolve that, although it does take time. So, there you go. You've just learned how to do an AI search optimization audit using Semrush's AI visibility tools. The next step is to then build out a full AI search optimization strategy. And that's exactly what we break down in this video. So go and check that one out next. Until next time, see you