Product of the week
(For a summary, click here.)
John Eischeid: Just to start off, just tell me a little bit about yourself, the work that you do, and that kind of thing.
Jason Dowdell: Yeah. So, I’ve been doing SEO since 97. Started and sold a couple agencies in that space, built some technology platforms. First web-based, SEO project management system, back in 2003. So we had a crawler pulled in all the metadata made content updates within that system exported them to the customers. So I’ve been around a bit. Last one was seotool.com, web-based crawler customized algorithm. Depending on the other factors that are present so scored every page to prioritize the issues and the solutions that needed to be applied. So I’ve been around for a bit. Currently head up SEO at Zen business and in light of where AI is headed and where audiences are headed and the nature of search and that’s what’s kind of got me into, “All right this is where we need to be, where the users are. And how do we do that effectively.” So yeah, that’s a little bit about me.
JE: I do have a question along the lines of metadata and that kind of thing. You mentioned before in some of your LinkedIn posts that LLMs are pulling content from Reddit, LinkedIn as well. To what extent are they also digesting the metadata as they would with traditional SEO, or does anybody know?
JD: Yeah, I mean they are ingesting it. We haven’t done enough experiments to say, “Okay, are they including or excluding the meta description?” Right. They’re certainly going to follow the robots guidelines for crawling and of course robots.txt only Gemini is parsing the entire DOM. So let’s say that you have schema data that’s rendered through JavaScript. So Schema app is a vendor we use it that’s in business and we have the JavaScript implementation. And that means that there’s rules where it parses content on the page and then uses JavaScript to literally add that content to the DOM. So it’s not in the raw HTML. So LLM crawlers, both for search and for the RAG side of it as well as for the model training, they’re not rendering the full DOM, so it’s kind of back to the old days of SEO, where up until past 10 years Google wasn’t rendering the DOM. So that if you have schema data in there which helps with disambiguation, which is really important when it comes to semantic optimization, which is kind of at the heart of LLMs, that content is going to be omitted. So that can impact your brand visibility. It can impact how the LLMs understand your brand quite a bit based on your owned content – so on your own website or number of web properties. Yeah, so.
JE: Okay, having said that, you mentioned also before that if LLMs are pulling from Reddit and LinkedIn . . . . I’m just kind of curious as to what role like the traditional website would still play, how much weight should people give to it?
JD: So I used to use this analogy to talk about the importance of pages, right – a page rank free analysis – or metaphor if you will, but basically that the homepage of your website is the doorway to your site. I could update that to say that it’s the doorway to your brand, so what that homepage looks like, the experiments you’re running, the messaging that you have there, there’s personalization. That’s the first experience that most people on an LLM are going to have with your site, unless they’ve engaged with you through a citation. So citations don’t drive the bulk of traffic. The bulk of traffic from the LLM is coming directly to your homepage. And it’s also coming through traditional search when people take the brand, because they’ve seen it multiple times through them educating themselves through their buyer journey within the chat agents whether it’s ChatGPT, Perplexity, an on-premise solution, right, you’re hosting your own LLM or hosting a model yourself. Wherever those things are happening, they have to get in touch with the brand so there that connection has to be made, right? And oftentimes it’s literally typing the brand into the browser going to traditional SEO and then clicking on the homepage. It’s going to – so to back up a little bit, your messaging has to be consistent, so, all of your advertising, right, needs to be consistent and targeting appropriately who your user base is. But then, when you get to the website, it needs to kind of overdeliver on the experience. I think I added in a comment yesterday that it’s like the brick-and-mortar stores that are creating Instagram worthy experiences, right? My daughters love to go to New York City.
JE:
JD: My wife and my daughters every year in January go to the city. it’s their annual girls trip and my daughters it’s literally a photo shoot experience for them. One of hem just graduated, which is super cool, but totally unrelated. so, there’s this one place they went to. I think it’s some ice cream museum or something . . .
JE: . . . Museum of Ice Cream. Yes, I know the place.
JD: Ridiculously over-the-top displays and whatnot, but they’re getting people in there through organic marketing, right, through socials. And the experience on the website needs to feel authentic, right? Because AI is being seen as kind of – it’s in the same way that people implied that Google put you at the top of the list so Google’s saying that you’re the best. So, it’s like Google is a person recommending you. The intimacy with AI is far deeper, right? I mean, literally this morning I was chatting with – you know, I get up at 4:45 go to crossfit at 5:30 I’m coming back home before I shower and take my youngest to school and I’m like okay I really need to get an AI pendant or something that will literally transcribe all of my calls and keep in mind that these are my personal goals, my professional goals, and give me an accountability for that at the end of each week, and tell me whether or not I’m making progress against those goals, and also to remind me of crap, because I forget stuff all the time, and conversations, I’m fully engaged present and I find that my thoughts are much more free flowing in those. But, there’s also the discipline of documenting, and just like I was late for this meeting, there’s so many ways you’re getting blasted every day. There’s only so much that you can remember, so, we need this kind of augmentation. So, I’m having this conversation and I’m saying, “Hey, what pennants are out there?” And by the way, ChatGPT did horrible. It was absolutely horrible, saying the Google pendant, which I can’t even find reference of anywhere, and I’m like “What about the rabbit thing/” And it was just terrible, so I went to the organic search, but my point is that for the most part if you’re asking general information right questions that you’d normally go to Google for, you don’t have to. You’re in this experience where you’re not being presented with ads, you’re getting recommendations, and it literally takes you through that buyer’s journey Getting educated a whole lot faster and at your own leisure. And the chat history is there. What we’re giving up for that is pretty significant, but the history is there. You can go back and reference it. So, we’re getting much deeper levels of recommendations and summaries and languages that we understand than we do in Google. And Google can do as much as they want, but they’re not going to catch up to that experience, right? So, to me ChatGPT, it really is the initial Google. There’s this very simple interface and chat really totally makes sense for this interface. I’m not bombarded – of course like all things, we’re going to ruin them – but right now I’m really enjoying that. So that relationship is going to be much deeper. I’m absolutely willing to give OpenAI tons of data about myself. And as I was having this conversation, I’m in the driveway and I’m like, I really need to put it in drive and turn off the car and get into the house. But I’m like, okay, I’m literally thinking about, okay, it’s going to know all these products that I have that I already have, the subscriptions that I have, because I want the most detailed recommendations for me. So, it’s enabling this whole one to one. So, if the website doesn’t deliver right on that one-to-one marketing experience, then you’re going to have an issue. Now, we’re in this age where the bulk of your competitors aren’t going to deliver on that one-to-one experience. So, it’s all relative, but the tide is being lifted, So, that’s why personalization is such a big theme for this year. because the LLM are creating this anticipation of “This is who has what you need.” And then if you get there and anything is off, then it’s like, okay, no, maybe not. So . . .
JE: Yeah, that brings us to a related question.If people are using LLMs to do research into a brand, competitors, products, and that kind of thing, to what extent did brands run the risk of being damaged by hallucinations?
JD: I think that — you know, in the same way that technology can move at light speed, right – so AI can become far smarter than we are, but our ability to process that is not as fast right we’re emotional beings. We do things, and I was literally thinking about the fact this morning that the way that we approach things is, “Wey these are the problems that I’m trying to solve”. The “jobs to be done” framework kind of summarizes that, and with LLMs you’re getting this “jobs to be done” like, “This is what I’m looking for. These are the things that you need to know about me in order to give me those,” and those are coming back. Sometimes those chats are not great. When I was asking it about the different types of devices that I can buy and my requirements, “It needs to be available to purchase right now. I want to know when I would actually receive this, because I don’t want to wait for six months on a pre-order or something.” And I also want to make sure that it’s actually still available, because I’m much more technical than most, I know that ChatGPT’s search limitations are pretty big. Google’s going to have a much better experience. So, I probably should have switched to Gemini. So, I think there’s a native understanding. “Look, it’s not always going to get it right. It’s going to say some things. But as a buyer, caveat emptor, right? it’s your job to make sure that – what you are – the decisions you’re making are fully informed or as best informed as they can be and AI is still in this kind of initial honeymoon period where could do all this stuff and we’ll go through the drought of disillusion and, “It didn’t really say it, so yeah hallucinations are there.” I don’t know that the brands are really going to be scathed by that quite honestly. However, there will absolutely be spammers that take advantage of those loopholes. The number of opportunities for that right now are just numerous that you’re not going to be able to count it. The problem is that spammers want to do everything cheap and really fast and see immediate results. and with these models, you’re not going to see immediate results, You’re not going to see a result until the next model comes out and it’s going to be trained on data. It’s going to be three to six months old. and the RAG piece of it is not strong enough to really be able to obfuscate or to pick up all the places where these spammers are pushing. Some will get through LinkedIn’s having a pretty big impact on the RAG for ChatGPT and increasing visibility of some of our competitors. And the contents of the post are garbage. But it’s having a bit of an outsized impact on it. But in the core model I don’t see it too much.
JE: More along the lines of pulling information from social media: If it’s becoming more – I mean you’ve kind of alluded to the idea that brands need to get their presence out on social media, they need to post, and put things out there – and I understand that if that’s what the LLMs are reviewing, but to what extent do you kind of run the risk of subverting the purpose of social media – which was to connect people – if you’re posting stuff on there just to get noticed by LLMs. Does that undermine the intended purpose of the platform?
JD: Yeah. I mean, I would say that and honestly, the things that have me so excited about LLM marketing is that when you do that, it doesn’t work. It’s not the same. So it’s kind of like if you’re going to proliferate a lie, right, or a conspiracy theory over hundreds of years, it’s really difficult to keep it going because the facts get misconstrued and whatnot. and if you think about the fact that with LLMs, what they’re really looking at is what is the essence of this brand, right? what are they saying internally on their site on their socials and whatnot? And then what is set being said externally by others about them, user generated content, and do those two match up? And the LLM, because it’s a statistical model, they’re literally looking at, okay, what’s the probability this is the next word? And are they discussing this feature? And so to me, if you’re not actually posting content that literally talks about your target audience, what they’re trying to accomplish, how your product and service helps them accomplish that, what are the pain points that they’re feeling, what are the emotional needs that they have, the place that many of them are in with either your competitors or without knowing what the solution is and you’re providing that to them, then they’re going to engage with your audience or they’re going to engage with you much more on social media, It’s going to be authentic. It will feel authentic. And the other part is that LLMs aren’t indexing all of your social media content, right? Only the stuff that’s publicly available is going to be indexed and not all of them are indexable. I mean, Facebook is pretty much a walled garden. and it makes sense, Competitively, Meta wants Llama to be, the model of choice. So they’re not all equal. And you have to look at, okay, which if you’re going to prioritize, which one you’re going to focus on, it’s number one, which ones have deals that are going to train the model and that are going to impact the rack. Reddit is one of those, right? So, Reddit’s an obvious choice. I haven’t talked about this publicly, but, two weeks ago, I said, look, what we’re doing in Reddit is not effective. And it wasn’t just about LLMs.It was about the fact that there are questions out there our customers are asking and people are complaining about and a lot of them are competitors. I mean it’s a wolf and dog’s clothing. I wouldn’t even call it sheep’s. And they’re saying all these negative things and I said, “Look we need to respond thoughtfully, because we need to counter what they’re saying with our own messaging, right? Number one. But also I want everyone else who’s reading these right to see, “Right, okay no they do respond.” Like it says, “It took me eight months to get a response.” Which is ridiculous. I mean, we have the highest rating in Trust Pilot of all of our competitors. We care about all of our customers. CS is the gem of our company, which just kind of shows you the character and the culture that we have. But it wasn’t reflective on Reddit. Secondary to that is, “I want to impact our visibility in LLMs, but I also want to make sure that the truth is out there.” And what was happening is it was distorted because we’d put less budget on this particular aspect of it. We’re not responding to everything. So I said to the agency owner, “Look, I want us to focus on this for the next two weeks. Respond to everything. Get the answers from CS, if the answer isn’t vetted, or you can’t get confirmation on it, ping me. And then we’re going to literally go to the next level and we will get that answer because I want to offer solutions. Just like, “Hey if we don’t have it we’re going to send you over to Bergman’s or whatever,” But the point is that I want to engage with our customers where they are right and make sure that they do get a real experience, whether or not we can drive conversions from it. Because it’s really about ma rketing your brand and so to me as an SEO who’s always been chasing not just rankings and click-through rates but increases in conversion rates and ultimately it has to drive conversions and leads and revenue. This is refreshing because it forces departments to align so brand marketing and social marketing and PR, paid channels, not only paid search but also our paid social channels, online video, display media, our offline, so podcasts and even – I shouldn’t say offline because we don’t do much offline – but literally every channel has to be consistent on the messaging. And that’s where getting back to it’s really hard to keep a lie going for a really long period of time. If everyone’s not on the same page, right, then the messaging is going to get messed up. and paid and organic teams typically do not get along. And so for us, that’s one of our advantages. We’ve always worked together. We don’t really care who gets credit for the conversions, as long as we’re moving the needle forward. I don’t know that our competitors are like that. I’ve consulted for almost three decades and I’ve never seen a team that works as cohesively like ours does. And, you know, it’s going to bite companies like it’s truly going to bite them, so that is interesting. I wouldn’t call them hallucinations, but you can create false narratives. But statistically speaking how long are you going to keep those up for right? Ultimately you have to invest in yourself, get your own house in order before you start trying to blow down the house next door.
JE: Okay. You did mention customer service a lot and I’m kind of curious about – like you obviously put a very big emphasis on it and this is at a time when a lot of people are turning those things over to chat bots, to LLMs.
JD: Um hum.
JE: So outside of SEO, to what extent does just having a conversation with an actual person still affect, you know . . . .
JD: Yeah, I think every industry and service and product is different. In some cases, people don’t want to talk to a person. but for us, if you put it in Google’s terms, YMYL, Your Money, Your Life, right? We do LLC formation and compliance. That’s our primary core product. We help you register a business so you limit your personal liability. It doesn’t get much more Your Money and Your Life than that. You start an LLC to literally separate all your business activities and protect your personal assets from anything that the LLC does, and so in that case you have to be able to talk to a person, right? And that person has to be knowledgeable. So how do we use AI? We index all of our transcripts, all of the calls even from sales and we put them into Snowflake and we have a RAG layer so automated answers and responses and all that stuff. So, we’re using more to enhance our individuals, right, the people that are answering the phones, but also the other aspect is you want to prevent people from calling if they don’t have to, Give them the best experience. And a lot of times they don’t need to call in order to ask these three questions. We want to proactively answer them on the page when they come to it. So, one cases one of the ways we’re leveraging that – and I think I posted about this, I was probably intentionally vague about it – but one of the things that we’ve done is we’ve literally taken that rag model and we’ve asked questions hey when customers call in and they say they complain about us not being like Legalzoom what are the 10 biggest things that they’re saying we’re not like Legalzoom on what do we need to focus on and what are the responses that we get? And we’ll go through those and sometimes it hallucinates, right, but that’s where you have that man of the middle, right, the person in the middle, the human in the loop, that says, “Okay, this makes sense. This doesn’t,” so we’ll have our editor literally get back with CS and figure out which ones of those make sense, which ones of those are actually accurate. And then we’ll update our product pages with that information, Because we want people to have those questions answered immediately. and those are the places where brands have such a huge opportunity. you put that content in there, those questions, those answers in a semantically optimized format: subject, object, predicate, semantic triples. Keep it short, but it has to be within the context window that LLMs can index. And you’re going to immediately see results. So, we did this and we literally saw AI overviews pop up for queries that have zero search volume. We don’t even check them because it’s so low. and we pushed all of our competitors lower on the page, right? They see the answer directly in Google or in the LLMs and great, we’ve addressed that before they even come to the site. So, I think it’s using AI to be as proactive as you can and then, reactive. There’s always opportunities for you to say, “Hey, we actually do care.” And so those personal touch points should have a bigger impact when they do get through to customer success.
JE: So, is there kind of a little shift going on in SEO going away from, knowledge graphs and, groupings more into trying to figure out what specific questions someone will ask an LLM?
JD: I would not say that there’s a shift from knowledge graph at all. If anything, there’s an amplification of the knowledge graph, and best way to put this is I actually track our brand and our competitor’s brand’s relevant scores in Google’s knowledge graph API. Because when we’re making changes that we intentionally want to increase our relevant score. Think about a page rank, right? I want to know where we are in relation to our competitors. I believe – this is my opinion – I believe that knowledge graph in Google’s eyes is being used to determine trust right so the EEAT and all the core algorithm updates that we’ve had and where big brands have gotten hammered you probably as I say – maybe we hit the limit?
JE: We’re at 12:29, but it’s gonna keep going . . .
JD: But basically, we’ve seen – what’s the best way to put this? – direct correlation between increases in that and us being in the knowledge graph and Google understanding who we are and us seeing a knowledge panel when you search for our brand in Google and how indepth that knowledge panel is. And I also believe ChatGPT is grounding their citations in Google search results. They return fewer citations, but I believe they’re grounded in the search results and AI overviews is one of those key areas. So yeah, knowledge graph is not going away. It is about indexing things and not strings. and if your entity is ambiguous, I saw somebody post today about, exact match domains, keywords in the domain as the brand name are going to make it more difficult for you to have your entity recognized in LLMs and surfaced properly, which it’s 100% accurate, but again, it’s one of those where even that little piece is so refreshing to me. I mean, having to chase Google algorithms for years. I mean, I can’t tell you how many audits I’ve done and how many customers I’ve had to dig out of penalties, algorithmic penalties and manual, at least a thousand algorithmic penalties at least and understand technically what Google’s doing, how they’re doing it, why they’re doing it this way, and then literally make the adjustments so that technically we’re literally doing everything perfectly for Google. Whereas now, no, it’s talk like a human, right? And keep your brand upfront and foremost at the top of the page. speak in terms your customers understand, like “Who is this page targeting? The content on this page?” Don’t forget about topic clusters and content hubs and all this other bull crap. Literally who is this page for? What are the problems that they’re trying to Okay, focus on that, right? And yes, There’s schema data. your headings need to be ordered and they need to be appropriate. But you also have to clearly your intent. Use your title to label the intent. It’s a product page, purchase, or shop, or see the review or, whatever your primary goal is for that page. Make sure it’s clear in the title. Make sure the H1 capitalizes on that. And all the stuff that’s at the top of the page is more important. And your schema data is critical, especially FAQs. If it’s a how-to guide, great. Mark it up. Put it into the stems, but also make sure it’s indexable. So the technical aspects of this went from 10,000 elements to 500 elements. And to me, that’s refreshing, and then you get rid of all the ads as well, and it’s like, thank God, I was terrified two years ago. and we got hit with a penalty, like a big one. And all the posts that I put up, I mean, it’s all legitimate. This is exactly what we’ve been through, me being the guy that is known for rescuing companies and saving them from these penalties, getting hit with one and I’m the head of SEO, like that sucked. It really sucked, very humbling. But our CEO was incredibly supportive. The C-Suite was really supportive. and we got through it, but it forced me to deal with a super weak area that I had in SEO, which was semantic optimization and entities and understanding really what that meant. So, yeah.
JE: We are almost out of time. We’re a little bit over, actually, but, is there anything else you want to add before we go?
JD: No, no, thanks for reaching out, appreciate it. I was really honestly really terrified to start posting on LinkedIn because I mean I don’t want to just be another voice out there. I know a lot of in-house SEOs that are terrified and me seeing not only the light at the end of the tunnel but a whole new frontier. It’s like okay I want to be part of this conversation. Of course balance it with family and my job and all that stuff but no so I appreciate you reaching out. It just helps to validate, that I’m speaking to something that’s relevant and people are responding to. So, yeah.
JE: And LinkedIn is very good for things like this that are changing almost on a daily basis. yeah. And if you’re worried about getting sucked into it, my suggestion is to just do half an hour, an hour every day. That way you keep it current, but you kind of have a limit. So, yeah.
JD: I’ve got five kids. My wife’s getting her master’s degree. My oldest just graduated from UF. The next one is going to be a junior at UF. I got three boys, two in high school, one in elementary school. I’m super busy, so I’m not so worried about it now, but just like people responding and the comments and, I mean, some of it, of course, is BS, but it’s been really nice.
JE: On that note, I don’t want to keep you any longer, because I’m sure you’ve got plenty of other things to do.
JD: Likewise.
JE: But, yeah, I very much appreciate you taking the time, and if I have any follow-up questions, I’ll let you know.
JD: Thank you.
JD: Take care, John.
JE: Thank you. Bye.
