Forensic SEO with Kristine Schachinger: Deep Dive into Google

• Who is Kristine Schachinger, the forensic SEO consultant?

• What has your SEO career journey been like?

• How detailed do you go into forensic SEO for clients?

• What is the SEO audit starting point for a site that has tanked?

• What's your mindset in regards to all the Google changes and updates?

• How do you think SEOs need to start handling things moving forward in the AI world we are in?

• Are Google now looking to rank genuine businesses rather than just information websites?

• Do you think minimal investment levels will make it hard for some businesses to compete online?

• How does Google look at the copyright and ownership of a piece of content?

• Does Google connect a person with the topic they usually talk about?

• Does Google treat a person as a specialist depending on the content they have published?

• What is the future for SEOs when it comes to Google SGE?

• Will Google SGE diminish clicks to websites?

• What's your thoughts on Google Ads when it comes to SGE?

• Will SGE require a massive shift in user behaviour?

• How will Google handle fact checking now they themselves are presenting the information directly through SGE?

• Is it important for SEOs to read and understand the Google ranking system?

• Is EEAT a direct ranking factor?

<<< Back to The Unscripted SEO Interview Podcast

The unscripted conversation between Mark A Preston and Kristine Schachinger

Mark A Preston: Welcome to the Unscripted SEO Interview Podcast. Yes, it's 100% unscripted, 100% unrehearsed, 100% unedited, and 100% real. I'm your host, Mark A Preston, and today our guest is Kristine. But, I want her to introduce herself because I'm a big believer in not taking perception as key. So I don't miss anything. Could you introduce yourself, please.

​

Kristine Schachinger: How much do you want to know?

​

Mark A Preston: I say, what's your SEO story? Describe your SEO journey. And what do you specialise in when it comes to SEO? 

​

Kristine Schachinger: So, my last name is Schachinger. And according to our German SEOs, I do not say it correctly which is true. I say it the American way. And then I started SEO in, as almost everyone, not in a direct path, right? There's like schools now, but when we all started, there was no direct way in SEO. So I was had a sociology and art degree out of college, which meant I was a barista for a year while I figured out what I was going to get employed for, because there are no jobs for a sociologist that like a PhD. And so they, someone had an ad in the paper for a web designer and HTML person. And I said I know how to do design. I have, an art degree. So can you teach me HTML? And they said, yes. And so that's how I got started in tech itself. So I was a front end designer and developer for probably six years before SEO came on my path and I moved to Las Vegas and I was working for a travel company here and we're spending 4 million a year in Google clicks and this is back in 2004, so 2005.

So I said, Hey boss, I heard this thing called SEO and I've been studying it. Can I take it over and maybe give it a try so I don't have to spend so much money on Google clicks and she said we have a company that's supposed to be doing it. But we haven't ranked for anything in six months. If you can figure out why I'll let you take it over. So I figured out why there was no index tag on the site. They forgot to take off. And so she gave it to me. So that was added onto my skillset. So at the time I was doing a hard cut, a hand coding front ends design accessibility. Usability, UX, and then I added SEO. And then over the years, just, when you go out on your own, designing websites is really a pain for people when you're just by yourself. And, because there's 15 people that'll give opinions and it never gets finished. I told one group, I'm like, if you have 13 people in the committee, six months from now, you will not have a website. And they had 13 people in the committee and six months later, they did not have a website.

So I decided it's probably better to get out of the website and just focus on SEO. Mostly I still do, like I did, was an accessibility lead for the federal government for three years. So I dip in and out of that. And then I also still build WordPress sites and still, but small ones, small businesses and stuff. And then about when the first algorithm started coming out, the. Big ones, not like Florida back, way back in the day, panda and Penguin, and we've changed to Hummingbird, Hummingbird was added, all that stuff. I found doing Forensic SEO and finding out how to recover sites was super interesting, and super satisfying because people, businesses are figuring out that they're not going to last very long if they can't get out of that hole. As we all know, it's like a 70 to 90 percent drop generally. So I find it super interesting, but I do all the regular SEO stuff too. I do day to day. I do site migrations. I do strategies. I do growth strategies. All the normal stuff. Only thing I don't do is I don't do link acquisition. I can tell them how to do it. If they have a team of like people, they can do like PR and outreach, but that's a, that, that's a whole set of skills and relationships that you have to have. And as a solo practitioner That's a little bit too much for me. And then what I do is when I do full scale SEO and I can't do all the things, I bring in people like at our level to do those projects for clients and then work with them directly.

​

Mark A Preston: Wonderful. Now you took, you mentioned that forensic SEO, is that similar to problem solving or how detailed do you go yourself?

​

Kristine Schachinger: Oh, with forensic SEO, I'm talking about basically site auditing for recovery, so I do a lot of site audits. I always audit a site before I do any SEO on it because I don't do cookie cutter SEO, so I tell the client we have, we do an audit first, we reduce, your costs, and we put your money where it needs to be. So I do site health auditing, which is forensic, but the real forensics are on the site, the algorithms roll out. And then they did devastate a site and then figuring out how to recover a site. So I know how most ranking systems work well enough that I know which levers to push to get the site back. Some of the newer ones we don't know enough about yet, but we have ideas. And then also manual actions as well. Like one client came to me, 24 months on a manual action. And it took us two tries, but we got all their traffic back. And then the craziest one I had, if you want me to take a minute, do you want me to go ahead? The craziest one I had was a 2 billion company. I walked in and I see that their fourth most traffic keyword doesn't make, it's just four letters, but it's not in their catalog. It's not one of their products. It's not any of the description. So I go search online and it's a porn site.

And I find out the site is what they call a cross scripting vulnerability, which means someone's able to divert traffic to their site by using the position of this site. Therefore, the biggest keyword was a porn term on a vulnerability that had been on the site. So long that the Department of Justice had fined them for it eight years earlier. And one of the big hackers in the world wrote an article about it, but it was still there. We go in and find out that they also had an open search box. This is what forensics is, right? You follow the the keyword, and then you find out it's a porn site, and you find out there's an access vulnerability. And then you find out that their search box creates pages that are indexable, and any words you put in there go in the title, the description, the H1, and the URL. That had been on the site for 20 years. So the site had 20 years of spam pages, but because it was a React site, they had no catalog of those pages. There was, unless you could crawl it. There's no way for them to know the page exists. That's why I always tell people to react sites. Every time a page is created, make sure you record that in a database. So we found out everything from ringtones, to Mozilla downloads, to things I'm not going to say on your program. Yeah. To really bad statements against religious figures. And then, of course, this porn spam. And then found out there were 13 million Russian porn spam links.

And because they matched the page, the words, Google didn't discount them, so they were ranking for, if you looked up that porn site information, the porn information, they're ranking in the top four because they're using this company's information. So that led to a manual action for user generated content being delivered by Google four weeks before Black Friday. On a $2 billion e-commerce site. So fortunately I had gotten every, I had gotten the VP to agree to fix it. Took a little bit of angling back in June. So we had already started, but we weren't on, we weren't there yet. So Google, that's when Google gave you URLs, to fix. So we just fixed those and we resubmitted and I was like, ringing a prayer. It's a big e-commerce site, and we fixed the URLs they gave us, and maybe they'll delay the penalty and the the devaluation is not a penalty. And they did, until right after the second week in December. So then we had to go through and find all the, we had, we couldn't blacklist terms because it was a fashion site. And like clothing and all sorts of things, apparel. So they used terms that you might use in porn. In clothing names or descriptions. So we had to do go through 300, 000 URLs to find which ones actually garnered traffic and then make sure they were all whitelisted to get traffic in the future.

​

Mark A Preston: Wow.

​

Kristine Schachinger: And got the manual action removed in January. We did all the work in December and got it removed in January, but yeah. So that's a forensic. SEO. That's not going through and just just doing the regular stuff. That's let me check your site, audit it, and see what problems you have. And a lot of sites, more often than you think, have serious issues they don't know about.

​

Mark A Preston: I was going to say, over the years, I've, yeah, I've come across a lot of unexpected things on sites when I've looked at it. Yeah, hacks, you name it. We come across both often. It's not obvious. So when you come across a scenario where, oh, but sites tanked or so much happened. Where do people start? What's the starting point? And they go from there when the, auditing the site.

​

Kristine Schachinger: I'm doing a talk on it at Brighton SEO San Diego, and that's anomalies. So a lot of people in SEO, just general speaking, but I have an academic background. So it's my nature to look for the anomalies because I did everything for a PhD except the final dissertation. So I spent years in academic research methods, all that sort of stuff. So when my eye sees an anomaly, it goes what's that about? And if I can't answer it, then I look into it because those four letters being the highest, the fourth largest keyword term on the site. Everyone just overlooked, but I was like what is that? That doesn't make sense. That's not a product I've ever heard of. That's not anything I know that they carry. So let me check into that. So anomalies are often the key to really big site issues and they don't always mean there's really big site issues, but I have to know the answer. I go down the rabbit hole and I have to know the answer to the anomaly. Why is this performing this way? And. And if it's performing oddly, whether it's in the data or whether it's on the site or whether it's, just in visits or, you look in GSC and the page is like, all these pages just dropped. That was one site. And there was a core update and we found that all those pages had a big technical issue. So those anomalies lead you down the rabbit hole to like the bigger issues, generally speaking, and the uncommon issues. I check for all the regular stuff too, but right before I deliver an audit, I check all my anomalies and make sure that all my anomalies make sense. Okay,

​

Mark A Preston: now that there's a lot, there's a lot of things happening with Google at the moment. Loads of updates, loads of things, obviously, machine learning has been here for quite a while, AI, you name it, and I feel as though a lot of SEOs are just confused. Do you feel that, I don't know, there's so much happening at the moment and with the SGA, what's your take on things? What's your mindset in regarding what's going on?

​

Kristine Schachinger: Which part?

​

Mark A Preston: I'm gonna say that it's I don't know, it's just like this, it just feels as though everything's just come at us all at once. But I know yourself you've been heavily involved in the research and how Google works and, the background and machine learning and everything. Is it just, I don't know. I just feel as though it was always gonna happen. It's just all this. machine learning and everything that's happening in the past is what's brought us here. Do you feel the same or do you feel as though there's a definite connection with obviously AI has been pushed on onto Google? But how do you think SEOs need to start handling things moving forward?

​

Kristine Schachinger: Just one thing about the Google. So a lot of people think Google was behind because chat GPT came out, but he just dumped it on the whole world. It's not like it was like new. It was version 3. 5 that we got right. But Google actually wrote a paper and I can't remember the order it's making. Dilatants out of debutants or debutants out of dilatants. And basically he's talking about search and where it needs to go and machine learning and large language models and all that. It was 2018. And that's where Sam Altman got the idea to go forward with ChatGPT. So Google has been creating these with large language models and using BERT. And starting with BERT. BERT was The big push forward, because prior to that, they can only disambiguate a sentence going forward, one direction. They couldn't mask words and have it guess. And so it was very slow, it was very time consuming, it was very expensive. And so BERT allowed them to just mask a word. And if I say, I'm going to go by blank and eggs, in the United States you'd say bacon and eggs, most likely terms. And so it could determine those, but it doesn't have an understanding of the sentence. It just knows from patterns that's the most likely word.

So that accelerated everything in the large language models and the machine learning when it comes to language because now the costs are greatly reduced, and the process time and amount of processing is greatly reduced. That's the four. Father of all that happened. When it comes to AI and search, we'll talk two sides of it. One using AI in SEO. Use it for, don't use it for, okay, here's how I say it. Don't use it for anything super important. If you have a page that needs to convert, hand write those titles and descriptions because you're converting people from the search and nobody knows a human better than a human and can do those turns of phrase and things like that. If you have a thousand blog posts that don't have descriptions, sure, use it for that. Use it for ideation, use it for newsletters or outreach emails that don't matter. We don't need a human touch. Don't do it wherever you need a human touch. Never do it without human review because they make things up. There's a lawyer that's losing his license because he allowed ChatGPT to write his briefs and it just cited things that didn't exist, law that didn't exist, because it's not a search engine. A lot of people mistake it as a search engine. It's not. A search engine goes out, you put a query, it goes out, it comes back with the data matching your query with all the ranking vectors, everything else. But large language models, all they do is the most likely predicted word in a sentence. And you can change the temperature to get a less likely word, but that's just the most likely word.

So I'll see people in our industry go, I did keyword research with it. You can't do keyword research with it. It's not doing keyword research. It doesn't do analysis. And there are plugins that might be able to help you using the large language model to pull in that data and rewrite it for you. That's different. But you're not doing keyword research if you go to chat GPT and do say, give me keywords. It's just not. The other thing people have to be careful about is not using it for direct content because although Google said we're fine with AI, they said we're fine with AI with lots of caveats and John has been going around for the last three to four weeks saying AI is a, is an anchor on your site, AI just rehashes and regurgitates models regurgitate what's out there, which is true. They can't go outside their training. They can't create, they can't make up something new. They can only do what they were trained on. And what does Google need with a trillion pages to say exactly the same thing that everybody else does in just a different way, right? But also the helpful content update is directed, not at AI, but what AI can't do. So some people say they're not detecting AI. You're right. They're not. They can, but it's too big of a problem. Large scale, right? Trillions of documents, trillions of queries. It's too large scale.

But some of the things in the helpful content update are things like originality, but our language models can't do originality. As John said, they just rehash the web. So if you're trying to make your content useful and helpful and add human experience, you can't do that with a large language model. We'll, and some people are like, oh, mine rank really well. Sure they do, just like when Penguin came out, not every site that bought links got hit, but a lot of the sites that bought links got hit. And I don't think if you're, if they think you're using AI that you're gonna be able to come back that quickly, you're gonna have to rewrite it all. So anything that's search engine facing that you need to rank, don't do it with AI use it for ideation, maybe. But or if someone's going to edit 90 percent of it, but they can detect how the AI writes because I hope you don't mind me quickly mentioning just because the guest is on our podcast when I do. Okay, cool. So there we do have called you Jim Hedger. And there's a guest, Gavin Klondike, who did, he's from the AI village at DEF CON. He's an AI researcher, a large language model researcher. And he explained, like, how they can detect it. And one of the things is called burstiness.

So humans write in different patterns. And we use mixed metaphors. And we do things that are just not grammatically... Like always 100 percent correct. It's a burstiness, like we write short and then long and short and then long, or maybe a little bit and then a very long paragraph. So it's not always the same. Patterns are not the same, but AI writes in a very predictable pattern and with very predictable words because they use the most likely word. Google does have came out in 2022 I believe it was, an algorithm that does look for content that's being rewritten using AI and is published again. So they have a specific set of algorithms for that, so they can detect AI, because otherwise they wouldn't have that algorithm, right? And that was an article in Search Engine Journal, I think, or Search Engine Roundtable. So just don't, you don't want to use it for that, because you don't want to take that chance. Like helpful content evaluations are like 90 percent 80 90 percent and you're out of the right. You're out of the rankings for two to three months They'll sandbox you now for that So there's no recovery for at least three months and then after that you have to wait for them to recall Rescore and then redo the algorithm. So you're probably talking six months down and it's applied based on the percentage Of pages it found not helpful. So if it's 10%, maybe you'll only lose 30 percent of your traffic. And these are just arbitrary numbers. But if it's 50% of your pages, you might lose 90% of your traffic. You have to rewrite all of that with humans. And then you have to wait. And you're going to be waiting probably five to six months before that gets re evaluated. And small to mid sized businesses can't afford Google Ads during all that time, right?

So that's why I say be careful there. As far as the large language models themselves, Be very aware that they hallucinate, they make stuff up, that's the real term for machine learning, I know you know that Mark, but if someone doesn't know that's literally what they call it. So just make stuff up because if it doesn't, it's not a search engine, so it's not pulling back information. It's just the most likely predictable pattern, also, there's no copyright AI. So you can, in the EU or the United States, as of two weeks ago, in the EU I know, I haven't checked since, in the United States as of this week. There's no copyright. So whatever you put up there, you do your images, or you do in Discord, on MidJourney or Dally, and then you go to chat. jpg and you put all that up, anyone can take it, and you have no legal recourse. So you don't want to do anything that's important like branding. You want to brand, create your logo, with, Majorni, and then create your branding information, language, LGBT, because anyone can steal it. And there's, it's not stealing, because there's no copyright. And then also, just be very wary, they don't do citations. Bing and Google have separate processes. That put in the citations at the bottom or put the links in to the SGE in Google or the Bing chat. That's not the large language model. So if the large language model gives you a citation, it's just making it up. Now, maybe it's right because maybe it saw it enough times that it's putting it in as a predicted citation. But it's also probably very wrong.

So the most important thing I say to anybody working with these is human review of everything. Make sure, and also make sure you do human review for biases. There's, Bias is based on gender and race and age and just stereotypes. Like one guy, I was on a panel about AI this week, and he did a search for plumbers an AI. Brought back the images and they were all like bearded men in their 40s with potbellies. He's maybe I guess so you want to make sure you want to really be careful about the problems of homogenization. You ever notice a mid journey unless you ask it every woman looks like some supermodel. Like supermodels, right? And all the men looked like they came off of a, romantic cover storybook thing. So romantic novel.

​

Mark A Preston: When obviously AI content tools started to become big, and Lots and lots of people jumped on it and think, Oh we don't need this content people anymore. We can just, for 20, we can write all the content we can use it. And I think it just the amount of content being published suddenly expanded massively. And I think obviously Google had to look at this and think we need to do something about it. We can't just, keep. indexing the same old stuff over and over again. But one thing I've noticed myself is it's the sites that's done better out of everything are sites that's a real business. They've got the EEAT fine, it's a real business and I found the sites that what I said, content sites are the, but there's no like specific brand behind it, are the ones that sort of been getting hit. Have you been seeing the same sort of patterns that basically, I think Google is looking for genuine businesses.

​

Kristine Schachinger: Yeah, I haven't seen this just because I haven't done a lot of research on what was hit from the HCU. I've heard people talk about it, but I don't, I haven't seen the actual but that makes perfect sense because that follows the HCU guidelines, right? And they want to, so right now, I will be honest, I think Google search results right now are horrifically awful. I was looking for the regional manager of my apartment complex yesterday. I got stuff from 2001, 2008. I got stuff in other states, even though I put Las Vegas, Nevada in the search. So there's something really broken. Right now with Google. And it's been broken for a while. It's been broken for like at least five months or so. I, but at least it's gotten a little better. I used to get all these weird ru sites. With nonsense in them. And it was like a general search. It wasn't like something super leisure specific.

But yeah, they're looking for authenticity, right? That's why they're pumping Reddit up to the top and they're pumping up like a user generated content, which we were told for a long time not to do because Google might hit your site for it. And now they're like, no, I turn it all back on. I don't find it particularly useful to get a Reddit result, like they're like, Oh, everyone's searching for Reddit, so now we'll bump up Reddit. And I'm like, I don't need Reddit, I'm looking for my DMV. I don't need people's opinions on my DMV. So I don't, I think they're in a bit of a conundrum about what to do because last year before chat GPT came out last spring not this past one, but the year before John was on Twitter and there are people saying my content has been sitting here for three months and nothing's indexed. And John said it doesn't have to just be good quality content now. It doesn't just have to be good HCU came out. So John was saying useful content. Useful content is authentic, original, has a human touch, has creativity to it. They don't need the same 50 things anymore. Now, maybe medical sites will always be that way because medical is medical, right? So you also have to think about what niche you're in.

But if I'm writing about Las Vegas, And I'm just like, getting chat GPT to write me stories, or I have someone here is like going to every restaurant or going to every night spot and they're writing their personal experience, putting posts into their own pictures, their own video. That's authentic. Authenticity. It makes people trust you. I don't know that Google's just about trust, but they're about that problem of, because whenever we think about Google, we have to think about scale. Like a lot of times people put out theories and I'm like that's great. But if you think about the scale and instantaneous nature, everything has to be returned to you. It's just, it's not possible, right? What is Google going to do? And next year they have a trillion new documents and 80 percent of them are rehashed GPT, Jasper, whatever, I guess SG is not going to let you write written content. What do they do with that? So they have to start creating methods to eliminate that from search. So they came out last week and said, they're not going to index all the content. You may have good content. It's not necessarily going to be indexed. And why? Because you're looking for my belief from what I've read in the ranking systems.

By the way, everyone needs to read the ranking systems. I see a lot of SEOs now go, I don't need to know algorithms. No, you really do. And the Quality Rater's Guide has zero algorithms in it, has zero ranking factors. There are no ranking factors in the Quality Rater's Guide that everybody's using. And EEAT isn't a quality, it's not a rankings actor. There's no ranking factors in Eat. That's just a good guide on how to write content. Google now publishes the ranking systems and what they do. So go read those and you'll see that basically from reviews. And HCU, they're looking for authenticity, originality, creativity, and human experience. Those are the four main things that travel across all the algorithms. So like you said, a business, a real business that has like real reviews and real videos of its business and people posting pictures and things like that should do better. While they work all this out, doesn't mean it's always going to do better. And AI content and rehash things are going to work for a period of time. But this time next year, since the the issues of machine learning algorithm that's running all the time, it just updates. That doesn't mean that's the only time it runs. Next year is going to get really good at detecting inauthentic, rehashed, algorithmic, AI content, probably. I tell my clients right now, keep your writers, in fact, invest more in good writers because you need that human touch. At least that's what I believe from what we're seeing and what I read in all the algorithms and what I hear John say. Yeah, it makes perfect sense.

​

Mark A Preston: I do feel as though that the investment required now into SEO, because of what's happening, that maybe a lot of businesses are going to fall by the side because they just don't have the level of investment required, to really make a difference and really compete online.

​

Kristine Schachinger: I think it's possible, but I still think the game is pretty much the same. In fact, if you're a small business, you might even have a hands up. As you can walk around your business and take video, you can have like a writer, like one writer who does all the voice and tone of your brand and has that creative human touch. So it is possible. It's actually the strength of the link graph to the big businesses. They'll never fail. But, oh, sorry, cat walking behind me. But the smaller business may actually have some advantages, because they can take hands on approach to creating that authentic content. So it is possible. If you feel you need a team of 20 writers to cover your business, that's a different thing. But if you're a smaller to midsize business, you only may need one or two part time writers to write your content for you.

​

Mark A Preston: Authorship, somebody owning a piece of copy, right? So if you've written a piece of copy, you put it on the site, somebody then copies that content and posts it on a higher authority website, then they're basically stealing the authorship from it. And I've been hearing that quite a bit at the moment. So when it comes to the copyright and ownership of a piece of content, how's Google getting around this?

​

Kristine Schachinger: They never have. I've worked with a lot of news sites and what I don't understand is if they can get Twitter a fire hose, so they get Twitter content instantly. Why can't they give news publishers a fire hose to let them publish their content to Google? Google marks it as the original piece of content and then anything else that comes up with its Writing or it's author or whatever, just doesn't get treated the same. But they don't do that. I don't know why they don't do that because right now, technically they can't do that. So Google doesn't really try to beat that problem. They don't try to fix that problem. They told everyone to put canonical on their syndicated content. Right now they tell everybody, don't syndicate. They're not gonna stop syndicating. That's not gonna happen. That's not an answer. But so that's always been a problem. It's always been a problem that cites. They don't even have to have more authority, they can just have a bunch of fake links to them until they get caught. So there's a lot of sites that get scraped by garbage sites, right? And they have tons of bought links to them and they rank for a while and that person doesn't care because they're just running ads. That's all they cared about. Google, it's just not a problem that Google seems to be willing to solve, unfortunately.

And it does get into the myth of authorship though. So the author patents were from Google Plus. And Google Plus at the time, I did a whole bunch of articles on this back then. It was an identity platform. And it wasn't just Google, it was all over the world. They're going to get rid of, they're going to get rid of passwords. It's called the, in our country it's called National Trusted Identities and something in cyberspace. And STIC. And they're going to, and Google is one of the first approved, there's a White House blog about it. Companies to do this. And so Google Plus, remember everyone had to have a log in to Google, like you couldn't use a Google product without a Google Plus, that's because it was your identity that they were tracking. So what they were going to do is they were going to get rid of the password, you'd have an identity provider, Twitter was going to try to be one, Facebook was going to try to be one, Equifax was one of the first approved with Google, PayPal was identity provider, the IDP, would validate you to the website. Are you over 13? Are you able to use this platform? Are you over 21 for an alcohol site? But they'd also track you and follow you. Unfortunately, they also have all that data and no real laws to govern it. They'll just read the ULA and that was supposed to be the only protection. And then Snowden came out and then all stuff went away. I don't know if they're related, but they felt like at the time they probably were.

So that's where that author patent came from when we used to connect ourselves to websites as authors. And they knew who the author was, they also knew a whole lot about you, they knew your schooling, they knew your jobs, they knew tons of stuff about you because everything you used in Google was now run through Google That died with Google There was, there's no author patent anymore like that's working. And if you think about it, it would take It's so difficult for Google to keep up with every author in the world, in every language, in every dialect, every pseudonym they use, every time they publish a piece of content, right? So what's much easier? They can use entities to know if the article, compared to the other articles in the corpus, writes with similar expertise. That's where expertise comes from. So it's I know I'm taking a tangent here, but just when you mentioned authorship, it's because there's a lot of confusion in the industry that there's somehow tracking every author, which doesn't make sense. That's scale. Remember, we talked, I said earlier about scale, right? So they use the entities to know if you're writing about this with expertise. That's what they do. There's a six billion dollar fake science industry that online looks completely real. There's a talk at DEFCON about it. People can look it up. And they have conferences, and they have schools, and they have pictures, and they all look like they're scientists, and not one of them exists.

And they put out fake science and Google would have no way of knowing that they were not real because they're so well done So doing authors is like a dangerous thing. Also an author can have many names, right? I could just put that, you know Something this author just wrote this or right now the problem on Amazon people are using chat TPT to write in the style Of like George Martin is Lord of the Rings. I'm not one of these. I'm sorry. The big, most biggest show of all time that I didn't watch. The one with all the dragons and stuff. Why can I not pick the name of that right now? Yeah. Anyway, so there's a bunch of authors, big name authors, where people wrote in their style and chat to BT, put it on Amazon and sold it under their name. Saying that they were them. And we're talking about huge authors. So they're now suing OpenAI because of this, right? So authorship is not a way that Google checks the legitimacy of something. They use the expertise. But when it comes to who published something, they don't track it well at all, and they don't seem to care to fix that.

​

Mark A Preston: So if, say for instance, I wrote an article on pets, and publish it on a pet website, It's because, doesn't Google relate me as a person, as the author, to SEO because that's what I'm known for and talk about, so it looks a bit odd talking about something else and publishing about something else. Is there that, is there some sort of connection between personal branding? And what do you usually talk about in the topics you talk about?

​

Kristine Schachinger: Not in that way. I think about all the writers who write for newspapers and write about things, different things every week, right? So it would be like, Oh, you newspaper author, we demote you because you wrote about the local fair last week and now you're writing about government, right? They write about different things. Authors write about different things. Writers write about different things. In the reverse though. Two things. One, if you're a really well named author, because I've seen people say, Oh, author works because we got all this. Increase in search, right? We use famous people. Sure, if you have famous authors, their entity knowledge graph is much bigger. So they'll come back for more terms because they have much more that's related to them in the knowledge graph, right? So they know that you write with expertise already, but they do know that this person's name, but also I can put that person's name on any article. There'd be no way to prove that they didn't write it. It wouldn't be, it wouldn't be a trustworthy signal for Google. As far as you writing about different things, no, but if you're writing on a site that writes about all these different things, like a guest post site, then that is a signal to Google because the site has no niche. The site has no definites, right? It's Oh, we wrote about puppies and then we wrote about. Flags, and then we wrote about medicine, and then we wrote about, because we're guest posting blogs, right? So that stuff they do know, but they shouldn't devalue anybody because they wrote about pets one day, because maybe I suddenly had a life changing experience with pets, and I wanted to write about that as an author. Why, it'd be wrong of Google to devalue people for that reason.

​

Mark A Preston: Can't they relate somebody as a specialist in a certain topic dependent on what they've published online?

Kristine Schachinger: If you're a doctor that's done a lot of research and Google knows you from Google Scholar as being cited in articles, yes. But they still don't, you don't get ranking value out of that, but Google does know the scholar, the scholarship of that because they do have Google Scholar and they do have citations but I could also be a grad student that's published the biggest piece of research that ever existed and I have no citations anywhere. Google's not going to publish that, right? So it's not based on your name or your identity. The entity graph is, so the amount of. Search you match based on your entity graph, but your actual name, what school you went to, what papers you wrote. I mean I could forge all that. We know famous scientists and doctors who have forged their entire histories, right? So it's not a reliable thing, and it's a hard signal when they can just use the entity graph to do all of that stuff.

​

Mark A Preston: All right. SGE. What's the future for SEOs regarding SGE?

​

Kristine Schachinger: That's a hard one. Okay, so this is my personal belief on how large language models work, and because Martin couldn't give me an answer on how theirs was working, because I asked him and three weeks later I asked and he still didn't get an answer he was allowed to give me. I don't think Google is using a true large language model like The reason is when Google said they were coming out with SGA, the experts in the field said it would cost a thousand times per query to write a Chachapati style for Google. Also, we see a lot of direct ripoffs of content, which doesn't happen often in the other large language models because they're just writing a predicted pattern of a sentence. And you'll see a lot of SEOs are like, Google took the content from here and from here. They also removed a whole bunch of rip snippet types. My feeling, it's just my feeling, they're rewriting snippets, featured snippets, things like that with the large language models, but they're not going out and fully generating like a Bing is. Bing ran out of processing power and so did OpenAI. They were literally not... Getting to the point, unless I can ever say the name of the company, NVIDIA, had not come to the rescue, they would have run out of processing power to do that kind of language model. Now imagine Bing has what, like 4% of share? Imagine like 96% of the world doing search on Google with a language model that Bing ran out of processing power for, right? So I have a feeling there's a little bit of... I don't know the right word. Is machine going on at Google?

But as far as appearing on the top, we now saw last week, they've shortened the box. Do you want to generate? And they've shortened it. If you're in the labs, are you in the labs? No. Or can you get it? I don't even know if you can get it, right? Because I know in some countries you can't get it. No, not over here. Okay. Originally it was like this big, long thing. Yeah. Four ads on top, and a PAA underneath. And, I'm sorry, sites are going to lose traffic. There's just no, there's no way that's not going to happen. But suddenly, the box is like this big. Do you want to generate it? And it only opens up like this much. And asks you to expand it if you want to, and then it has the sites right here, and not a PAA, the PAA has been moved down. So my guess is the tests are showing Google their ad money will go shoo, as they still use 85 percent of the revenue from the ads in search for everything in alphabet. Google, but alphabet. So they can't lose that ad revenue. So I'm guessing it's all going to depend on how that works for them. They are putting writing and image creation in the search SGE as of this week, which I just find is weird. Like maybe they know more than I do, but it just seems like maybe they should have a tab that says create and then have people do it there and put ads around that, but when I'm searching for something, we're really going to start writing in the, I don't know, I guess maybe. But I think the big concern for SEOs is, one, we don't know how things get into the box. They have little cards that go into the box. I don't think those are going to get clicked a whole lot, first of all. Just from my experience with user interface. And then your site, how far is it getting pushed down? So if you see big drops, don't assume it's an update unless there is an update. And even if it is, still check. Do you have results that have a SGE box and is the SGE box still small? I'm probably not affecting you that much. But if it's the big one with the four ads and then the PAA and then the first link? There's no way people aren't going to lose on that, and I don't really have an answer for that right now, because we also don't know the final condition of the SGE we won't know until December, the experiment runs out in December, I think it's 12th or something. Yeah. It's scary though.

​

Mark A Preston: Yeah. I'm going to say that because everyone I've spoken to. Nearly everyone's saying it's going to ruin us click throughs because basically SG is providing everyone the answers themselves. So they've no reason to click on any of his websites. But just think from my own personal perspective, that wouldn't be good for Google. Because they make the money from the ads. So if no one is clicking on anything, then they won't make any money. So there needs to be a fine balance there, not to fully answer the queries but maybe it's more the top end of funnel queries that it's going to impact more than the money keywords.

​

Kristine Schachinger: Yeah, I totally agree. And I think I saw them you add an SGE to like the people also ask. I was like, that'd be the perfect place for it. It doesn't really belong in search. It's not a search engine. So therefore, it can be wrong. Like the show I do, Echology Dave Davies was the host I replaced, and Dave Davies name still appears in the SGE, or at least it did two weeks ago, in Google, because there are some other places on the web where it still says Dave's name. But we could also get into serious issues, like medical issues. They're putting the SGE on your money or your life queries, which I didn't think they would do. They're putting them on news queries. There is stuff. It's called the 10 OWASP. The OWASP generally are how a site can be hacked, the most pop, the most common ways. So for a regular site, that's like cross site scripting and things like that. And they're listed and you test the site for those. So there are 10 OWASP for large language models. And how you poison them, how you data break them, how you make them give you information that they shouldn't. But last, two weeks ago, Bing ate itself.

So Bing's chat, the snake ate its tail, right? Bing's chat had a false piece of information in it. About a mathematician, that I've never heard of. A mathematician went, oh, look at this. He's also an AI researcher. Posted in a blog. Bing search picked up the blog information as factual and returned it in a search result as though it was a factual piece of information. It's a really dangerous loop for something that hallucinates. And they don't think they can get rid of that problem. They don't think they can get rid of it making up stuff. I follow a lot of AI researchers on LinkedIn. They say, they have papers that are written. We will never get rid of hallucinations in a large language model because it's just predictive text, right? So I think it's dangerous for Google to put it there. I know they did it because, they put out ChatGPT and then there was Bing. But you know how much market share Bing's gotten from having that there? Zero. Zero percent market share. You know how much the usage has gone up? Fourteen percent. So their costs have gone through the roof. OpenAI, just two months ago, they thought they might go bankrupt this year because of the cost of ChatGPT. Also, these are heavily environmentally what's the right word? It's a very heavy resource load on water and electricity. They're, you thought like crypto used a lot of those things?

This is like that on steroids. It's environmentally not a good thing. So I don't know how Google's gonna go forward with all that. But yes, if the SGE is sitting on the top in the full expanded version, it will destroy Quill. If it's up here, and it's generate it yourself if you want to. I find it annoying. I find a lot of people find it annoying. Sometimes it's helpful, but it's rare that I really need it. I'd rather go to a search result where I can read an article and get more information. It's not just this one piece. So I don't know. It just depends on how regular people interact with it. Because I'm not a regular person that way. But I just, I don't know what you think, but I just feel like it just totally depends on how Google decides to implement it and how destructive it'll be, but if it's fully implemented as an expanded mode, fully written out, whether this big, they literally take up like a third to half the page, it'll destroy the clicks underneath. But what happens if Google destroys the clicks underneath, they're not going to buy ads because they won't be able to afford to. So if you take away all their traffic, they can't buy ads, they're not going to just buy ads to make up the traffic because they can't. And where are those ads even going?

The four above, which everyone's going to blind, you get this blindness, so you'll skip it. That's what happens, that's why Google's always changing the placement. And if you look in the SGA place where they put things, No, it's going to click on those tiny little cards that are barely readable with a little half a title. I don't know. It's really hard. I just think it was, I honestly think Google, my personal opinion was shiny object chasing fear of like big, like everyone taking them over. And it turns out, neither chat GPT or Bing have done exceptionally well with it. And Google probably should have found a really much better thoughtful way to add it where they wouldn't be the parasite. It's always mutually parasitic, right? Google uses your content, they get ads, but you get traffic. If Google is just taking your content, writing SGEs, and then you're not getting any clicks, what's the purpose of Google for anybody but Google at that point? And they're also under a big antitrust lawsuit, right now, in court, this week. I don't know. I don't know what you think, but I think it is.

​

Mark A Preston: I was going to say that stripping it right back away from SEO, I just think it's going to take a lot for people to understand how to change their search behavior. There's, they've had this box for forever and they've had results. And suddenly, if they presented, we all, we, it's, from an outside perspective, if they suddenly see this page and it's Oh, looks like a circus. It's everything, it's, I'm just feeling as though there's a bigger impact here beyond just where we lay things out. It's going to take, I think it's going to take a lot for the normal user to start using the conversational search. Because that means you have to change the whole behavior on how they search.

​

Kristine Schachinger: And how many times when it brings back something incorrect, they decide never to use it again because they can't trust it. And it's going to bring back incorrect things, it just not do that. It doesn't fact check. So when Google came out, was it? Yeah, Google came out and said they're going to highlight things that are commonly known and not commonly said. That's their consensus engine. So for the featured snippets, I know you probably know all this, but just for anyone who's listening. For the featured snippets, they have a consensus engine. And if it doesn't fit consensus, they wouldn't put it in a featured snippet. They'd use it anywhere else, just in the featured snippets. But, they're using it now, appears, and that's the box to show you what's common and what's not common. It's not a fact check, so what happens if I believe the uncommon thing, it just confirms my bias, and then I find out Google was wrong. I want to do with something important, medical or financial or any of the money or life stuff, news. What if, right now, I can guarantee you there are nation states that are data poisoning, which isn't hard to do with large language models, data poisoning large language models to get them to say things that support narratives. When Google returns that result to people, will they trust Google?

Like I could damage Google's entire brand reputation, right? There was somebody at chat GPT. I always love this story because it's so SEO 2000. He did white on white text, right? And this is before they took the break at chat GPT where it would ingest URL. And so he did white on white text. He put a special phrase in there and he gave it the link. And then I don't know how long you waited a couple of days or whatever. And he went back and it was there, right? This special phrase he put in. Now think about brands. Information, news, if you can make the models do that, ingest your information as the SGE, then there's a lot of damage that can occur too. So I don't know what the breaking point for users should be with Google to go, I can't trust Google anymore because I use this SGE box and it gave me the wrong information.

​

Mark A Preston: Yeah, I think that the whole fact checking thing is such a just interesting topic for me, because if, as before, you did a search and you clicked on a website and that were it, but now surely Google has to create something relating to fact checking, just so they don't get sued. Because now Google themselves are presenting the information, not third parties. So if, for instance, I search something and Google gives me something and I go ahead and follow that information and it, damages something, then, there's a whole legal ramification behind that. So I think it's a really complex thing, but I do think that in order for SG to fully become its own entity, that some sort of fact checking must, based on whatever, must be impressed.

​

Kristine Schachinger: And yet they can't. So the researcher a few years ago, where people started thinking Google was fact checking, didn't read the whole article, the paper, and at the end of the paper they said it's not feasible to fact check. It's too expensive. We can't do it. So my favorite way to think of this is there is fish oils, right? Omega, Omega oils, right? Yeah. So Omega 6s are very inflammatory and very bad for you. Omega 3s are very good for you. But in the American Heart Association in the United States and in like other medical journals, it says just take omegas. It doesn't separate out three and six. The British Journal of Medicine says sixes are bad, threes are good. So how would Google determine what was a fact? Both are highly reputable journals, medical journals that they've been published in, but the new information, the less popular published information is don't eat omega 6s, don't eat, they're called PUFAs, polyunsaturated fats, and they cause inflammation. Like we need very little in the American diet is 16 to 1 ratio, and it's supposed to be a 1 to 1 ratio, right? So 1 to 1 is fine. Your body needs that. 16 to 1, highly inflammatory, heart disease, diabetes, all good. How would Google do that?

On every fact in the world they can. They can do the consensus model where they're like, a lot of people wrote about this. A lot of people didn't. But let's just take ourselves out of our history. In the United States, slavery in the 1800s would be a common thing, right? So if there was a Google search engine, the most common pages would be like, Slavery is good, and this is why you should have slaves, right? Obviously horrible thing. And I'm using this as an extreme example, just because, just to make it really obvious, that you can't easily do consensus either. Because what's true today doesn't mean it's right or true tomorrow, literally. Like AI literally changes every single day, right? Try to keep up with AI with fact checking and consensus, right? So that's why I think it is dangerous and just like you said, legally to be putting it on your money or your life queries, because Google judges your money, your life at the query level, not at the site level. So they know what queries are considered your money or your money, your life. Keep it off those. Why would you put it on that? But that goes back to my belief, which is it doesn't belong in search engines. If AI engine. Great, put create put all that there, but it's not a search engine and what you put in that box is going, I just published a graph yesterday from an AI researcher on the comparisons on how wrong they are and like, none of them are good. They're all like over 60 to 90 compared to each other in how wrong they are, how often they're wrong. So it's Yeah, you're right, it's a big legal issue for them. And so I think as SEOs, we just have to follow and keep abreast of what they're doing. So know which way to pivot. Because it's going to change, and it's going to change based on ad revenues, legal issues, and antitrust issues. Those are the three things I think are going to be the big movers for them.

​

Mark A Preston: Wonderful. The time is absolutely rocking on. I could go on forever with these subjects, but obviously we have to call it a day sometime. Now, is there anything... That we haven't covered already that you feel is super important for the people listening and watching this needs to know.

​

Kristine Schachinger: No, I think we covered most of it. I just like to reiterate that just a couple things. One, go to Google's ranking system pages, learn the ranking systems, click in, click all the little blue links for each one. So you understand what the algorithms are doing. Two, the thing about. EAT and the quality raters guide is actually written by Google on those pages that they don't have ranking factors. So they're great guidelines to like how to write like content. They're just not an action manual. Like here's how this is a car not this is where you put the steering wheel, right? So there are a lot of things in there because the quality raters guide is just a QA testing document for quality raters when Google makes an algorithm tweak and some of the SEOs thought it was like the secret to the algorithms and John has Every single Google person has said it's not but it's also in the documentation it says it's not so just be careful not to use that as your only information on how to create a site. Learn the ranking system so that you know what Google's really looking for with the algorithms that rank things. And just for the AI content, Google has the HCU, and I'm sure they're going to have more and more algorithms that are going to try to keep out regurgitated, rehashed information, and it doesn't have to always be AI content. You can just do that yourself or with spinners or whatever you're using. But invest in human writers for your content that needs to rank.

​

Mark A Preston: Brilliant. And is there anything that the audience can do to help you and where can they find you and what sort of conversations would you like to have with people?

​

Kristine Schachinger: Oh, that's nice. How could they help me? Hire me for audits when you need them. I love doing site audits. I'm really good at finding things and I'm really good at recovering sites. I have a very high success ratio. And then LinkedIn is a really good place to find me for business stuff. My Twitter tends to be more personal in nature. LinkedIn. It has kschackinger seo, cause I... Was like just playing with that a while back and it stuck and I can't get rid of it. So I was just you know, it's like I'm the only Khristine with a K Shackinger on the internet if I commit a crime or do something bad, I'm gonna have to create another me to hide me So it's just S C H A C H I N G E R. You can find me so and webcology our podcast Wonderful.

​

Mark A Preston: Many thanks for joining us today It's been super interesting.

​

Kristine Schachinger: Thank you for having me. I really appreciate it Thanks.