Transcription
So, there's a video going around right now that's telling local business owners that Google IO, it's their Google's annual event, just ended SEO as they know it. And the fix is a brand new playbook, new tools, new files, a new way to build your site for the robots. Now, I run an SEO agency for a living focused on local SEO. So, I'm going to go ahead and watch this video with you and break down what's accurate and what might need a bit of a rework.
Now, most of what changed at IO doesn't change anything about the fundamental work of SEO. And there's one move in here that he tells you to make today that Google itself has said it does nothing for your visibility. I'm going to show you exactly where this falls apart. So, let's go ahead and get started.
Google just ended search as we know it, and the evidence is already sitting in your Google Search Console. Your impressions are going up, but your clicks are going down. And that's not an accident. That's AI answering searches before anybody even reaches your website. Combine that with what Google just shipped at io search agents that not only browse the web for you, but can book for you, purchase for you, and even submit leads for you without you ever visiting your website. And you have a complete rewrite of what it means to get found online. And if you don't adapt, your SEO, your leads, and your sales will quietly disappear.
>> So, this is clearly a bit of fear framing here. Uh, not surprisingly. Uh, that being said, the fundamental change that Google talked about at IO was that search becomes a center for controlling AI agents. What that does mean is that for 25 years, we've been building websites to rank on Google, but then writing the content of those websites to convert users, real humans who are reading the content. That will change. Uh, that is what Google announced that now we have to write websites to get recommended by AI and write websites to convert the AI. Humans won't be looking at your websites anymore in the near future. But just because an AI is reading your website and not humans, that doesn't mean your leads and your traffic and all of that are going to go away completely because the AI uses fundamentally very similar techniques to read and evaluate your website content that Google's own algorithm does. But in this video, I'm going to show you exactly what changed, the one mental model that explains all of it, and these strategies that you need to be doing right now so that these search agents recommend you and not your competitor. Let's get to it.
I don't like pointing right at the camera, but that's just me. There were a lot of incredible announcements at Google IO, but there were two that are going to impact the way we search. The first one is the fact that the search box is now turning into your own personal dashboard. You can ask, "How do I build a balanced workout plan?" and a few more other things and Google not only looks through your information but creates a personalized dashboard that you can use over and over and you can also share with people. This is going to be the death of a lot of other traffic particularly for websites that have things like recipes or workout plans. See you later. Those are not going to exist anymore.
So the AI agents still need somewhere to pull the data from. So the point like it it depends on how those uh websites are being monetized, right? Websites or workout plans that monetized based on AdSense. Yeah, those are probably going to go away because the AdSense revenue is going to dry up, but uh there still are a lot of people uh who don't necessarily want an AI to be giving them recipes and giving them workout plans. Uh, so uh the right monetized program, those can still exist, those still work. They just probably won't be using as much Google to drive traffic over there, but there's still Tik Tok, Instagram, YouTube, a variety of ways to build an audience to drive traffic, at least for now.
And number two are these browser search agents. Now, this isn't specifically new. We've seen this with Perplexity's Comet and a few other AI companies have their own versions, but when Google does it, they automatically have a billion users behind them. So this really makes a massive impact and you need to know how to optimize for these search agents. What you need to start thinking about is the fact that your customers now have a concierge service. These are the search agents because they know so much about you or about the customer that they're going to search with all that knowledge and information behind them. Let me give you an example. Let's say in this weird example I owner of a Vietnamese restaurant somewhere in Melbourne, let's say Fitzroy. If I type in find me an accountant, Google's not just going to search normally with these agents, accountants near me. Absolutely not. It has all of the previous information about me. So, it's going to see, okay, let's find an accountant, but for a 12 person Vietnamese restaurant in Fitzroy with 1.8 million revenue that uses zero and prefers in-person meetings, that is all the information that it's going to do the search with because it wants to find the perfect accountant in this instance for me. So we're no longer optimizing just for humans or for the AI search engines but for the search agents.
>> Yeah. So he's absolutely right here. This is the big change. Sundar Pichai has been talking about this for a while. Uh, right now it's the AI overview that has basically decimated informational search query click-through rates. But with the Ask Maps feature that Google introduced in March, with uh the search box becoming uh a home for search agents itself, we see this really impacting local search. The map algorithm, the fundamental three-pack based on the Google Business Profile, it shows the top three results. That algorithm hasn't changed in 10 years. That streak is likely going to end uh within the very near future where Gemini is now going to choose local businesses instead of uh humans, instead of Google's algorithm showing you three and then a human looks at them and and chooses them. This has massive implications. One, as he talked about, the LLMs get to know you the more you talk to the LLMs. This means that hyperspecific attributes become incredibly important for these searches. Not only hyperspecific attributes that it's learned about you over all of your interactions with it, but the prompts themselves are typically much larger, much more rich than what people typically typed into Google search. One implication of this, which is very good for Google, means the more you use that LLM, the better it's going to do for you. So if Gemini integrated in with search gets better results to you the more it knows about you, that is a massive competitive advantage in terms of people going and using Gemini instead of ChatGPT or instead of Claude or the other AI LLMs.
So the fight here is how to make sure the AI agent has the information that it needs about your business to understand if traffic is down, how do we measure things? What's the next real KPI that we can follow? Well, this really comes down to two main things. Your search appearances and your citations. And how to find this data? Well, there's actually a free tool that is very much slept on. That is Bing Webmaster Tools. I know, who would have thought I would have started using Bing. But Bing Webmaster Tools, which is their version of Google Search Console, has this fantastic set of data that you can have access to. If you go to your Bing account, then you go to the AI search performance tab, it gives you all of the citations from AI and all the times you're getting cited, which is amazing. So, if I scroll down here, I see that over the past 3 months, I've been cited a total of 8,600 times, the average pages that are getting cited, and the queries which my website is being cited.
>> Now, this is absolutely incredible. I completely agree with it. This recently, very recently rolled out on Bing Webmaster Tools and it got my team to go install Bing Webmaster Tools on all of our clients because of course we didn't have Bing Webmaster Tools installed on the clients before because who cares about Bing, right? So, this is very good information to have. It's the only way that I know of to get truly trustable data about what AI is recommending you for and those queries. Now, Google Search Console is also currently rolling this out to smaller groups of users. So, I would anticipate within the next couple of weeks to months that Google Search Console will also have this data for AI overview and Gemini query specifically. Very good to have this visibility, which we really just don't have right now other than Bing Webmaster Tools. Really good information. And whilst I'm not getting the the traffic that one would used to get, I'm still getting cited, which is a good KPI to understand.
Now, we need to know that this is only for the Bing search network, which means just for GPT, which is all well and good, but let's be real here. The big competitor here is Google. So, we need to know, are we getting cited in Google AI overviews? There's another tool that you can use for that. I'm going to leave a link for you to try out DataForSEO for free. You don't need a credit card or anything. And there's a section here that is AI visibility. I'm going to go to the brand tracker section. And here I can put a website like let's put theversel.com. I got to make sure I've got Google AI overviews selected. And I'm going to go to analyze. This is going to give me all the times and the AI search visibility of Versel as a domain. So it's getting these many total mentions. Where is it getting mentioned? Mostly from the website, but also on YouTube and Reddit. And the mentions over time as well as the actual questions that are making the AI.
>> So, this is the problem with tools like this and there's a bunch of them that exist. This is one of them. I've tested a bunch of them. Uh, for the most part, they're all garbage, right? They don't actually give you anything that's useful or actionable. And we can see that by even just looking at the LLM answers mentioning you that it's generating. It's just not particularly helpful in terms of figuring out what you need to do, change, improve to get cited by the right uh types of queries. Uh, the problem is one of the key problems for this is that AI LLMs, as we talked about earlier in this video, they know so much about you that the search prompt is not always even super relevant as to why you got recommended and someone else didn't get recommended. Or vice versa. So the problem here is that with Google search, a handful of keywords that you could track, you could see your rank position, you could see what was going on. With LLM prompts, the amount of hyperspecific attributes that are so important to getting recommended, getting cited, it just these tools just don't track it in enough detail to be able to give you actionable information. And in my mind, information that gives me the wrong impression, information that's not actionable, that I can't do anything about, or that is incomplete so that I don't know what I'm supposed to do to improve it. I would rather have no information than have that information. Wrong information is worse than no information. And if we look at even this example, LLM answers mentioning you, a lot of these just aren't super relevant to versel.com wanting to be cited more. Like how would they take this information and then move forward to improve their AI visibility? It just anyway, this is why I'm so, so excited about uh what Bing introduced and very much looking forward to seeing this data in Google Search Console because then all of the tools like this, we can freely ignore them.
Site and mention Versel. So now that the new KPI is really more about search appearances and citations, you need to think of your website not just for a landing page for humans, but you need to think about it as a data API of your business for these agents. And websites have already been something like this anyway, but you really need to think about having all of the correct information and all of the information that you can on your website because essentially the ability to give the agents all the information they need to make sure that they recommend you. But here's the thing, you need the right type of information to get cited. Because having the right information doesn't mean having all of the information. See, what's really going to win now is you getting hyper-niche in understanding who your customer is. Let's say I'm still that accounting, or I've reversed the roles this time. I'm the accountant this time and I want to market myself from an SEO perspective. I can't no longer say, "Oh yeah, I'm an accountant for small businesses." That is way too broad and it's not going to bring me the right traffic because I'm not going to give the search agents the right information. I need to get deeper. For example, accounting for hospitality businesses in Melbourne. And that isn't even niche enough. I need to go a step further. And I would say things like accounting for restaurants in Fitzroy that uses Xero. That's the accounting software. And you might be saying, uh, there's no search volume for that. I'm going to get rarely no clicks. But remember, nobody's getting clicks anyway practically. So, you need the right clicks. And the more niche you get, the better information the search agents have to make sure that you're the one cited because you have all the exact requirements that a user might be looking for.
So yes, nobody's getting clicks, but that still means you need to target a niche that someone is searching for, right? Uh, someone has to be asking Google or ChatGPT or Grok or some AI tool, otherwise it won't work, right? So, if you said, "My specialty is in clearing snow out of the parking lots for Cuban restaurants in Miami." Yeah, if anyone looked for someone to clear the snow out of Cuban restaurants in Miami, you'd probably get recommended. But the problem is no one is searching for that. Okay? So, there is a fine line there. Okay? And what he's talking about is basically what we've been talking about since like 2018, which is entity matching. Okay? So remember the T in ChatGPT stands for transformers, and Google actually invented the transformer technology in 2017. Google has introduced it into its search algorithm before ChatGPT 3.5 was released. So Google understands entities in the same way or a similar way that ChatGPT does. So what we're talking about, what we want to do is to figure out entity matching. Okay? So your service, the service you provide is an entity. The location of your business is an entity. The customer and what they're looking for are entities with attributes. Okay? Uh, so historically, if we think about ranking on the Google three-pack, the old map pack, there are three primary rank factors. Proximity, relevance, and trust. Well, with Gemini taking over, proximity largely goes away. Okay? Gemini does not care about proximity in the same way that Google's old map pack algorithm does. So we still have uh authority and trust, but proximity is replaced with attributes. So what we want to do is produce content on your website that talks about those attributes. That's entity matching. So the fact that you are looking for restaurants in that suburb of Melbourne who are using that specific accounting tool. Awesome. Put that on your website. Write about it. Put a a subsection on one of your pages about it. But don't make that the entire focus of your website. We can entity match more than one thing. We can focus on more than one attribute. Uh, it's even more powerful as a side note if those attributes go into your reviews. If his Vietnamese restaurant wrote a review and said, "This is the best accountant ever for restaurants who are using this accounting software." That would be even more powerful uh than just having it on that website. So, when you're asking for reviews, don't just tell your customers, "Hey, could you please leave us a review?" Instead, send a message to your customers and say something more like, "Hey, could you say what we helped with and what the outcome was?" You'll get much more specific reviews with attributes. And attributes are the magic sauce for getting recommended by Gemini. Now, like I said, customers putting in your reviews, that is amazing. But at the very least, you can put these attributes on your website yourself. And you can take a look at the Search Console data for your website. The Search Console data will show what you're being ranked for on Google search. Yes, it doesn't have the Gemini data yet, but you can see some attributes that people are actively searching for on Google. If those attributes are relevant to your small business, make sure those attributes are talked about on your website.
New KPIs that we need to look at. We know what type of information we need in our website. How do we know if our website is search agent-friendly, that the agents can actually access our website and do all the things they need to? Well, let's do the test. To do this test correct, and by the way, it's free. You need to download a special version of Chrome called Chrome Canary. This is the version that is used by developers because it's updated almost daily with new features. Some of these features in Chrome Canary, not many people get. So, I need you to download that. Once you've downloaded and installed it, you need to do a couple of things here. You want to go to this URL, and I'll leave it on the screen here. And you want to make sure that you enable WebMCP for testing. This is critical because by default, it's just by default, but you need to make sure that it's enabled. Now, we're ready to go. We can test if our website is ready for the AI search engines or the AI agents, sorry, that are searching through the website. I'm going to pick a website here.
With the website loaded in Chrome Canary, you want to go up here to the three dots. You want to go to More Tools and we're going to go to Developer Tools. Now, by default, you might have been placed in the Elements tab. You want to go to the Lighthouse tab. And if you don't see this report here, it's because maybe you've got the doc sited in the side like this. I like it at the bottom. So, I'm going to place it like here, like this here. What you need to do is make sure that in the categories you have this Agentic Browsing category selected. I've got the desktop selected and I'm going to analyze the page load. Now this calls the Lighthouse report, which is just a classic report that every SEO should know about. It essentially tells you the performance of your website from a speed, from an SEO perspective, from a technical report perspective. Really good to have. But now there is a very new and important feature here because these are the normal data sets that we get: Performance, Accessibility, Best Practices, SEO. But now there's a new one: Agentic Browsing. That is the one that you need to make sure you optimize for. So if I go down to the website, I see here, okay, my agentic browsing. I'm missing a lot of things. The accessibility tree is not well-formed. So now you know if your website is actually optimized for this or not because you can get up to three out of three from the agentic browsing, and that's where you know that you're actually getting optimized.
So real quick here, like this is an actual audit, but you don't need Chrome Canary anymore. This is out of date. Like this, he posted this a couple of weeks ago or a week ago. When did he post this? 10 days ago. He posted this 10 days ago. This is already out of date. Uh, you don't need Chrome Canary to do this, and in fact, Google is going to roll it out soon to PageSpeed Insights. So you'll be able to see it right whenever you run a PageSpeed audit. But it is important to know that it's going to measure browser readiness, not measuring anything related to how likely the AI is going to recommend you. Basically going to analyze a lot of the super technical things that are inside Search Console, right? It's, you know, whether your buttons and your forms have labels, whether things on your page jump around after it's loaded. Uh, very, very technical things that don't really have a huge amount of impact for whether an LLM actually recommends you or not. Uh, most of the LLMs that are talking about local businesses, recommending local businesses, they're doing it through other devices. They're doing it through other centers like the Google Business Profile, which is already structured well. When they go to your website, they're not actually trying to interact with forms. They're just reading it to try to learn more about you. Uh, the fundamentals they're already getting from, yeah, the Google Business Profile if they're ordering food from you. We saw in Google's own IO presentation that they're doing it through DoorDash. They're not trying to place an order on the website itself. So a lot of that section is kind of packaging to make it feel secret and hidden when it really isn't.
You can do today to make sure that you are optimized for this new type of SEO. Well, thankfully Google has confirmed the one thing that's going to help this is setting an LLM.txt file on your website. This used to cause a bit of drama on the internet because people thought that LLM.txt files were a waste. Turns out they're not. So, how do you put >> No, they still are. You put this file on your website. Let me show you really quickly because it's very easy to do. All you got to do is if you're on WordPress is make sure that you have a plugin installed called Yoast. Right? So, I'm going to look for the plugin with you. I'm going to go to plugins. I'm going to add a plugin and I'm going to search for Yoast, and it should be the first one that you find here. I'm going to install. The free version for this purpose is going to be more than okay. So, I'm going to install this. And then you need to activate it. With that plugin installed, we can go to Yoast SEO because we're going to have that tab. Now, we want to go to settings within the Yoast application. You want to scroll all the way down until you see this one here, which is the LLM.txt file. It's just under the AI tools. All I want to do is enable that and save changes. And that's it. Well done. You've set your LLM.txt file on your website. You're on your way to becoming a lot more friendly for the search agents.
Now, I'm not saying LLM.txt is a waste of time and you shouldn't add it because it is different for agentic browsers versus the LLMs that we're using today. But the LLM.txt is essentially not being used anywhere today. Let me show you. This is a study that was done at Limey and they said out of 500 million AI bot visits across the 90-day window, 408 LLM.txt directly. Basically, none of them. Okay. Now, this is different for agents that are crawling, which really doesn't happen very much now. So, I still suggest it takes 30 seconds. Go ahead, implement the LLM.txt. Just don't expect it to meaningfully change anything anytime soon, even for agent browsing. So, not fake. Uh, but in the world of SEO, if something takes 30 seconds to do, then you probably should just do it. So, the rest of this video, he's going to pitch his community again. So, we'll go ahead and move on. I just want to call out the LLM.txt file that, you know, he told you to install today. Uh, one move, big fix, right? Uh, Google's own AI search guide came out and listed it under the stuff that you don't need for visibility, right next to the other tricks that AI optimization services try to sell. Now, yes, Google says a lot of things that aren't actually true in terms of showing up and ranking for search, but when you look at the actual bot traffic, the actual data, the crawlers really aren't looking at it at all. That being said, it is not useless. It is helping it will help an agent read your site a little bit faster. It's it's a real thing, uh, but not the thing that's going to help you get found. This is a a browser checkbox that's being sold as the secret.
So, if we strip the whole video down and look at what's left, it's really getting specific about who you serve. Make sure your site says exactly what your business is. Hand the machine clean, exact information so it chooses you. Honestly, that's not a new playbook. That's entity matching. It's the same fundamental signal that's been winning since Google stopped reading keywords, which happened in 2018. The tools changed. The file is new. The work isn't different. So, every time somebody tells you the game just got rewritten and here's the brand new thing you have to go and install and look at, check whether the people who built the machines are actually using it. Usually they're not.