AI Search Right Now: What Marketing Leaders Need To Act On

Michael Transon 00:00 – 00:25

There was a study of a couple hundred like high stakes buying sessions with users where they were asking very high intent queries and questions in AI mode. And most of those read the answer and they just made the call right there and then on what they were going to do. And they never clicked anything.

Michael Transon 00:36 – 01:01

Hey, I'm Michael Transon. I happen to be the founder and CEO of a search agency, but I'm also a bit of a professional skeptic in an industry that sells a lot of certainty. So this is The Search Signal. Here's where we talk about what is changing in search and what it means for you. But before we get into today's episode, let's run through the news. There's three things today that I want to talk about that I think are worth your time.

Michael Transon 01:01 – 02:28

The first one is about how Claude, which is, as everybody knows, Anthropic's chat assistant, picks its sources and it's a bit of a strange one. So as you might know, Claude doesn't really show its work on citations the way that a lot of other tools do. But this week, Profound was able to dig in and what they found was something like 80% of the URLs that Claude cites comes straight out of, if you can believe it, Brave's top 10 results. If you don't know Brave, it is a smaller, kind of newer search engine. And against Google's top, Claude's citations only line up about a third of the time. And even out to, you know, the top 50 of Google's, you're only seeing about two-thirds representation. So what that means is Brave's first page is doing a lot more work here as it relates to Claude citations than Google's first five pages put together, which is pretty wild. Claude also only runs a web search in about a third of their prompts, which is crazy considering ChatGPT is doing it on something like 90%.

Michael Transon 02:29 – 04:12

So, what I would take from that if I was you is if your buyers are leaning on Claude for their answers, strangely enough, your ranking in Brave might actually matter as it relates to that. And to be frank, almost no one's looking at Brave right now. Then a couple of things came out from Google this week. First, two employees from Google, Mueller and Split, went on Google's own podcast, and they pushed back a little bit on this trend that's happening in the industry of converting your pages into markdown files specifically for AI. Split's point in the podcast was that Markdown actually strips out the structure, which is things like your navigation and your internal links that both, you know, regular search and also AI systems use to understand how a page fits in within the structure of the rest of your website. And Mueller added that these systems can already turn your standard HTML page into plain text on their own. So building a separate markdown is effectively redundant and not worth anybody's time. So if anybody that you're talking to is pitching your team on some cool or sexy markdown strategy for your website as some sort of innovative AI play, Google's basically saying it actually might work against you.

Michael Transon 04:14 – 06:50

You'd be stripping out the structure from pages that are already performant and are already ranking. And then the second one, and I'd say this is probably the most important one for this week, and it's pretty good timing because the whole second half of today's episode is really focused on this: Google actually finally gave Search Console its own report for AI. Finally. So for the last probably two years, the question has been for most people like, am I even showing up in, for example, Google's AI overviews? And there was up to this point really no clean or easy way to answer that inside of the tools that we've already been using. And now, with this update, we can actually see how our pages show up in AI overviews and how our pages show up in AI mode. So it also breaks out the numbers on their own instead of burying them in some of the regular numbers that you have typically seen inside of GSC. So the catch is that it is only impressions for now. So you can see that you showed up, but you can't see whether anyone clicked your site if it was used as a citation or as a recommendation. It's also slowly rolling out. So it is starting first in, I believe, the UK. So unless you're in the UK, you might not have it yet. And then riding along with that update is a toggle. And that toggle is going to allow you to pull your content out of AI features without touching your normal rankings. And that went live this week. So if you've ever been interested in or curious about how to actually take content out rather than trying to improve your performance in AI rankings, the only way historically to block AI overviews was to essentially try to tank your own organic search results. And I don't know why anybody would want to do that. So nobody could really use that. Now it's a lot cleaner. So I will say this pointly for most people, I would absolutely not recommend flipping that toggle and turning on that switch. I would not recommend it because pulling yourself out of where the buyers are already looking to, I don't know, make a point or whatever you might want to do that is a bit backwards, but if you care, the control exists now. So if you've got a real reason, it's there. You can take action on it. Links for all three of these items are gonna be in the show notes if you wanna read on them further. But that is the week and the week's news. So let's get into today's episode.

Michael Transon 07:04 – 08:44

Okay. So last episode we got into what moves AI visibility. And the big finding from last episode was that domain authority, which is the number that most search programs live and die by, barely predicts whether an AI platform names you or cites you. But what it does predict is how often your company gets named in other people's content out in the world and next to your category. So if you didn't catch last week's episode and you're listening to today's episode, I would definitely recommend you go back and listen to it. It's linked in the show notes, or you can pop into our podcast and grab that first episode. Cause that episode really talks about the what and the why. And then today's episode, we're going to be talking about the how. And as it relates to the how, I want to talk to you about four things today. First, I want to talk about building AI visibility and presence across multiple AI search surfaces instead of just one. Secondly, I want to talk about how to figure out where you're actually ranking and what your AI visibility is right now. And then third, I want to talk about the build itself, which is we'll talk through the actual type of work that you can be doing to earn more visibility in AI search. And then fourth, we'll land by why I think ultimately taking action on that work and actually doing it is going to be worth the effort. As always, everything that I mentioned today, all the studies, all the data is going to be linked in the show notes if you want to dig into it afterwards. With that said, let's dive into it.

Michael Transon 08:44 – 11:01

So a lot of companies, when they started getting into AI search visibility and wanting to improve their performance, started in a similar first location, which is Google AI overviews, which makes a lot of sense because you know that's the surface sitting right there in the search results that they were already tracking from traditional organic SEO. And the assumption underneath it for a lot of people was, you know, if I show up in the overviews, the AI overviews in Google, I'm probably going to show up across the other AI search surfaces as well, which might feel logical and might make sense, but that doesn't really hold up. And the reason is, you know, for example, Ahrefs did a study, I think it was at the end of last year, and the overlap between what gets cited in AI overviews and what gets cited in, for example, another Google product like AI mode, it was something like one in seven citations were showing up in both at the same time. And like I said, those are features inside two Google products, not separate companies, the same exact company. And so doing well in one of them barely carried any translation into one of the others. And so we saw the same kind of thing also in our own research. We looked at about a hundred thousand different AI search responses. And what we found was that each of the platforms that we analyzed was sourcing its answers in its own very distinct and unique way. And, you know, that changes what gets you in front of a buyer on that specific surface. So for Perplexity, for example, leaned really hard on news outlets and specialist publishers. And then, well, they also focus on a lot of stuff with, you know, data in primary sources that were really behind the information. And then we had Google AI overviews, which was tracking actually a lot closer to regular organic search results. So in many ways, but not wholly, ranking the normal way, for lack of better phrase, was a lot of the battle there.

Michael Transon 11:01 – 12:27

And then we had Copilot, which we looked at, and that was pulling a lot from places like LinkedIn and some of the B2B trade publications. And then you had ChatGPT, which, if you're not familiar with, is pretty well known—they pull a lot from Reddit, Wikipedia, and Quora. A lot of the big editorial sites get a lot of citations in ChatGPT, and it also tends to favor brand names and companies that it already seems to know in its corpus. Another company, SparkToro, ran a prompt analysis. They looked at about 3,000 prompts and they looked across the big platforms similar to what we had done. They found another very interesting thing, which was that if you run the same exact prompt twice on the same platform, there is a less than 1% chance that you are going to get the same exact set of brands back. And what that means is basically these platforms are sending you different sets of information every time you ask it the same question. So naturally, you hear about that information and you might think like, my gosh, AI visibility is like absolutely chaotic. Like, how do I even do anything about this?

Michael Transon 12:27 – 13:26

And I will tell you, I don't personally read it that way. The way I read it is just we need to stop thinking about AI visibility the same way we think about ranking for one specific keyword. The thing I'd actually watch for, and we'll talk about this throughout today, is something called share of voice, which is how often you turn up across the whole range of questions that a buyer might potentially ask, and not whether you land in one very specific answer or one very specific prompt for one very specific platform. The goal is if you are in most of them, you've built something that actually holds durability because it's not really riding on, you know, one signal or one citation or one change to the platform that could move, you know, overnight. And one day you might be showing up in a response and one day you might not be.

Michael Transon 13:26 – 14:24

So I'd also say right before Google's I.O. conference this year, Google had put out its first, I would say, official guidance on how to optimize for their AI features in search. The core of that documentation was very simple. And what they claim was that the features that run on the same ranking and quality systems as regular search. So The fundamentals that you are already doing for just traditional SEO, they essentially still apply. And they also named a few things that you can also stop doing, which I loved, which are things like—and I know a lot of people did this, no shade if you did, but it does not work—things like uploading LLM.txt files or having very AI-specific schema uploaded on the website.

Michael Transon 14:24 – 16:30

Google's claim is most of that, specifically and very practically the LLM.txt file, is just not necessary. So I also want to say, and I don't know if you can tell by the tone of my voice, Google has told everyone things many times before that just do not hold up. The release of the internal documentation from the search team in 2024, that leak made that very clear. If you have not seen that, you can go check that out. So what I would say though is I believe most of it. And the reason why I do is because it, you know, not ironically, really lands in the same place a lot of independent research already has, and lands similarly to a lot of the research that we have done on our own at Victorious, and also what we talked about in the last episode, which is getting the brand name out in the world. Getting third-party content, third-party mentions, along with a really strong entity-based architecture on your website is what predicts and drives AI visibility. And these other fly-by-night or sexy tactics or even core SEO tried-and-true metrics like domain authority, just do not do it. So what the guidance doesn't tell us is how long this AI visibility also stays free because Google does like to charge for monetizing the tools that they offer the market. We've seen that happen with Google Shopping and we've seen it happen with the local pack. So, in my opinion, the move that brands need to be considering right now is building while the space is open and while the space is free. So let's talk about how to do that.

Michael Transon 16:31 – 18:15

So before you build anything, it's really important to get a read on where you currently stand and how you currently perform. It's really simple to either check by hand or you can use a platform to do it. So if you want to do a very basic check, an N1 study, you can just open up an incognito window and you can run your queries and try to do it without signing in where you can, because all of these LLMs will personalize the responses that they give you based off of your history. And you ultimately do want to see what a cold buyer is going to be seeing when they are doing those similar searches. So most of these tools will allow an incognito mode, ChatGPT or Claude. You can click that and then do your search there. I would use and pick whichever platform you think your buyers are most likely to use. And then I'd just start by running five or so odd questions, ones that they would ask when they're, you know, weighing options in your category. These would be like commercial ones, like, you know, best thing for situation or whatever, or how much does X thing cost? I'd run each also a couple of times because like the SparkToro research study that we talked about, there is a chance that it will change. And there is a chance that just doing it one time might not give you all the information that you need to know. And what you're doing in this is building a mini share of voice number. And, you know, the goal being out of all those runs, what share of the answers are you and your brand actually showing up for?

Michael Transon 18:24 – 19:17

We also want to keep two separate things in terms of metrics in mind when we're looking at these responses. This is the same split that we talked about in the last episode: citation rate and mention rate. Getting cited is when they link or they attribute a portion of the answer to you or your website. And then brand mentions is when they're actually mentioning your brand by name in the answer and potentially as a recommendation. So the citation is what gets you high intent, theoretically clicks from LLMs, although the click-through rate is low. And then the mention rate is what builds your name with the model over time and gives you hard recommendations to your potential buyers. So they do do different jobs. So I wouldn't definitely, you know, average them together. They have their own unique purpose and value.

Michael Transon 19:17 – 20:08

And then here's the part that turns that from, you know, more of a diagnostic into a plan. So I would write down the competitors that are turning up in these answers instead of you. And then I would also write down what are the outside sources that are being used as citations in the answers that you're getting from the queries that you're using. Those citations and the sources that are being used essentially become your target list for your third-party content strategy. So, you know, doing it by hand is a, I would say, a good gut check, right? But you can't reasonably do that every single week. And it would be a lot of work to manage it with a spreadsheet and your incognito window. You will get discouraged and you will not be able to do it at any scale at all.

Michael Transon 20:08 – 22:19

So what I would say is once you're past that gut check, or if you're already past that gut check, you're gonna wanna get a tool that watches it for you. And there is like a whole category of these platforms out there right now. And you may have heard of these: Profound, Peec, and then the one that we recommend and we use, which is Scrunch. So I'll give you a quick read on each of these. They're not totally interchangeable; each of them has their own quirks and their pros and their cons. I'll start with Profound. Profound is probably the largest and best known, I would say. They're also, I would call, the enterprise option. So they've got pretty broad coverage. They've got, from what I've seen, pretty much every model, including the newest ones. They can, if you want to pay for it, basically run unlimited prompts. They've got a lot of data that you can export. So I would say if you are on or you run a big team and you have got analysts on staff who are living in dashboards and have the time to do it, it can be a very powerful tool for you to use. The flip side is it is really, really expensive, and there is a big learning curve for this specific platform. And I would say Profound leans more towards showing you data rather than telling you what to do with it. Now Peec, on the other end, is way more affordable. It's the—I would say it is the affordable one. It's pretty quick to set up, and the UI is generally clean. The other benefit is it covers a lot of different languages and a lot of different countries pretty cleanly, so it's more international. Now, the catch on this is similar to Profound: it's mostly just a monitoring tool. So it's going to tell you where you stand and it kind of just stops there. So you are on your own for the actual doing. Now, all three do share that similar monitoring and not doing, right? You need people to do the work.

Michael Transon 22:19 – 24:06

But the last one, and the one that we typically reach for, is Scrunch. And I'll talk you through a little bit about why, because, you know, you do need to take it with a grain of salt. All these companies do have a lot of comparison pages on their website about why they're the best and why everybody else sucks, so you know, everybody's—including me, you know—talking about what they think is the best option, which just happens to be themselves. But what we like about Scrunch is it goes, you know, past the score in points at the actual work that needs to be done or has been done that's been getting you to where you're at right now in AI visibility. So Scrunch oftentimes will show, you know, the specific pages and the lists that the AI is pulling from for the queries that you're wanting to look at and also shows where you're dropping out of the buyer's journey, which is, you know, exactly that target list I keep talking about, and they can help assemble that list of sites that are being cited in the queries that you are looking for rather than you assembling that by hand. So it sits kind of in the middle ground where, you know, a mid-sized company, if you're one of them and you have an agency and you've got them on staff and they can help run this stuff behind the scenes and they can act on what they see, this is a good option for you. It's a couple hundred bucks a month, so it's not the cheapest, but it also tracks a few less platforms than Profound. So if you are, you know, a huge enterprise, I would definitely look hard at Profound. And if you just want like a cheap pulse check and you don't want to do the handwritten work anymore, I'd say Peec is probably just gonna be fine for you. But for most companies that are trying to go from blind to actually moving in this space, I think Scrunch is probably where I'd start.

Michael Transon 24:07 – 25:14

Let's talk about what Victorious does. So, you know, like I said, we use Scrunch, but regardless of the tool, right? When we run this process for a specific client, we typically start by working with, you know, 50 plus questions or queries that we want to check in AI. And we'll group them by topic, and we will run these queries on a consistent basis—this could be every three days, every seven days, twice a month, once a month. It really just depends on what works for that client. And then we benchmark their performance against the competitors that they are typically losing deals to. So we'll get that information from our clients. Now if you want to just run like five or whatever, you know, a couple of times, I do think it's plenty to see where you stand, but it's important that you know that for us, that's typically how we establish our initial baseline. And we do that before we even start working with a client when we're still in the initial discovery phase of deciding if we would even want to work together.

Michael Transon 25:14 – 28:00

So if you follow this process and you've got your baseline, and let's say you did start to build your target list of websites that are third party content that are being used as citations and therefore influence the responses that buyers are seeing for questions that you want to show up in, we should talk about the actual work that closes the gap and how we can create a strategy and execute work that is building on itself. So the biggest piece, and this is the thing that honestly comes back as the strongest predictor in AI visibility—and that's proven in our research and in pretty much everybody else's research—is first and foremost getting your company named in other people's content next to your category. So I would picture what your diagnostic process turned up. Let's say, like Perplexity, for example, keeps citing the same couple of trade publications and maybe a certain, you know, best-of-category roundup on a site, and you are nowhere on those publications at all. Right. Those publications, that piece of content that is the roundup, that's now the job, right? You need to go and earn your way into it. You need to pitch the publications that run it, and you need to start getting covered in the publications and the content that the AI is reading and that your buyers are reading. So this is digital PR and just classic outreach. And it's not necessarily, although it could be helpful, but in this situation, it's not necessarily just cranking out, you know, more blog posts. And I would say this is a piece that a lot of search programs are underfunding, and it is expensive. It takes a lot of manual effort and work. The response rate can be pretty low, but it is the number one predictor of visibility and should be funded because it is the number one strongest predictor. So alongside of it, I would also say you would keep a steady, steady drip of fresh links, fresh backlinks that are pointing at specific commercial pages that you care about on your site and not just the homepage. So having a cohesive off-page third-party strategy with strong foundational backlinks plus this outreach is very important. And the second piece is also very important, which is making the machine actually certain of who you are.

Michael Transon 28:01 – 29:00

And I would say this is the one that a lot of people skip because it's a little bit unglamorous, if I'm gonna be frank about it. But for the, you know, if you're anxious and you're trying to figure out what to do about this, I would say this is also the most controllable thing on the whole list. And here's the problem that it solves: maybe there is another company that exists in the world that has a name very close to yours, but it's in a completely different industry. And when the model isn't sure which one you are, it gets cautious because it doesn't have certainty, and it might actually reach out for a competitor who it's more sure about instead. And you can fix that. So what you're doing is getting yourself more effectively integrated into both the individual LLM citation corpuses and then also Google's traditional knowledge graph, which is a structured map essentially of who and what exists in the world and what Google uses to understand what companies and what topics exist.

Michael Transon 29:00 – 32:22

And these are the types of structured knowledge bases that AI models really lean on to know that an entity is actually real, a company is actually real. So when you've got a very clean entry in the knowledge graph—and if you don't know what it is, it's that little panel that sometimes pops up next to the search results when you do a branded search on the right—when you are effectively creating a strong knowledge graph presence, the models describe you correctly and they also reach for you more effectively. And then when you don't have that, they ultimately have to guess. And if they have to guess, they don't like guessing, and they are, you know, more likely to cite a competitor than to cite you. And I would say this work's also very clear, super concrete. You're stressed out about this; I would calm down on this. It's very easy, right? At its core, I would start with one page on your website, usually I would say it's your about page, and I would treat that as the single source of truth about your company. I would do things like put complete organization schema on it—you know, your name, your logo, what you do, when you were founded. Then I'd use the sameAs property in that schema to link out to other third-party profiles to confirm that you exist. And this would be things like your LinkedIn, your X account, or your Crunchbase—whatever you've got out there that you own, third-party profiles that you own—so Google can effectively fuse them together into just one entity instead of what happens when you don't, which is a bit of a scattered list and loose mentions of your company. Then after that, I would go create a Wikidata entry for the company and I would source it properly. Wikidata is the open machine-readable database that Google leans on hardest for the graph, and it's a way lower bar than landing on a Wikipedia page. And also, the thing that might cause issues with this that I want you to get ahead of is inconsistency. So things like your company name being written four different ways across your site. Maybe you have it in standard case in one location and you have it in camel case in a different location, right? Pick one version of everything and make it match everywhere, okay? If you are a local or a service business, your Google Business Profile is going to be a very quick path to this, also doing a citation audit and a citation cleanup so that your name, your address, your phone number is consistent across the entire list of directory websites that you are currently on. You do that work, and over a couple of months, you're going to go from a name that the model wasn't sure about to one that is definitely much more confident recommending and saying out loud.

Michael Transon 32:22 – 34:04

And then the third piece I would go to is depth. Okay, that might sound general, but this is where I would start if I was you first, if you want to start this week or in your next sprint cycle. I think this is the fastest win for most brands. So the way I would do this: take your single best-performing commercial page on your website. Okay, this is one that already ranks, might have a really strong link portfolio already, and go through the process of making it more citable. What I mean by that is go section by section and rewrite each section. You can do this with your content team, or you could do this with your search agency. You can have your search agency take this work on your behalf. Rewrite each one so it opens up with a plain and complete answer to a question that a buyer would ask, and write it so it would still make sense if you just took that paragraph out and you read it on its own. I would absolutely swap any generic content or generic lines in the content for specific numbers if you have them and your own data if you have them. And then also make sure that nothing important is buried behind something that a crawler cannot reach. They will be reading the HTML, right? They will not be reading JavaScript, okay? We need to make sure that we can render this content in a way that the crawlers can actually read it and understand it. So this is a couple of hours, maybe on a page that you already own, and then once that's done, you build outward from there.

Michael Transon 34:04 – 35:07

Right, a main commercial page you've got. Then I would hope that you would also have already a network of other supporting pages on the website that are already addressing and covering specific sets of questions that buyers are asking, so that when the AI is doing the decomposition work, the query decomposition work, which we talked about in the last episode, we've already got answers for every piece and there's no gap for it to route around you to a different website or to a competitor. So one last thing before we move forward, why this is actually worth doing. I think it is, and the reason why is because AI traffic is, in my opinion, becoming maybe the most valuable and high-intent traffic that you will find on your site. It's also behaving very differently than just a typical normal SEO or organic search click. The reason is because most people in AI mode are not clicking anything at all.

Michael Transon 35:07 – 37:25

There was a study of a couple hundred high-stakes buying sessions with users where they were asking very high intent queries and questions in AI mode, and most of those users read the answer and they just made the call right there and then on what they were going to do, and they never clicked anything. You also see it in regular search too. The top result used to pull around a quarter of the clicks. Right? That was a lot. And now it's closer to like one in 10. So we're getting a lot fewer clicks in traditional search. That's not gonna change; zero-click is probably going to be here to stay. And the majority of the consideration process and the buyer's journey will continue to happen in these AI search platforms, and they will not be going to your website before they make this decision. So, what that means is the ones that you are going to get from search, the ones that are going to be coming from AI, are going to be worth a lot more because that person already did their homework. They did it inside of the platform before they came. They compared their options already. They already narrowed down what they want to do, and they're so close to buying by the time they land on your site. And I would also hold this next part a little bit loosely because the data is still, again, it's still early. But as of right now, the AI referral traffic is converting better than regular search, but also better than virtually every other platform and significantly better than what it did even just a year ago. So I can't promise that that is going to continue to be the case and that it's going to hold, but it's something that we are continuing to watch and keep an eye on. But it also does line up with what you would typically expect as we understand how people use these platforms.

Michael Transon 37:26 – 39:49

Okay, so if you're gonna do something with what we've talked about today, this is probably where I would start. So, number one, go and get your baseline of your AI search performance this week. You can do either the incognito window or you can work with a tool. If you work with a search agency, you should already have this information available to you. But if you do not, you know, incognito window, five queries your buyers would run. Run each of them a couple of times, write down who's showing up instead of you, and also write down which sources the platforms are continuing to cite. And don't forget that last part is going to be your target list, right? If you want to watch it ongoing instead of doing it by hand, or if you want to do it at a larger scale, that's what a tool like, for example, Scrunch is for. There is a fuller breakdown of the platforms and of the tools in the show notes, which you will be able to check out after the episode. The second thing, like I mentioned, make your pages citable. So I would not say publish new stuff yet. I would not stop it if you're already in the motion of it, but the focus should be on taking your best existing commercial page, rewrite it so it is answer first, use your own first-party data wherever possible, and make sure nothing is buried where LLMs and LLM bots cannot read and index the content. This is a couple of hours and it's probably the fastest way you will get to improving your AI visibility. And then number three, build the presence the diagnostics pointed you at. Okay, we need to earn our way into the lists and into the publications each of these platforms is pulling from. Okay, these are fresh links that we can get that are pointing to our commercial pages. And then I would say lock down your entity signals so the machines actually know who you are. Things like I mentioned—schema, like organization schema, sameAs schema, Wikidata entries, keeping your name consistent everywhere, cleaning up your citations—that type of work is what is going to earn us a more effective knowledge panel presence and is what a lot of models lean on when understanding what brands are about.

Michael Transon 39:50 – 41:03

Then lastly, third, build the presence that the diagnostics point you at. So we need to earn our way into the lists and into the publications that each of these platforms are pulling from. And we also need to make sure that we're continually doing a good job of building standard traditional link building to our commercial pages. And then lastly, we need to lock down our entity signals so that these LLMs and machines actually know who we are, right? I said this earlier: organization schema on the site, sameAs schema that links out to your various profiles on social media and third-party review websites, any profiles that you own, a Wikidata entry, and your name consistent everywhere. And if you're a local business, Google Business Profile optimization and making sure that your local citations are clean, accurate, and consistent. That type of stack is what is going to earn us a more consistent presence and profile in things like Google's knowledge panel, which as I said earlier is the type of system and understanding that a lot of these models lean on.

Michael Transon 41:05 – 42:01

And then I guess lastly, lastly, I would say track share of voice, which is how often you are showing up against the whole sphere of questions across all of the different models. This matters more than rank, and we need to track it regularly, at least monthly. Because as we said, one single snapshot of one search of one prompt at one time doesn't really tell us the full picture and does not tell us where the answers are moving. What you want to see is the share of answers climbing across the questions that your buyers are asking and then benchmarked against the competitors that you are ultimately competing with. Thanks for listening today. If you have found today's conversation and episode helpful, I'd recommend you subscribe to The Search Signal wherever you found this podcast episode. I will see you next week.