Latest / The Search Signal / How to Build a Brand Entity for AI Visibility

Transcript

How to Build a Brand Entity for AI Visibility · The Search Signal

• michael: One of the worst strategies you can take at this moment is producing as much content as possible. The fact is the larger the share of the content on your website that isn't specifically about what you do, the more confused AI systems get about what you do. Hey, welcome back to The Search Signal. I'm Michael Transon. I'm the founder and CEO of a search marketing agency called Victorious. And here at The Search Signal, our goal is to help marketers just understand and take action on what is happening in the quickly evolving world of search marketing. So we've been running search campaigns ourselves for nearly 13 years. We work with some of the biggest and also the fastest growing brands that are out there. And we bring what we learn from those experiences with those clients, plus a lot of our own very interesting first party research to The Search Signal to help you with your own work. So, over the last four or so odd episodes, we have been building a full funnel AI search strategy from the foundation all of the way through to measurement. And then today, I want to talk about something that all four of those past episodes really depend on, which is something called your brand entity, which means the way that these AI systems understand who your company is and then also what it should be known for. So, we're going to spend today connecting the work that needs to be done for brand entity to the brand building that marketers have always done. And then we're going to spend a lot of today's episode on some practical work that we can do both on our own website and then also on other people's websites. So, before we do get into today's episode, let's talk about what happened this week in the world of search marketing news. Okay, let's start with ComScore. They are an audience measurement company and they had released their latest AI report this week. I think it was on September 22nd. It showed that buyers are spreading their questions into more and more AI assistants than they were at the start of this year. So per their report, ChatGPT still handles about half of the prompts that they measure, but that is down from about 70% earlier this year in January. And most of that share they report, went to two different competitors, Gemini and Claude. So Gemini, Claude, ChatGPT, each of these assistants, they read the web in their own way, and they also rely on and lean on a lot of their own sources for their answers. So the takeaway for this is the way one of these is describing your company can differ from what another one might be saying. And then the next update this week that we'll talk about is Google is testing links inside of AI overviews that are sending people to an AI mode instead of to a website. So an SEO had spotted a couple of links at the bottom of an AI overview this past week. And it opened up a follow-up conversation in AI mode. Google has not commented on this, so not sure personally if this is gonna stick around or not. But Google did confirm something similar in Discover, which is the newsfeed product that you can get on your phone. In that, Google's testing something called a Dive Deeper button. And from what we can see, that opens up an AI-written overview of whatever topic you're looking at. And then also has links to some related stories. And then also, I think they use some community reactions. But the point being is that both of these updates are keeping users inside of Google's answer and product for much longer. And then the third update this week to talk about is with Google. They have started to email business owners when their systems detect a spike in spam reviews on a business's GBP, their business profile. The email from what we see says Google is going to remove the reviews so they don't affect the rating of the business. And then Google is going to pause new reviews on the profile for a couple of days while it deals with the spike. So, point being is your reviews are one of the places that these AI systems do go to find out what company is like. So if you get a burst of fake reviews, whether it's positive or negative, is going to change what they read about you. So I picked these three news updates this week on purpose because each one of them is about how much your buyer's opinion of your company is now being formed inside of an AI answer. And that answer pulls from your own website, but also from what everybody else says about you. And that's what the rest of this episode is about today. Okay, let's dive in. I actually want to start talking about brand building. And I want to talk about how most marketing budgets have treated brand building. So for the last, I don't know, 10, 15 years, the marketing that got funded was often the marketing that you could attribute, where you could like, you know, draw a proverbial or real line from the dollar that you spent on the marketing to the revenue that actually came back. Brand work really never fit into that very well because the payoff of a strong brand is that people already know who you are by the time that they're ready to buy. And there's really no click that you can track for that. So a lot of marketers got to the point where brand was really the hardest line item to defend in front of their CEO or their CFO or their board. And as a result, it either gets cut first, or it might get folded into something else that can be better measured. But if you've been spending, years judging your marketing by what you can attribute, you would probably expect a lot of your traffic to go away suddenly if you decided to cut back on the performance marketing and paying for clicks. And one company is a great example of this, Airbnb, where that did not happen for them. So when the pandemic hit in 2020, Airbnb, if you don't know this, cut its marketing spend by almost half, I think. And it was a huge cut. And most of that cut came from performance marketing. I looked into this. So on the earnings call in early 2021-ish, Airbnb's CEO said that their traffic had come back to where it was in 2019. Before they had started spending on marketing again, and almost all of their traffic was coming in via either direct or via unpaid, which I looked into Airbnb's annual report and found that that includes both brand marketing and search engine optimization SEO. So the pandemic did play a big part in that. And since then Airbnb has grown its marketing budget back past to where it was in 2019. So it is just one company and kind of a funky year, but I still think it shows how much a company's demand can come from people who already know who it is. Now, here's the point I'm trying to make. Getting named in an AI answer can create that same kind of demand because a buyer who hears your company name in an AI system is going to come to your website already knowing a lot about who you are. Building a brand entity is the same exact work that marketers have always been doing to build a brand. You just decide and work through what your company is best at, and then you work to describe it in the same way everywhere the company can show up. And then as a result, you can earn coverage on websites and in publications that you don't control, so that other people are repeating that same brand description and information. And when I say brand entity, I mean, as it relates to AI search, your company as a thing that these systems can identify, and then facts that are attached to it, like what you sell, who you sell to, where you're based, who runs the company, right? The way that I think about it is the who and the what, where the brand entity is the who, meaning your company, and the topical entities are the what, which means like the subjects that you ultimately want people to connect to your company. Like, for example, if you do payroll software or you sell commercial insurance or whatever it is that you ultimately sell, right? But what you're building is the connection between the two so that when one of those subjects comes up, your company name is likely to come up with it. So when somebody is asking ChatGPT which company that they should work with for commercial insurance or which payroll software might make sense for a company with a couple hundred employees. You really want your company named in that answer. And we've said this throughout the whole series that mentions of your company on other people's websites are the main driver of whether you're going to get named. But the work that comes before these mentions is getting your own properties to say clearly what you want to be known for, so that when other sites ultimately mention you, they are reinforcing a description that your own website has already stated. Okay, so now that we've established that brand entity work is the same brand building you have always worked on, and that you should be doing it so that your company gets named when a buyer is asking you know an AI system which company to work with. We need to talk about what has changed and how your buyers are finding companies. And then we also need to talk about how these systems decide themselves, what your company actually is. So the fact is, your buyers have already moved a lot of their research into these AI tools. G2 is a software review site, if you're not familiar with it. They did a survey of a bunch of B2B software buyers this past spring, and about half of them had said that they now start their research with an AI chatbot more often than with Google. And many of them had bought from a company they'd never heard of until a chatbot had named that company. So, point being, for a growing share of your buyers, an AI answer is now where oftentimes they first hear your company's name. And before AI, a buyer formed their opinion of your company over weeks and months. They, might have seen an ad. Or they could hear your name from somebody at another company. A lot of them might read a couple of reviews. Eventually, they do make their way to your website, and somewhere in there they decided and made a formed opinion about what they thought of you. What's different is the AI systems now do a shortened version of that same process, but they do it on the buyer's behalf. It reads your website, it reads what other websites are saying about you. It includes your listings and it includes all your reviews. And then it hands the buyer just one clear summary. And a lot of buyers do go forward with that summary. Now the system builds that summary from whatever it can possibly find, including, for example, random pages on the internet that might even only mention your name in passing. I found a really interesting piece of information earlier this week, about a company called I think it's called Wolf River Electric, and they do solar installation in Minnesota. What they had found was that Google's AI overviews was telling people who searched for the company that it was being sued by the state's AG for deceptive sales practices. Now, the AG had sued for solar lending companies. And one of the pages the overview cited was a newspaper story about that lawsuit that in passing mentioned Wolf River near the end and did not say that Wolf River was even a part of the case. But AI overviews said it did. And Wolf River says that customers had started to cancel literal contracts because of this. I think there was one that was like 150 grand that they lost. And as a result, that company is now suing Google. Now, again, you're more likely, as your own company, probably run into a much smaller or less intense version of this. Oftentimes, more most frequently, it's where your website might describe your company one way, and there might be a directory listing from you know, maybe a couple of years ago that you forgot about that still describes a service that you might have stopped offering. But the point is the system has to decide which version is going to be correct, and it might pick the old directory. Or it might also combine everything that it's read into a single description that could be wrong. The system uses whatever description it ends up with when a buyer asks which company they should work with. So your company could get left out of that answer or described in a way that doesn't actually match who you are today. Now that's how these systems put together one description of your company from all of the things that they read on the internet. And then then they use that description to describe whether that they should actually name you. But let's now talk about whether your own team, your own company could even give that description themselves in the first place. I think we assume that companies know what their brand is before they go out and they build it. And the fact is a lot of them do not. You cannot create a clear identity for your brand if you don't have a clear idea of what your brand is yourself, meaning what you're good at and what you do. And if you cannot answer these two questions clearly inside your own company, it's going to be very hard to get anybody else, human or machine, to answer them for you. Most teams can list what they sell, but a lot fewer teams can say in a sentence or two what they do and how it's better than the alternatives that are out there. So at Victorious, we use something called the hedgehog concept for this, which is it comes from a Jim Collins book called Good to Great. And it was published, I think 25 years ago. I think it was 2001. And that book asks three questions, and the answers get drawn as three overlapping circles. And the first question is: what can you be the best in the world at? And The author Collins is very specific that he means the best in the world, which is a much higher bar than just something that you are like just good at. The second question is: what drives your economic engine? Meaning the work that makes you money in a way that you can repeat and also scale and grow. And then the third question is, what are you deeply passionate about? Which means the work that the people in your company care the most about doing really, really well. So the subjects that are inside of all those three circles are the subjects that you want to be known for. And for most companies, this ends up being a handful of specific subjects. So let's use like a made-up example to show what that handful of subjects might look like for even your company. So let's make a business up. Let's say we run marketing for an insurance company and let's say we sell to mid-size and large businesses, right? Our list might be cyber liability coverage for maybe manufacturing companies or workers comp and how claims get managed, DNO coverage for the company board, maybe commercial property coverages if like let's just say you're a local business with a lot of locations. because you know your your underwriters are some of the best in the market at those and they make up the most of your revenue. And they're the lines that your team also cares about doing very, very well. Once you've got a list like that, every page you write and every publication that you are pitching should connect back to one of those subjects because those are the subjects that you want your company named for when a buyer asks an AI system, which insurer they should go with. So most of what you do with that list is work your marketing team actually already knows how to do. You might want to do boilerplate content at the bottom of your press releases. So when it gets republished on a bunch of sites, it describes your company in the same words and ways that your homepage uses. And maybe you've been working on building that consistent description for a long time. Or let's just say the PR placements that you're working on to help get your CEO quoted in a trade pub are next to some of the subjects that are the same kind of mentions these AI systems learn from. And whoever keeps your listing accurate is maintaining the pages that these systems read. So those practices can stay the same and you can point all of them at the specific short list of subjects you pick which fit inside your hedgehog concept. So now that you've got a short list of subjects that your company wants to be known for, your brand, let's also spend a little bit of time flipping the script and talking about what to stop doing. And one of the first things I want to talk about is publishing content about the subjects that aren't on that list. So, fact, we live in an era right now where pretty much anything, almost anything, can be done and done cheaply and infinitely. And nearly everything I read about AI search is a big long list of things you should start doing and add. What I don't hear a lot of talk about is what you should stop doing. Collins also wrote about this in Good to Great 2. He had found that the companies that went from the title of the book, Good to Great, kept something called stop doing lists and used them as much as they had used to-do lists. So one of his examples, I believe, was Kimberly Clark, which is, I think in the early 70s, they had made most of their money from paper. And their CEO had concluded that the company was never going to be the best in the world at coated paper. So he decided to sell the company's coated paper mills, including the one in the town Kimberly, that the company was actually named after, and put all of that money into consumer paper products, brands you might know like Kleenex and Huggies. So 25 years later. Kimberly Clark was beating their competitors like Procter and Gamble in most of the product categories where the two of them had competed. For many companies out there today, content about subjects that they don't sell and they don't believe that they can be the best in the world at belong on the stop doing list. Now the flip side of that, it's oftentimes the content that they least want to give up. And I talk about this with prospects a lot. And what I tell them right now is One of the worst strategies you can take at this moment is producing as much content as possible. The fact is the larger the share of the content on your website that isn't specifically about what you do, the more confused AI systems get about what you do. So a system can read your website and see you write about 50 different subjects. And then it looks for evidence that you offer those things. And guess what? It can't find anything anywhere else on your website. So what is it left asking? Do they do this or do they not do this? And if they don't do this, why are they writing about it? And if the system isn't sure what you do, it might actually leave you out when a buyer is asking which company to work with for one of the things that you actually do sell. Now, the closest thing I've found to evidence for that type of confusion is Google's literal own guidance on helpful content, which asks site owners whether their website has a primary purpose or focus and whether they're producing lots of content on many different topics in hopes that some of it might actually perform well. And Google also says its usual search advice, that same advice still applies to things like AI Overviews and AI mode. So whether your site has a clear focus is part of how Google literally picks the sites that it uses in those answers in AI mode, Overviews, and traditional. So companies that produce outside of their subjects, they are usually gonna do it for traffic. And at a couple of very well-known companies, this has been going on for years. And the pages about the subjects that they don't sell have lost a ton of that traffic. So HubSpot, in my opinion, is the biggest example of this. And, you know, for years, HubSpot published about pretty much everything. And it grew their organic traffic enormously with people who were never gonna buy HubSpot. And there was an independent analysis that I found of their traffic estimates. And it said that the blog went from about three quarters of HubSpot's total organic search traffic at the start of 2024 to under half of it by the end of that same year. And the biggest losses were on pages, this is funny, like famous quotes and how to write the shrug emoji. I saw HubSpot CMO had said that searches where nobody clicks have taken a big share of their blog traffic, and that their performance on buying intent searches are as strong as ever. Now, I think both of those things are probably true. And the pages that fell hard were still the ones about subjects that HubSpot didn't sell. But there was another one, ClickUp. If you have not heard of them, they make project management software. They went through the same thing more recently. I looked at this, they had pages ranking for things like WhatsApp status updates and resume templates, and found another independent analysis of those traffic estimates, and those pages lost almost all of their traffic before the start of 2025 and the spring. And so did the blog as a whole. Okay, so point being why I bring this up, if you're deciding whether to publish on subjects that you don't sell, you would be not generating any traffic. And in these two examples, the traffic kind of went away on its own. So at Victorious, the way that we think about this, the answer to our hedgehog question has always been about search, search marketing. And for 10 years, we have said no to publishing content about anything on our website outside of SEO. I dare you to go try and find some. I think that consistency is a big part of why we get named consistently in a lot of AI answers when there are thousands and thousands of other marketing agencies out there that say they do SEO and AI search optimization work. And a lot of them have published way more content than we have ever published. So what I would recommend at this point is very clear, publish only on the subjects within your hedgehog concept on your list of things that you actually do. That still can include the top of the funnel content that we talked about in episode 11, as long as each piece is about one of your literal subjects. And for most teams, the content budget can really stay the same and just get more pointed at fewer subjects. And we can go into a lot more depth on each of them. If we go back to the insurance example that we are we were talking about earlier, a guide on, I don't know, cybersecurity limits, for example, for manufacturers or how workers' comp might get managed. Those can be inside the list because those are actual policies in our example that we write. A guide to writing a business plan or to planning for your own retirement, those would be things that are outside of it. No matter how much search volume those subjects might have, because we don't sell anything related to them. So now that we've covered why publishing outside of your hedgehog concept as it relates to your brand works against you, let's talk about how you audit what is actually already on your website, things that you've already published that may or may not fit this requirement. So you can run this audit with a spreadsheet. You could run it with tools that you probably already have. We can start by exporting every page on our website that we know is indexable in Google's search results and is accessible by AI bots, which you can get from, you know, you can use a site crawling tool like Screaming Frog, or if you want, you can go into Search Console and then you can add two columns next to each URL, one for the clicks that the page got from Google over the last 12 months, which came from GSC, and then one for conversions that it drove over the same 12 months. You can get that from Google Analytics. Once we have put together, that list, I would strongly recommend we go in manually, use a real human being. You can use AI if you want, but I would recommend a human do this. Tag every one of those pages as either on subject or off-subject against that short list we built with the hedgehog concept. And then at that point, you can see how much of your site is about what you actually do and sell, and then how much of your traffic and your conversions come from each of those groups. So I do want to set an expectation too. Some of the off-subject pages might be bringing in a lot of traffic. And a page about a subject that you don't sell can actually be one of your best performing articles. And still make your company harder to describe. So there might be a reality where you need to consider deleting a bunch of content on your website. And I get it, a lot of teams would be hesitant to say it lightly to do that. Because you know, they've put a lot of time and they've put a lot of money into building those pages, and they rightfully will be worried that their traffic will drop. I saw an example of IBM had deleted close to 80% of the marketing pages on their website, IBM.com, in 2022 and 2023, to try and make a very complicated website a lot simpler. And that person who led that project had said online that the traffic and the conversions both went up afterwards. So deleting a large share of your website doesn't necessarily mean you are going to hurt your overall traffic. So when we do this for our clients, most pages end up in one of four groups. And where we keep the pages that get traffic and are about one of your subjects and then combine pages that compete with each other for the same topic into one page and redirect the others to it. Another group would be remove pages that have been live for maybe a year or more that have almost no traffic and no connection to anything you sell, and then improving pages that have potential, but are maybe underperforming. So if your site's new or you're in the middle of scaling it up, I would say most of the pruning may not apply to you yet because there's a good chance your pages just haven't been live long enough for them to be judged fairly. But let's talk about this. For the off-subject pages that get the most traffic, I would strongly, strongly recommend that you work with somebody who has gone through the process of auditing and pruning content before you make any of these decisions. So one of the worst things that you can do for your own search performance is getting rid of content by accident that's actually valuable and helpful for your buyers. And then the other risk, though, is that you might be trying to fix a problem that might not exist, or you're fixing a real problem in the wrong way. CNET is a really good example of fixing a problem that doesn't exist. So in 2023, that company deleted thousands of old articles, and they did that believing that Google rewards sites that look fresh. And if you're not familiar with Danny Sullivan, he speaks very publicly for Google Search. He responded to that by saying deleting old content because you think Google doesn't like old content is quote, in his words, not a thing. So I'll say this before you delete a page. You should be able to say what the problem that deleting it fixes. Like a page about a subject that you don't sell is a good example. That makes it harder for these AI systems, to understand what your company does. Now, while you have the inventory open, you should look closely at the pages that describe your company directly. The home page is probably the most important one because when an AI system needs to know who a company is, the homepage is probably the first page it's going to read, let's be honest. And it also might be the only one that it reads. So the homepage should say in very clear words who you are, what you sell, who you sell it to, and if you're local where you do business. A lot of the home pages I look at say way less than that, and they lead with some catchy cool tagline that could belong to any other company in any other category. And AI systems are going to struggle to understand what that brand is about when it's going to that one specific page on the site. Now, the about page, let's shift. The about page or the company page, if you have them, should carry the facts that a system would also want to know about any company. These could be things like when you were founded and who founded it. Do you have an HQ? Where is it? What industry are you in? Obviously the services and the products that you offer. The places if you're localized where you serve, I would say the kinds of customers that you typically work with. And then if you have any awards or certifications that you've earned, they go on there as well. I would also say separately on a different page or linked to it, case studies are also very helpful because case studies also show these AI systems something important, which is the kind of customer that you serve and the results that you got for them. And I would also say last but maybe not least, if you've got a team page which can help connect the people that are on your team and their credentials to your company's name. So beyond those pages that I just listed there, every page an AI system visits is a chance to reinforce your brand. So, I'll give you an example. When I talk to somebody at Victorious who's getting ready to give a talk somewhere, I always tell them the same thing, which is that. Everybody who listens to someone speak walks away feeling a little more positive or a little more negative than when they walked in. And it is pretty close to dang impossible to walk away feeling neutral. So I think about AI systems visiting our websites the same way. Every page that they read is either going to add to what they understand about, for example, us Victorious, or it doesn't. And a page that never mentions anything about the brand is going to leave the system an opportunity to infer and guess. So I would say the simplest version of this, let's think about your blog. I would put a short block at the end of every blog post. If you've got a newsletter sign up at the bottom, maybe next to that. That says who your company is and what it does. Very simple, right? For us at Victorious, it might be something like. Victorious is a search agency that helps companies get found in Google and an AI search. And then maybe like an extra line on what we've been recognized for some of our awards that we've won. So the point being is the system is able to read one article when it goes to your site on its own and can learn about your brand and who published it. And if you have accolades in that block, it will give the system a reason to treat the article as coming from a credible source. So your pages about your subjects could have the same blocks, like your topic pages, because you know, every page about, for example, in our insurance example, commercial insurance that clearly comes from your company is just another piece of evidence that connects your company name to commercial insurance. And when a buyer is asking about one of your subjects, the system reads that page and they might never see your home page at all. So they might be missing the context that would be helpful to have when they are looking for that specific service that we might offer in relation to who our brand is and whether we should be recommended. And then after that, we want to make sure that our company's identity is actually machine readable and indexable. So we talked about this in episode 10. I will link to that episode in the show notes so you can go listen to that. A couple other things that are important is first organizational schema, which if you're not familiar, just a very small block of code on our site that labels our company's details in a structured format, like your name, your logo, your address, your founding date. So the machine knows exactly what each one of those things is. There's also inside of that schema a property called "same as", which can list your company's other profiles like your LinkedIn page or your crunchbase, any of your review site profiles, also your Wikidata entry, which we'll talk about in a second. So the system does know all of those things describe the same exact company. And that also is very helpful for certain companies where people call your company by more than one name, for example, like Victorious and Victorious SEO for us. Wikidata, I was gonna talk about this. That does the same job from outside of your website as the schema does on your website. This is a free public database. It's also run by the same foundation that runs Wikipedia and it stores facts about your company as very simple statements that machines and AI systems can read. But it also gives your company a stable ID so that your schema and other sources can point to that stable ID. You also may have seen that Google published a guide this year on optimizing for its AI search features, which lists overfocusing on structured data as a mistake and says structured data isn't required to show up in AI answers. Now I think we can trust that it may be true for Google, but Google doesn't speak for every AI system out there. So my view is that schema is cheap to add, and at minimum, it removes some ambiguity about who you are, but I wouldn't expect it to just fix a massive brand issue or a page whose content doesn't actually say who you are in the first place. Now that your own website, if you've been following along with this, says clearly who you are and you've established an understanding of your brand. Maybe you've used my recommendation and example of the hedgehog concept. Let's now talk about other people's websites. Because as we covered in episode 12, what other sites say about your company decides mostly whether you get named on a buying question in an AI search. So the goal with brand mentions is to get as many mentions of your company online as you can in content that also talks about the subjects that your brand should be known for on your hedgehog list. You may have heard this called co-occurrence. I'd use that language, which means your company's name and the topics about your brand appearing together on the same page. And it's how these systems learn to connect the two. Ahrefs, which is, if you're not familiar, SEO software company, they studied brand visibility and AI answers late last year. And how often a brand was mentioned across the web line up with how often it showed up in answers from ChatGPT and Google much more closely than any of their actual link metrics did. So at Victorious, we call this mention building. And we do this because, link building is a thing, but it's different. And that's about going after backlinks specifically. And that is a different type of signal than what we're talking about with mentions, which is just mentioning your company in a third-party piece of content. Also, when a lot of marketers hear digital PR, which is another example of this, they think of like press releases and getting a big story and a big publication right? And mention building is one part of what a digital PR program does, which, for the sake of this conversation it's focused on getting your company named alongside of your subjects on sites these systems read. Before you go and get new mentions, you should first look at what is already out there. And we went through this in an earlier episode on AI brand reputation. You should start by pulling up your directory listings and your profiles on review sites and on sites like crunchbase and your LinkedIn company page, any literally any coverage you've earned over the last couple of years, and compare each of them against how your company describes itself today. And that comparison is going to give you two lists. And the first list is going to be descriptions that are wrong. Descriptions that are incorrect, like a listing that might still advertise a service that you had dropped a couple years ago. Or even might leave out a service line that you launched in the last year. And then the second list is places your company should appear and does not. Now, wrong information, like what's on that first list, that can end up in AI answers, even when your own website says something different. I saw a Washington Post report last year on a guy who searched Google for a a cruise ship's customer service number, I think. And the AI overview scanned websites and it ended up accidentally giving him a number that scammers had posted on a third party review website and they said it was Royal Caribbean's actual number. And so what did the guy do? He called it. And it was different than the Royal Caribbean's own website number, but he used the AI overview number and he ended up giving his credit card to a scammer. So that's an example of when this happens. So for that first list, somebody hopefully can go from your team and log into these profiles that you do own and fix them directly. And the only cost there is going to their time. For coverages in sites that you don't own, you can reach out and ask the publisher to correct it, or you can publish or try to publish something newer. That can state more current facts or information and then get enough visibility that those systems might read that information instead. But that doesn't work every time. We should fix the profiles that we own before we go after any new coverage because these systems are reading these wrong profiles and information today. And fixing them is going to help take a wrong description out of what they read. SEMrush is one of the only companies that I have seen publish before and after numbers on whether this kind of work actually changes how often AI answers actually name a company. And they measured it with their own tool. They started building relationships with the owners of pages on other websites that AI systems had cited very often to get SEMrush described accurately on those pages. And at the same time, they were reworking and adding content on their own site. And across the buying questions they tracked, how often an AI answer named SEMrush compared to its competitors went from about 15% to about 25% over just six months. And they ran all of this at once. So they can't really say which piece of those tactics drove what, though they said that the work beyond their own website, that third party mentions in correcting information and clarifying information look like a really big part of it. Now that's the first part of the list. For the second list, what I would recommend is for more brand mentions, is starting with places where you can create or claim a profile yourself. And I would start with, if you haven't already done it, Wikidata, because it's free and anyone can create this entry. And at a minimum, the entry that you can create should cover what your company is, where it was founded, where it's HQ'd, its official website, who founded it. After Wikidata, I would recommend then you can start to claim all the other profiles and directory listings that your company doesn't have yet, which means review sites and directories in your industry. And then I would strongly recommend to fill out each one of these as completely as the form will allow it. And try to use the same description of the business on each one of them. Again, this mostly costs your team time. You can do this this week. After that, that's when earned coverage comes in, which means getting written about in publications and on websites where you had, you know, no hand in the writing. And I say this goes last because it it costs the most, honestly. And you also honestly control it the least, even though it is the type of content that these systems cite more often. Nobody has measured these against each other. So the order in my opinion comes from my judgment around cost and control of the process you can have. Also, just quick side note, LinkedIn is also showing up a lot more often in AI answers right now for B2B questions. And most of what gets cited from LinkedIn comes from posts and comes from articles people publish under their own names. And we're gonna cover that, just kind of a teaser here. We're gonna cover that when we do an episode in the near future around running a mention building program. And again, if you have not yet learned about what it means to build an off-page strategy for AI search, there's a recent episode that we ran on The Search Signal that we will link to in the show notes, and you could read and listen more about that there. But now that we've covered the work on your own website, and we've also talked about the work off of your website. I want to take a second before we wrap up to talk about how long it takes for any of this stuff to actually show up, which depends on honestly, whether AI systems are going to be searching the web before it answers a question. So some answers, in AI systems come from what the model learned when it was initially trained, which was based on the web as it was in a single moment at a certain time and date. And then some answers come from that same AI system running a live web search and reading pages in real time right then. If the system searches the web, a change that you make can show up as soon as it reads your page again. And there was a software vendor that tracked a small set of changes. on a software company website early this year and found that the typical change showed up in Perplexity within a couple days. And then in some of the other platforms that they tracked, and this includes ChatGPT search and Google AI overviews. The updates were within about one or two weeks. So now if the system doesn't search, the change has to wait until the model gets updated. And when for example, when OpenAI released GPT five in August of last year, its training data ended in September of the year before. So anything that changed about your company after that wasn't in what that model had learned. Now most platforms are going to do some form of both. And whether they search depends on the platform. It depends on the question that gets asked, which is why you might see one platform pick up a correction that you have done within a couple days while another might continue to repeat an old description for much longer than that, maybe months. And because the timing is different on every platform, I would recommend you start setting up tracking when you start doing this work. So you can use a tool, Scrunch or Profound or Peec or any of the bajillion AI platform tools that are out there. You can set these questions up and it will ask these platforms on a schedule so you can see how often these platforms are naming your company for the subjects that you want to be known for, and you can keep a log of the profiles that you are completing and when you're updating it and the pages that you are changing. And can see when the changes show up on each of the platforms and when. We covered the rest of what and how to track AI search performance in the last episode on measurement. So I would definitely recommend listening to that for more information. But pulling this all together, let's take a step back. Building a brand entity is brand building that marketers already know how to do. And the difference that we're in now is that AI systems read everything published by your company and about your company, and they pull it together into a description and they use that description to decide whether to name you when a buyer is asking which company to work with. Getting named means knowing what your company is best at, making every page you control say it, and cutting the content that says something else and getting other websites to repeat it. So thank you for listening to this week's episode of The Search Signal. That's it for me today. If this was useful, please go ahead and subscribe so the next episode will find you. And I will see you next week.