Schema Mythbusting: The Biggest LocalBusiness Schema Study Ever (and the Surprise ChatGPT Result)

Jake Hundley, David Quaid, welcome back to the show. Thanks for having me again. Great to be here. Also, um there is a slight delay on my on a microphone. I just plugged it in. My old microphone is busted and I plugged it in and now now you guys uh now it looks like I'm speaking 2 seconds from the past. Well, it's great to have you here because even with the delay, it's great to have you because you just completed the most extensive schema test ever done. You did the most comprehensive test about whether or not schema is worth it both for SEO and for LLMs and you are sharing the results. This is going to be an exciting episode with actionable outcomes. Yeah, I had Jake on. It had like um I don't know. I I went in with uh obviously like the hypothesis I think all of us had which was like schema does absolutely nothing for rank. Um, and the problem was is like I did the same like a very similar thing with the geotagging XF study that we covered last year. And um, there was there were a couple criticisms of that study which I which I I actually agreed with which was one was um actually the only one I could really think of was that there wasn't a control group and a test group. So as we went through time with the study like anything could have impacted it like an algorithm change or anything like that and the entire test sample could have been manipulated by that um by the algorithm change. So in this one um I was like okay we're going to have a test group and a control group and we're going to go like we're going to go through time through the study so if there's an algorithm change they're both affected and the one that improves the the the one that improves is going to be the winner. Um and uh going through this like the whole methodology and everything. I wrote like eight pages of methodology before I had it reviewed um by other people in the industry. You guys included Joy Hawkins, Darren Shaw, David Hunter from Local Falcon, Ma Yoast all reviewed it and um big names. It was a lot a lot of preparation and uh it'll still get some criticisms which even in the study I address the the uh potential criticisms and um why things in the study are the way they are. Well, so I we'll talk about the criticisms, but I guess what exactly did the study prove and then also what did it not prove? All right. So, it proved uh that local business schema. So, I I feel like there needs to be a distinction between rich snippet generating and non-rich snippet generating schema. Um because that's what we tested. So, local business schema which does not generate rich snippets. uh it does not or it does nothing for rank on Google maps, Google uh or Google mobile. Absolutely nothing. It also does nothing for search engines. So Google, Bing, Yahoo um and um and then uh it does nothing for some AI or LLMs. So Gemini and Grock, it didn't do anything. Uh but the weird thing was is that with over 92% confidence it absolutely impacted Chad GPT. So that was like the the one thing where uh that was like the one area where I was like I don't know maybe maybe not and then ended up being like pretty ironclad there. What was the impact with Chachi PT? So for um for both I have the report up open here but for both the uh like the general query. So we had like very similar to the XF data study we did like a general query which was like service in target area. So for example like lawn care in Chicago or near Chicago or something like that. Um it was yes with like 92% confidence local business schema improved the position and so we'll put an asterisk here because a lot of people say that you can't have like positions in LLM but there is a whole call out in this section about how we come to that conclusion but it increased it by three positions. So, uh, if you if you don't know, if you if you type in like lawn care in Chicago, usually chat GBT is going to give you between three and five answers. And so, going three positions is quite a lot. Um, and then the share of AI voice, which is like similar to like share of local voice if if you guys ever use geog grids and local falcon. Um, but it's basically the percentage at which you do show up uh in in the results. and that increased it by 10 points uh with over 90% confidence. So if you're showing up 40% of queries, you're going to be showing up 50% of queries. That's basically what that uh uh that says. This method of marketing is so effective, I had to make sure it wasn't against Google's rules before I kept using it. It's a form of SEO I call compact keywords. Whereas most SEO focuses on putting up articles to answer questions, how, what, when, compact keywords focuses on putting up dozens of pages that sell to searchers who are actually looking to buy. These pages rank on Google and convert so much better than normal that when I discovered this years ago, I couldn't believe this was allowed. It's less work, too. The average compact keywords landing page is only 415 words. Compact Keywords is a 13-hour deep course on getting sales with SEO. A customer said, "We spent nearly 18,000 in the last year and a half on marketing and SEO through different agencies locally, and that did nothing. We decided to take the leap on the compact keywords course. We're now getting about 6 to eight calls per day on a good day, which is just unheard of." Another customer said, "Give it to a junior employee. Have them follow it exactly as Edwards laid out. You don't have to do anything, and you're going to gain a six-figure SEO level employee just by having them go through this course." Compact Keywords is about setting up an SEO funnel that brings you sales for years and years and years. It works with AI. It's less work than traditional SEO and it makes way more money. You can get it now at compact keywords.com. Back to the podcast. Are you familiar with Mark Williams Cooks uh Mark Williams Cook's duck test? Yes, I referenced it in the study actually. Yeah. Awesome. So can you explain um how why chachept is not looking at schema as just like normal HTML as normal text? No, I this this so this this study for me like I have my theories um on why like Chad GBT isn't but like I I do want to preface it by saying like this study doesn't tell you why anything. It just tells you what is and what is not. Um, so the biggest thing with this is like my theory is um I think I even say it in here. I say like look unless like OpenAI built their own index that we don't know about like I don't know. But also um uh I do know like the only thing I can think of b from all the speculation out there is that because chat GPT does not have an index it needs a way to process information quicker and faster. And so the only thing I can think of is that schema aggregates that information quicker. And so when you're asking it for certain like details about things, it doesn't have to crawl to figure out is this a business or is this like an aggregate site or um does this actually have a location or does it not and is it near where the person is. Um that's like the only thing I can think of. David, what do you think? I think the whole thing about schema is interesting. I I spoke to a group of CMOs offline and I was like, "How come the schema messaging resonates with you so much?" And they were like, "It's because it gives us control, right? We feel like we're losing control, right? Google's taking everything away from us." And I've always thought like that's just how Google is anyway, right? I didn't realize it was a recent issue. But to me, it just seems so odd that whether you have Google or or um chatb building its own index to like scan the whole web trying to be the most relevant engine and then to dismiss everything because it doesn't have schema makes no sense, right? and the fact that schema isn't validated, right? Which means, you know, it's it's why Google doesn't trust keywords as a as a as a meta tag or why the description doesn't help as a ranking factor or why everything on your site is the publisher side and what what becomes a ranking factor is how other people talk about you or what your clickthrough rate is. Those are the factors that that that you you're just the relevant side of the algorithm, right? And the third party validation is what sets where you are in in the relevant side. We're all claiming to be relevant. And if you had to look at it from a Google point of view where let's say everyone is trying to get spam you and get into your index, you can't trust the publisher. And I think that's where people feel let down, right? They're like, "But I built a good company and I built a good brand and I deserve to be number one." And that's everyone. Like even if somebody starts SEO today, they're always like, I can't believe that other company ranks above me, you know? It's just like this this thing that people have. And so I don't know why people hold on to schema and think it gives them this special advantage, right? It it just makes so little sense. I can see where, you know, you have to have schema for a job listing and why it's so vital and why it's so vital for um hotels and flights and things like that. And it makes some sense there. But also if you look at um what's in a HTML document and what when you download a web page and you could have a blog post say with a thousand words let's say the average word is five um characters that means your your blog post is about 8 kilobytes in size right and then you have all of this JavaScript and all of these images that could turn it into like a 200 kilobyt but all it needs is that text right and just downloading the body text is a 5 kilob 5 to 8 kilobyte sized exercise whereas downloading all this like CSS and schema and everything else just suddenly bloats it and now the c the the LLM has to tokenize and scan everything. It's just so much extra work that just doesn't make any sense. But I think that what I love about Jake's study and in in this is that we're we're we're facing this deluge of of um geo propaganda, right? Whether it's on X or LinkedIn or or Reddit that's like we've analyzed all these things. We have all these things in common and most of them are just recycled SEO myths that died under SEO and that are now like reliving their best lives under go. Um, but there's no pointing at um, how the study was done, what the data, what the methodology, who the experts were that reviewed it. Jake started this in November, right? Um, and we've been talking to him online on our LinkedIn group about this. And that's how long it takes to study 39 companies. Yet these people are putting out like infographics created in half a millisecond on LM every day telling you here are all the absolute facts right and I think it's it's such a a a short lesson for people like you have to apply critical thinking when you read these things I saw someone on LinkedIn say uh look even like this person was like what do I need to do for SEO and chachi pt responded you should use schema and the person was like look even chachi BT is say that you need to use schema, but like that's just because it's quoting it's referencing what like probably the most viral SEO articles say, which is use schema. That doesn't actually mean you need schema. That's I thought that's I thought that's what people would get from you know our exercise with like fun fluencer and king of SEO is that we can put anything in front of an LLM you know and and if you look at the other thing that's being traded and and we had this on on our episode about um with the AI tool is like you know people saying that LLMs are scaring the whole world web and then taking all of the companies who have a succent value proposition and remembering those that's a massive exercise Right. Um why would it why would it do that to all that information and and then yet it's so susceptible to us overwriting it in a query fan out from yesterday, right? It it just it just makes no sense. And I think we're at the I saw something very similar in SEO like 26 years ago where people were saying like, "Oh, Google knows when you bought your domain name and Google knows every fact in the world because it's got access to the largest database." And and that was like in the year 2000, right? And I think we're at this, we're still at this sort of like unicorn phase where LLMs can do anything and they know everything and now we're going to solve cancer and population growth and food and hunger and we're going to get to Mars because of LLMs. Um, and we haven't yet seen just how basic they are, right? I we think we know, but the the most people like the three of us know and a lot of viewers know, but the three of us. Did you uh did Did you guys see Mark Williams cooks uh cats.txt experiment? It's it was brilliant. And Jake, you were say you were going to say something. I think I I think I did. Um I remember something about it, but I can't remember all the details. Yeah. So, he puts up he he creates a new standard called cats.txt TXT and has actually before this went viral because then when it went viral it's completely ruined and Chachi PT knows what's up. Before it went viral, Chachi PT was confidently saying, "Yeah, cats.txt is going to help you get cited more in LLMs because that's what the cats.txt standard website said. This is a thing that works. It's Mark Williams Cook. He he he got I think cattxt.org and put out a LinkedIn post about it. I I'm not sure if he did any more than that to like get it cited. I just asked uh recently, I think it was the day after their schema report came out, but I just asked cheap chat GPT if uh local business schema affects uh rank in Google and it cited us first and said no. Oh, so that quick this my article I I submitted it to with no links. I submitted to indexing as soon as I published it and it was uh it was indexed within I think two hours. I think we're going to see I think we we we're going to see a problem that and and it's the same problem that we see with uh sort of like flat earth, right? So there there was a big problem with flat earth and people making flat earth videos and um that that poor guy who was the limo driver in Vegas who built the steam powered rocket unfortunately passed away in his experiment. Right. and there was no need for him to do that, right? But if you look at it, NASA or Oxford or all of these universities, they they they're not going to put out a 100,000 videos to match all of the flat Earth videos, right? They're just going to say like this is dumb. Um the same with Google, right? There's the Google um QRG guide. There's the Google document that says, look, if your if your job is to create EA, stop. That's not how EAT works, right? And if you ask John Mueller or Dr. Marie Haynes on blue sky that that's what they'll tell you. The problem is that a a thousand other agencies will go and write a document saying it is critical schema is amazing and so when it does a query fan out and it gets 30 documents Jake is one of them at the moment but eventually it's just going to realize hey there's 29 other documents that say schema is amazing and it starts to drown it out because it's a consensus machine and that's one of the problems with LLMs and and trusting LLMs. And I think the other thing that we have in society is that it's sort of rude to push back on people. Like I do it on LinkedIn and like everyone's like um you know, but it's easy. I I you go to Reddit, someone puts up a post says um I did this, this was the outcome. All you have to do is push back and say like where's the data and they disappear or they block you. They it just turns into vaporware, right? Because it's very easy to push back on these things. You don't have to be an LLM expert. But if you're not doing it, we're just going to be stuck in these cycles where people are putting their time and faith and it's a malinvestment into doing the wrong things. Yeah. A mal investment. Jake, what did critics of your study get wrong? And what criticism of the study do you think was fair? Um, I'll start with I think the criticism that I think uh is fair uh just because it's the easiest top off the top of my head. The the most fair thing and there's really only like one thing that comes to my mind is that it it wasn't long enough which cuz our ours is 5 weeks. Um to that I say it doesn't really matter how long it is as long as it is over a month I think which is what ours is. And the reason why is because as long as you can verify um indexibility uh before and after or like a a new crawl before and a and after the schema was placed the control and the test period then it doesn't matter because that information is now ingested. So you are now uh you are now being indexed and all that is being considered into whatever ranking signals that Google wants to consider. Um the other criticisms on the LLM side which is uh well you know if for the for those of of for those people who think that LLMs are training all the time some people will say like well an LLM only does their training data like once every 3 years or something like so like you're never going to hit that test period of whether it's ingesting and and learning from that right away. But like that doesn't make any sense because as as everyone states that like like you guys are saying like you can be in an LLM tomorrow and it's like well they didn't do training data so like clearly they're they're creating consensus somehow every single day. So um that's probably like the only one that's like actually fair. There's a lot more that we can address on here after um after a little bit after this that um people might be having in their heads about it. Um, but it's it's really easy to just kind of like rebuttal and and explain why we did the study the way we did. Um, but uh, as far as the Sorry, what was the other question? I I asked what what was what was fair and what wasn't fair. And like I also I also saw like I saw a bunch of people giving you criticism and you just responding. Did you even read the study? Like you're you're saying things that were mentioned at the at the beginning of the study. So yeah, there was one guy who didn't read it. Um, in fairness, I think he skimmed it, but if you skim it, you will miss the methodology section. Um, because the methodology section is like this big. It's not even a full fold of the uh of the screen. And the reason why is because I wanted this study to be applicable to everybody who reads it. Like, there's going to be people who skim it. There's going to be people who look at the abstract and just see the gra the infographic and be like, "Yep, there's the answers. Great. Don't read the rest of it. I don't care." Um, and there's going to be like mega nerds like like us for example who are like okay well where's the you know tell us where the data came from. There's a methodology section you can jump link to it from the table of contents and um everything is in collapsible uh arrows. So like there's pre-est and client sorting environment cleaning and benchmarking control and test period schema plays considerations things like that. And if you expand those, the article gets really long. Uh it's like it's over 7,000 words and all those most of those words are right there. So, um yeah. So, it was frustrating because I spent over probably a 100 hours just writing this. Um and um I mean you guys know I I was sending you teasers in April like uh you know just going through the data. Yeah. You sent a lot. Yeah. And here we are in June. So, um, uh, so I was like frustrated and I said like, "Look, I I spent this many hours reading, like writing this. The least you could do is read it." And, um, and he responded, he goes, "I missed I I missed the methodology. I didn't even see that." But he was like, "What's the local business schema place?" Because you can't just say local business and then not sure cuz like it could be really long and it could be really short. Like, so I actually have the exact schema I placed. Obviously, it's anonymized, so you can't like it's just random information in there, but um and then another criticism I think uh another person had was that it was only 5 weeks and um the argument didn't really make a lot of logical sense and I think it was just a lot of stuff that's perpetuated by geoists where it's and and even and even people who are hardcore schema ad advocates on the on the SEO side of things, but it was that Um, Google needs more time for uh, entity resolution, which is a a term I absolutely hate and despise. Um, because uh, Google like his argument was entity resolution takes time and it compounds over time. And then he started talking about how like different systems ingest that information. So I push back like what systems are you talking about because there's only one that we need to reference and that's Google's ranking system. like I do they have multiple so um he didn't respond to that but then the other one was the entity res the comment about entity resolution and compounding over time because he started bringing up like NAP citations and like like other entity signals and I said like hey we got to stop right here because now what we're doing is we're not isolating variables because if we're talking about compounding over time from other signals is it the other signals or is it the schema and so Um, but also just on the entity resolution thing on its own. Like I, it doesn't make any sense because Google, this whole concept of entity resolution has been around since Snap citations began to be important. And like the whole reason why it was so important in 2012 to make sure name, address, and phone number were identical on every platform. I'm talking like it's not just like ST on one site and then street on another site. Like remember when like people were insane about that? Now it like Google doesn't need that. It's fine. You can have ST on one site, you can have street on another one. Your your phone number can be in a different format. Um Google can res resolve those entities through that. It doesn't need the local business schema to do that. And because one that can be manipulated too and if Google is primarily built off of authority and consensus and that's how they want to rank then you having majority of leverage and control on your own property doesn't make sense. It's a conflict of interest. So um the the entity resolution thing just didn't make sense from a Google from the Google perspective because they haven't needed that for years. I remember sharing um your XF um data post on a on a on a subreddit last year and um somebody got so annoyed cuz I I think that that they worked at an agency where that was their job was to take all the photographs from the client and add the data and then upload it. They got so angry they flamed me for a year. They flamed me everywhere and they actually created a Google account and left a negative comment on my um Google profile. That's crazy for for my company. They were so outraged by it. Oh my god. Well, you see, you weren't the visible name. I didn't quote you. Um, it's it's kind of like when I posted the Google study saying that they didn't uh that you didn't need geo for um somebody two people accused me on tech SEO of um of clickbait and and bait rage and I was like and they said like the straight out with the title and I'm like the title was Google's post. I just copied it as Google wrote it. They didn't go after Google, they went for the messenger, right? And that's how you know they're so wound up emotionally and that that's scary. I knew this. I think it's scary. I I've I've been like joking to some of my friends who aren't in the SEO space and they're like, "Oh, what are you doing this weekend?" I'm like, "Oh, I'm like lighting the internet on fire." Um, and like cuz like if this is people's job like specifically for SEO, like this isn't good. like it's not a good day to be uh someone who is heavily invested in schema SEO especially I keep want to say this especially that schema does not produce rich snippets um like local business schema because it does absolutely nothing for SEO now like what my big worry is is that um I guess maybe it's not a worry it's it's more of kind of like um caution is that like because the study showed chat GBT is has has such an influence over it. Um I think people are going to cling to that even harder now. Um and and that's okay. Um I think they will too. But we've also shown that you can rank number one in Chad GBZ without any schema at all. So that's so so people also need to be aware that this isn't like this isn't the way to like this isn't like the way right like you need schema to be in chat GBT. That's not true at all. Does it help? Probably. I mean, this study shows that it does. I mean, if if I'm go off of my own results and someone says, "Does does local business schema uh impact chat GBT?" I would say yes, absolutely it does. I spent six months trying to do trying to figure it out and it does. The other question this doesn't answer is um said the impact is small. No, the impact on THP is big. It's It is. I thought you said it was like from 40% to 50%. Oh, 10 points. Yeah. I guess if you think that's small, that's small. I think I think 10 points can be big. Um, but like if uh that's visibility, right? So that's getting into chat GPT. So let's say 10 points is small. Okay. If you're if you only show up 10% of the time and now you're showing up 20% of the time, you're still not really showing up in Chad GPT. So um but if you do show up in Chad GPT, this is where I think the really big thing is. This is where like if you're like the fourth recommended according to this, you would be bumped into number one with the local business schema. Well, that's big. So, yeah, that's big. Um, as far as visibility goes, um, yeah, that it's that it's like showing up in chat GBT every time. Um, schema is not going to do it. Like, you're going to need a lot more than than that. Um, and that's that consensus. Uh, well, so let me let me ask you this. After doing this study and actually and David, question for you, too. So because of the effect that local business schema has on chatbt is it worth it? I don't think so. Um so one of the things I've wondered about chatbt and local business is um because you can like when when I'm when people were asking me does catch use Bing or does it use Google? was like, "Well, if you go to maps, if you do a map search, right, like pizza near me or whatever, you could see the Google link on the map in the bottom lefthand corner of your Chrome browser." And I was like, if you can't see that that's being delivered by Google, but then I wondered how how is chat GPT passing my location data to Google and how accurate is that? Because it the map didn't match the Google map exactly, right? But um I don't think it would change anything for me because I come from a world where I didn't use schema for those things and it never affected anything and I'm not about to change. If if if I find I can't do something and I can't get into first place, then I'll reconsider it. But if everything we're doing just keeps hitting with the same success rate, I'm not changing it if it's all broken. I'm g say it's worth it. Um, I'll say this because it's easy. You take the template I have in my article and you just put your stuff in it and then throw it onto your homepage. Done. That's it. Um, it's now like is it something to like promote and say like this is the way this should be like like local business schema should almost be talked about like metadescriptions almost like is metad descriptions worth it? Yeah, sure. Do you need it? Like you don't really like to rank. No, you don't need it. But it's easy. Just just do it, you know? So, also I think we come from two different worlds. I'm my world is very is almost never local unless I'm helping a neighbor out with a business, right? I mean, so the the way I look at it is uh I I stopped putting in meta descriptions like 12 years ago. Um because Google has its own like lab where they look at, for example, what's the screen size I'm going to, right? The port size. Um if there's a W, it takes up three times as much space as a capital I or a lowercase L, right? And so it will create the the title and the snippet and it will reduce the snippet if it's got site out callout links. I prefer to get call out links at the bottom of the snippet. They're like little round buttons which are point to H2s on your page. I found for me those work better, right? And so when I started those posts on on Reddit, the first one um was a post with a project that we reviewed for in fintech where we built a satellite site. We had 25 back links to the site. It had 50 pages. Most of them were like FAQs and paas. And I just noticed when I was looking at SEMrush, the number of SER features for the site was something like at the time 19,000. It's now revised it down to 9,000 retrospectively. I don't know why Simrush does that. I have to look into it. But it was it was 19,000 and I posted it with a screenshot going no schema 19,000 SER features and it was every single feature with zero schema. And I'm like if schema is so important then it shouldn't be here. Other pages with schema should have have taken the place. If you're building a system that says schema helps, then it has to reorientate its index where everything without schema has to be dropped, which makes no sense from a relevance engine point of view, right? It it's undoing it. The second thing is it can't have a duopoly where it just jumps from one thing to another randomly depending on the page. It has to be consistent. So if you don't if you can rank without schema, that should have been the stopping point in in my mind. So, um, it it's it's hard to believe that either Chatbutt or Google would invest so much money in a in a relevance engine and then like restack by the scheme of of of the presence of schema. And that should have been the critical thinking point that I think people stopped at but don't want to stop at. And it it makes me think of like when I was an internal SEO or when I've worked at big companies with internal web teams and you do something that requires the web team to give up on a lot of tickets and they have to focus on something like a rebuild or something in the CMS that blocks you from doing something in CMS. And the way as a leader you build their support, their buy into that project is you show them the results of what they do. And every team will automatically credit what they did as being the credit for the heavy lift, right? While also trying to give credit to the team. But when they have a private chat with the CEO at the water cooler, they'll be like, "Oh, did you see like we did this on the CMS and SEO went up?" And there's that like overcrediting of systems. I think sometimes that that rewards these. And if you look at the different I think there are different tribes in SEO, right? I I I don't think that this morning. I was thinking that in my head. I'm like, why did why is SEO so tribal? I was literally thinking that. If you look at tech SEO and the other SEO subreddits, the the the difference is chalk and cheese, right? Like if if you go to most SEOs and say like, "Hey, what what do I need to rank?" There's um authority, there's um topical authority, there's audio system. If you go and if you go to a tech a web engineer, it's a sitemap, meta description, schema, page speed, and your tech stack, right? Uh, and you can even ask an LLM, and this this shows like the flip-floppiness. If you if you say to um most LLMs like, "What are the most important parts of my website?" And you put in like text stack and CMS, it go, "Oh, yeah, yeah, your tech stack's really important." Google's never evaluated a tech stack. that was the case, Wix and Web Flow would be wouldn't be working, right? Um and and and so these it's these tribalisms I think that that skew the different points of view and again compound the problem for the person who's just trying to do SEO at the end of the day who I think you know I can imagine for them this is very confusing. David, like who wants to sit and listen to a study and it's like just should I do it or not? That's how they're thinking, right? David, why do you think uh why do you think the study found an increase in citations with local business schema? I don't know that that I that I have no idea. Um I think so what what's interesting about it is as Jake says um unless KTP has its own index, it's largely using Google's index, right, which we know and we know it's using it through SER API and Google is suing SER API. um that that requires another set. I I would imagine you would have to start looking into how what's the difference between a Google map listing and a chat GPT map listing. Is it how chat GPT sends information and it doesn't have as much access as Google does if you're on a mobile device? I I don't know. There's um cuz you uh say the the the thing that I think so I'm I'm diving into for our own agency. I'm diving into building like a a brain like a client brain for each client and then we just if a client needs something we just ask the brain and then it produces the the assets and then does the thing we need it to do. Um and like one thing I noticed about this or learning through this process is that everything has to be converted to JSON like everything cuz that's just how the that's just how the LLM is going to ingest it properly. And so knowing that schema is written in a JSON format, my theory would be that if it does use Google for example um to to source the information as the search engine, it still has to go to those pages, crawl it and review it. And so the only thing I can think of is that like the faster it can do that, the faster it can develop information. So in order to to kind of beat the the time cut off, it'll ingest the sites with the uh the schema quicker um because it's in that language that it understands quicker um versus just trying to crawl all the content without getting like exact pinpoint uh details that it needs for the fan out query. Yeah, I I I I don't know. I've never I've never looked at it. I I think because are you talking about the way Google delivers the information down to chatbt for it to build the map on on its own? Um have did you did you test at all the difference between the results of the Google map and the chat GPT map? Yes. Yes. There's actually um I don't know if we can share screens on this, but there's actually a um I'll show you uh let's see map map. Um there's a like it shows the map outputs here. So it shows that uh like this is a normal geog grid and you see like it's very radial and like that's how Google ranks things. So, this client specifically, they actually live like right here um in the corner of this, but we're optimizing for this area here on their site and everywhere. So, like we can show like we can rank really well on their on their target area, but where they're living, they're not as well, but we're not optimizing it. But you can see the radial from it. As we get further away, it's just red. Uh Chad GPT geog grid, it's just it's random in that entire service area. It's random because we're targeting the query. we're targeting is like right here. Um, but it's it's all over the place. Now, granted, like every every single week we're doing um a uh uh 81 queries within Chad GPT because it's a 9 by9 grid and and we're we're testing this in every single spot. So, um it is a bit random and I don't know I don't necessarily know why, but to the point too, like this is what it looks like. So I zoomed in on that one grid here where it's it's four. I clicked on four. And just like with a geo grid, you can kind of see the the results that it takes that screenshot of. And so this is how you can kind of get that output within um local falcon's uh uh UI, but it shows you basically this bulleted list of things that it's recommending. The first one's a competitor competitor. This is actually a directory site and then it was our client. And then additional context. So, this isn't a competitor or directory. It's just it's just no entity. And so, when it shows you the actual quote rankings or the positions that I'm talking about in the report, um it shows you here. So, really where the third one because the fourth one the third one's a directory. But, um but yeah, as far as like how it sources that, I'm not sure. But I do know that when you ask ChatGBT a uh a question um like like you're looking for a business near you, it'll mention things that are said on their site. So clearly it's going to their site and it's and it's ingesting information aside just from what's in SERs. So if it's ingesting things on their site, how is it ingesting things on their site? And then what does it prefer to ingest? Um cuz it's there's got to be a preference if it's saying schema improved position in the output. Well, it looks like the geog grid is completely upside down. They're not following the Google placements. Yeah. I mean, it's just random. I like if it was upside down, I feel like it'd be the like radio in the other direction. But yeah, like when I say upside down, I mean it's random. That's what I meant. Yeah. Yeah. Um it looks like it's completely random, which means maybe Maybe it was just maybe the fact that schema it was associates with with a sort of a higher position is just random. It it could be like I mean um the difference is is that the uh it could have like a wider area coverage too. Um I don't know. But like basically what happens is like at the beginning the grid like the all the points are random of where it says it's ranking better at. But then all the points become more and more saturated in that randomness of improvement. So even though it's not directional, all the points in the grid that we're using and maybe maybe if we went wider there might there might have been more of a pattern, right? Like if we did a a 600 by 600 grid across the US or in the entire state or something like that, maybe maybe we would have saw the pattern. I don't know. Um but we did a we did a um it's a really long term. It's called a a two sampled Welch's one-tailed t test to determine confidence. And to determine the confidence, you basically have to find the standard deviation of every single uh test subject that we did and then put that into a formula to determine yeah the the test group outperformed the control group but with what confidence level. So there are certain situations in which the schema uh test group performed better than the uh control group but the confidence level wasn't high enough to determine whether or not yes or no that improved or did not improve. So we only take confidence levels at 90% or higher. Anything below that is just a modified coin flip. Right? So like 50% would be a coin flip and then anything between 50 and 75% technically is just a coin flips without enough uh sample sizes. I think just something that's occurred to me um because you said this the other day Jake um you know if you go to Bing the results are just so terrible right that you end up having to go back to Google and I wonder you know a couple of years ago I I remember going to meetings and people were like oh I'm so sorry SEO is over it's you know it just it's dead and um I was like yeah sure that's it's but it's had a good run um but as you said if if if the results from being are that terrible that people keep bounc bouncing back to Google. What does it say for chat GBPT and and how how much it's stalled, right? It's absolutely stalled as it got close to that 1% of Google size. Um, you know, I I can imagine like if you're looking for a SAS product and the same five SAS products are there, but they're in a random different order in a Google search result, not the end of the world. But if you're looking for, you know, if you're on a family tour and you want to sort of like get the family vacation, you want to get to, you know, Grizzly World or um having the maps work is pretty important. And and if it doesn't get you the the to your favorite pizza joint or your favorite Thai restaurant, you're going to give up on Chat GPT and bounce back to Google pretty quickly, right? Because that that that time is money, right? Or that that time is valuable to people. Um, you said yourself if you go to you go to Bing, you you you get the search result, it's so terrible, you end up bouncing back to Google. Um, if if Chat TPD's results are so different from Google and Google has been so good at it historically. I makes me think a lot about why if if that's contributing to Chatp's failure in that sense. Yeah, it could be. I mean, every day, every day we get more and more uh people finding us using uh quote AI or LLMs. Like, we just got one yesterday. Someone was using chat GPT, which is interesting because we don't have schema on our site. Um, so I I really don't know. I mean, I'm I'm a Gemini fanboy, so I use Gemini a lot, but I don't use Gemini for finding businesses. I use it just for top level research. Um, same with if I use chat GPT2 and I still end up going to the actual search engine which with the actual index that's going to give me the results that I'm used to. Question for both of you. What do you what do you guys think? Is it fair or not fair to sell local business schema as a visibility tact as an AI visibility tactic? Oh, I'll go David first. It depends if you if you're if it's like if it's 10 minutes work and it's like under $100, I don't care. Um if that that's not what I'm you know that that doesn't worry me if if it's built into the website and it's easy to do and you got Jake's template, go for it. It's not going to break anything or hurt anything as long as you don't put your faith in it. That that's what worries me, right? when when people say like I'm trying to rank my website and I this the you see a lot of posts like this you our business is fading my mom and my my mom and dad's business is failing I really need to help them get higher ranking that's what worries me um I don't care about an ego argument about um schema and I'm not trying to put people off schema or say schema is completely useless uh what worries me is people putting stock in it and and because a lot of people do something and say now I have to wait for it to happen and that could be two weeks and two months and that's pointless that's lost time that you can't ever get back. So, do it, but don't rely on it is my answer. Right. Yeah, I agree with David. I actually think it should just be like a value that you just include in your SEO already without it being a separate line item. It's just just do just do it. Um I think the biggest thing that's going to be difficult though that that I find um probably frustrating even on our own side even before the schema stuff uh that we did was uh you don't it's it's this tribalism that we talk about like between agencies and SEOs and um especially if you're an independent SEO and and you're selling your services to multiple businesses. Um, and I it's one of the reasons why people have such a hard time sharing clients because one thing that people don't know about is what they're paying for and what they're not paying for with SEO. We talk about paro SEO all the time. Um, and it's like, okay, if they're paying 500 bucks a month, yeah, don't expect 100 blog posts. Don't expect the schema. Don't expect all this other stuff. Um, but like that's no reason to then criticize the person doing the SEO simply because they're missing a few of your checkboxes. Like the end at the end of the results at the end of the day, the only thing that matters is that someone is getting results for their clients. End of story. And and um if there's things that they can do to improve, like maybe like, hey, yeah, why don't you have local business schema? And you you know, it could just be like we haven't historically prioritized it and we've gotten results without it. And I know I was just telling um what was it on Reddit? Uh I was I was responding to a post where it was like what's our local backlink strategy for our clients? And for the last 8 years, we've never had one. We don't I mean it's for a local business, we have been able to get the results for our clients that are needed for 9 years with a 90% retention rate without a local business, without a local link strategy. And um if frankly if somebody wanted that, they're going to have to pay more because you're just prioritizing relevance, right? Yeah. And then you just submit to Google Search Console. Yeah. And people it's a lot less competitive than people think. And like you don't need you don't need the back links. And we we've tested back links before like getting like local. Are these for new are these for new new websites or or primarily websites that have been around for a while? I mean all of the above. like the the like if you do a new like a brand new website and then you have it relevant for you know not not competitive terms but like high intent terms and most high intent terms are actually not competitive and you just prioritize relev relevance but it's a new site and you submit to Google search console what type of results do you see in what timeline it so it depends on the location um but I will say on general I usually we within like two to it's all this also depends on seasonality with our industry, but I usually say within like 2 months, we're seeing really good results. Like like I'm talking like a,000% conversions year-over-year. And uh or you know, like in our onboarding form that we have uh our clients fill out, we ask like how many how many leads from your website do you get per month? And usually the answer is like five five to 10 or something like that. And you know, right when April hits, like so if they onboarded us in in February and like right when April hits, they're getting like 30 40 50 leads um just from H1's page titles, you know, uh making slight modifications to their their GBP, which I don't even think moved that much of a needle um between what they already had and what we were, you know, were doing. Um I I think like Google puts more weight into a site uh than people realize uh for GBP ranking. Um obviously after after proximity and after reviews I think like there's like all these other not like um uh levers you can turn like like services offered and uh like location which I know like Joy has done some stuff about this where it's like adding locations doesn't really change anything but um after that it's like you make changes to the website and you would see geo grids fluctuate. One of the toughest or one of the most interesting rank positions is if you look at Manhattan, right? And Edward, you'll appreciate this more is like, and it's a bit like the subway map where like Jersey doesn't exist, right, on the western side. It's just like Manhattan's like the furthest point. Then you have like Brooklyn and Queens and the Bronx. Um, and uh, Staten Island is connected to the south, tucked into under Jersey. um is is that if even if you have a Manhattan location, sometimes you can have SEO agencies ranking in in Brooklyn because it's only like three or four miles away, right? Which is very very small if you look at say like the Midwest of the US. Um but if you can rank an SEO agency in New Jersey state in a Manhattan search, uh that's that takes a lot more than just proximity to do if you you know because it's out of state for a start. Um, and so it's it's in those hyper competitive, you know, where you have like a thousand SEO agencies listed in New York that I think maybe backlinks start to play because I' I I haven't changed my profile in New York in 6 years, but the amount of traffic going to it has gone up 5x in the last 3 months or two months. Um, and so I can only put that down to authority going from like 20 to 30. Have you been have you been making a focus on link building over the last couple of months? Uh apart from losing 90% of my links according to Sam right. So I went from 5,000 to 500. Um but what I rank for has gone from 8,000 to 14,000. But you said your authority increased. So authority on Seamrush is calculated in its tweet using your traffic and rank positions. And so that increased my my backlinks didn't increase my total back because I don't know where those back links came from. If you look at our site um and our competitors, so if I type in like lawn care marketing company, uh we show number one, I think I might have illustrated this in the last time I was on, but we show number one under paid results. And then our back links or the number of linking domains that we have is 493 and we have 1.5,000 links. Now, if you look at our second competitor, they have about same same amount of domains, but they have 108,000 links. And right underneath that, we have 11.6,000 domains linked with 456,000 links. Yet, we're ranking way way higher than them. And they're these are both lawn lawn and landscaping marketing companies. So it's like there is a difference between I guess the quality of backlink and the relevance where it's like I think people get too fixated on like the word backlink strategy and they just say backlink I need more and I need to get this but it's like I'm I did this research study for the best backlinks that I will not have to worry about another backlink for another six years. So, I know like we're we're I'm looking at uh doing a co-publish with Moz right now, getting in Moz's publication, and they're reviewing the raw data right now cuz they're not going to take my research study, which is funny because they provided the MA part of the data, but then there's also like the local Falcon part of the data, too. So, they're like they're being extremely thorough about it before they're letting me in their their publication gates. Going to be cool. Uh, I'm going to go back to uh I'm going to go back to schema and um I think this is a this will be a cool quote to end with. So, this is from Href's article. We tracked 1,885 pages adding schema AI citations barely moved, which I guess kind of contradicts what you found, but there there's a passage at the top of this that I really love. Um, and that is schema markup tends to live on betterm maintained, more technically sophisticated sites and those same sites publish stronger content, build more authority, earn more links and do all the other things that gets pages cited. Schema could be doing the real work, but it could also just be riding the wave of every other signal. I think we we see uh we see correlation not causation a lot with schema and that passage from hrefs sums it up pretty well. It would have been better for them to just not publish that and save their breath cuz I hate I hate when people publish these things that say uh we analyzed 6 million pages and we found this you didn't find anything. You found you wasted your time and you're wasting my time by reading this because this is just this is all correlation. That's all this is. Oh, you mean with adding schema? No, with um yeah, I mean just a study whatever they're saying like it's it's like well no they're saying they're saying that they're saying that AI citations barely moved by adding schema. I don't like I really don't care what they're saying because it's all it's all correlative. They're not they're not doing a controlled study like you have you have to actually do like like set the control. Um I guess I don't like let me ask this like did did they have a whole methodology? Did they they do they do have a whole methodology? They have a long section on this is a new one cuz they published one at the end of last year. No, this Yeah, this was published May 11th of this year. So this is actually less than a month old. And uh Yeah. So they they they have like they share what they did in it. Yeah. Then that's I mean what they say actually could line up with what this report is saying too because if they're saying that all these sites have all these other things that are also moving the need that are highly technical then that 10 points isn't going to matter because that 10 points of that that it increase in chat GPT for example or probably taken care of by something else. And so like what we're dealing with here in our study is also local businesses. And so I made it very abundantly clear in the introduction and the abstract that this is local business schema um and it is for local businesses. So when I say like in the conclusion when I say does it affect the answer is does it affect local businesses and that's where this comes into play. Now you can make all kinds of arguments for other industries too and that's not what the study shows. Yeah. I I think also Jake, I don't know if you want to reiterate what your study found, but to me it said that local schema didn't affect ranking for local businesses in Google. Yeah, I didn't it didn't touch it in Google at all. Um yeah, not a single one. The only one was um for one query there was like Gemini showed like a higher confidence interval than most things. Um, but there's no explanation as to why cuz Gemini failed every other test. And the so the only thing I can think of is that if I extended the if if we had a larger sample size, which which I would argue is almost impossible to get in a controlled environment, um, then uh that percentage might be even lower. But yeah, gentlemen, thank you so much for this podcast. David, you're going to be back on tomorrow for a for a geo debate. Yay. Looking forward to it. So excited for you. Those are so much fun. And and Jake, thank you again. Thank you again for doing this study. Um I'm I was happy to to have connected you with people and and Jake, thank you for coming on and explaining all of this. Yeah, thanks for having me. I felt bad cuz the last time I was on I I I tiptoed around the what the methodology was going to be and I said I didn't want to share it cuz people would criticize it and I'm like cuz it was only half done. Like there were things that we didn't know that we were going to run into that we had to to pause and rewrite the parts of the methodology and that's exactly why. So now if you go back to that last video and you find those comments about that, you're more than welcome to read the methodology. Now buddy, go check out the study. It's going to be in the description of this episode. This is episode 1,070 of this show. 1,070 days in a row doing this podcast. Thank you again to my wonderful guests and for all of you out there. If you watch us on YouTube, thank you so much for watching. If you listened on Spotify or Apple Podcasts, thank you so much for listening and I will talk to you again tomorrow. Bye now.