Chat With Mark Williams-Cook | Building Future-Proof SEO Strategies In The Age Of AI And Migrations

Hello, welcome back. So it's Sean. This is of course another Chat with Sean episode, and today's guest needs no real introductions. In fact, I'm thinking about to say, if, if someone's been named the most on this channel by other guests, it's probably Mark. But anyway, we'll do the intro and then we got some like good topics to go over. So Mark Williams-Cook has worked in SEO for over 20 years and is the director at the agency Kanda. And is the founder of the very popular tool, AlsoAsk. So yeah, thanks for joining me today, Mark. Um, it's a real pleasure to have you.

Yeah, it's great to be here. And congratulations. I saw you hit, I think it's 25,000, was it subscribers on the channel? Yeah. It's massive since I came on near the beginning, right on a, on a panel. Um, and it's been really cool. Yeah. Seeing you grow and in the intro there, seeing so many of my friends like Cindy and Julia and Ferry and stuff.

Yeah, it's been like, that's what I love about this community. Like, it, you know, the people, don't matter how, you know, where they're working or wherever, they're so accommodating and willing to have a chat. And some real big names like yourself, like all these people that I thought no chance they would want to speak to me, especially in the early days. But I didn't, yeah, I never expected this sort of growth, but it's been fascinating and it's been such a learning curve, uh, for many reasons. So yeah, thanks for that. I appreciate it. It's a testament to the good job you're doing. Cheers, man. And yeah, we've got some good topics to go over, some very generic ones, I guess, just to get your feel for what's going on. But, you know, on that, what do you make of the current search landscape? Because it's such a, as you know, with like what the kind of posts you do about what's going on in the industry every day, you could be writing like a whole bunch of things that are happening. So what do you make of it and where do you predict things going for the rest of the year?

Yeah, I think there's a lot of kind of fear and FOMO in the industry that's starting sometimes unwarrantedly from the top down. And that's around the biggest changes we've seen, I think, around attribution and reporting of particularly SEO. So that's changing user behavior and, you know, zero-click searches, people learning things in AI overviews and on Perplexity, Claude, you know, ChatGPT, those kinds of things. So the crutch of the reporting that we've had for so many years that made it relatively easy, at least to demonstrate growth drivers, you know, this is how many clicks we've had, this is how many visitors we got, isn't this nice and easy to mention? That's, you know, very quickly evaporating. You know, in some of our clients, we've seen people with stable visibility lose like 30% of clicks. And interestingly, not The revenue. So the revenue is still coming through. So the good stuff is still happening. It's just happening in places we can't track. So I think there is a lot of fear around working out, you know, what are we spending? What are we getting from that? And that filters down to practitioners who are getting a lot of pressure into, you know, tell us what's going on. How do we measure it? And then of course, in the old adage of marketing and advertising, nothing sells like fear. So we have a whole new generation of different sizes and shape tools that are being sold as, you know, wonder pills to fix this ailment we all have. Some are helpful, some are less helpful. I don't personally think there's a great solution yet. And I'm not saying for that reason the tools are bad. I think it's very difficult to have a satisfactory answer to some of this, but that change is, I think, staying and it's going to continue in that direction because the type of things people are doing online means they're getting similar results for less brain effort. Again, for better or worse, there's all kinds of interesting conversations around what happens when you start switching bits of your brain off and relying on AI. But that's one thing we've always seen with technology, which is if you can give someone a way to do a thing that is less hassle or less effort than the previous way, eventually people will migrate to doing it that way.

Yeah. And actually, you're not the first one to say that where they've seen lots of You know, we're not lost, but seeing drops in traffic, yet the bottom line hasn't really changed. So yeah, I dunno what that really says in many ways. It probably says that a lot of that traffic was like not that meaningful anyway. It's just kind of like people having a look around and trying to find those answers that now they're finding on AI overviews and stuff.

Yeah.

But yeah, in the sense of generally, like, where do you see it in terms of Obviously there's a battle going on, you know, will Google go full AI and such like AI mode? You know, that's the question on everyone's lips. Well, how long before AI mode becomes the default, if it does at all? And will ChatGPT continue? You know, they're going up and down in terms of like their market share and how popular they are. And it's, and then obviously they've had been in the news recently. A lot of people got turned off from them and went to other models like Claude. So it's like, yeah, where do you see that all playing out?

Yeah. So I've said for a long time that I think, you know, OpenAI versus Google, Google's going to just slowly tear chunks off them and eat them for breakfast at their leisure. I mean, one of these companies is making like record-breaking profits and the other is burning through money at a record pace. And again, where I always said that the real challenges for the AI space would be Apple, Microsoft, or Google. And the reason for that, I believe, is to make helpful AI assistants, whether you're talking about like agentic systems that will go and do things for you, or even kind of chatbots on the very basic level, they need personalization to be effective. And The people that own the hardware, i.e., Android, iPhones, or even to an extent, you know, Windows computers, and actually some of the destinations like Gmail, YouTube are the places where you can get really in-depth kind of imprints on what that individual user is about. And doing it in that way on those devices, Also gets you around kind of a lot of privacy issues as well, because you can ring fence it. And OpenAI has not got any way to get that kind of data. You know, they have very basic limited memory within the chat of things it kind of knows about you. They have their own browser they could potentially pick up stuff from. But when you compare it to someone like Google, you know, that's got Chrome, for one, they've got Android on millions and millions of devices, hundreds of millions of devices, probably billions of devices. It will be billions. You know, the search engine, YouTube, you've got Google accounts across lots of major services. Google said that the personalization is going to become the product essentially. And I think that's very close to where the future's going there. So yeah, and we sent out some data from SimilarWeb on Monday on our core updates, which was the desktop usage of the major LLM platforms. And that showed ChatGPT is fairly stagnant at the moment in terms of traffic. And Gemini is just on still this growth march. It looks like they're around about a third, approaching half, from memory of the kind of desktop. share. And the models Gemini have got are fantastic now. You know, they've closed that gap on a lot of the independent tests. The Gemini models are beating OpenAI models as well. And yeah, OpenAI's kind of politics recently obviously hasn't been particularly popular. So if I'm betting on the company that has all the history, the infrastructure, you know, hundreds of billions of dollars at their disposal versus the one that's running off funding and has none of that, I'm going with Google 10 times out of 10. So yeah, I think from a consumer's perspective, we are going to see obviously more AI in the future. I think agentic will be the way things go, i.e., they are doing the legwork for you in terms of searches and retrieval. But I still think there are some pretty major hurdles to get over to get these systems to be reliable? Because we've also seen a couple of U-turns with the e-commerce stuff. So like ChatGPT being an example where it was very much lots of bells and whistles over a year ago that they were integrating and you can buy directly through ChatGPT. And then they went very quiet on that. Then it kind of came back with Shopify integrations. Now it's kind of taken another step back because they're just not Reliable enough and secure enough to do those things at the moment.

Great points. And yeah, I think with OpenAI, I could see them more focusing on just like, they obviously took up where Anthropic kind of turned down the, well, they didn't want to do some certain things and OpenAI sort of jumped at the chance to take the government contract. I can see them kind of living off that sort of thing where they're working closer with organizations like that, like the governmental sort of side of it, rather than, yeah, I think the general part, I think Google will dominate that sort of search side of it and maybe a few others can pick up the other pieces. And, but when I had Eli Schwartz on, he was saying about how similar to what you are saying, obviously Google has all the infrastructure, they have all the data, they have all these different things and Meta sort of have He was talking more about ad platforms. So obviously Google has a very strong ad platform for all sorts of businesses, small and large. And Meta sort of have a similar thing set up with obviously Facebook, Instagram. They've obviously got WhatsApp. So he thinks that Meta could be like a dark horse in all this as well. Obviously they've just bought, I think I saw this in the news this morning, they just bought the kind of AI agent-based Reddit website and Malt book, which was interesting. Obviously don't know what they want to do with that, but it's just, you know, they've got their wearables that isn't really taken off that much yet, but who knows? But that's what you were thinking. I dunno what you think about Meta's position in all this.

Yeah, I guess from the, I'm not as close to the paid search side of things anymore. because I don't work in that area now. I mean, Meta for me, from when I did work with them, that just as a personal gripe, that Meta ad platform is like the worst thing I've ever used. I saw him, I'd love to get like to watch a livestream of Mark Zuckerberg trying to like set up some ads on Facebook to see how he gets along. I mean, they've taken some atrocious bets though, that whole What were they calling it when they tried to launch their whole like VR thing?

Do you remember that? The metaverse.

The metaverse. There we go. I tried to scrub that from my memory, but I just kind of watched that unravel, you know, jaw on my desk. Like, this looks like 10-year-old like VR stuff that we've had better stuff in video games for years and years. And there's just like no good use case. For this. I don't know if— so I've got a Quest headset now, and I don't know if you've ever tried working in VR. So they have like an app where you can do that, and it's just the most— and this is coming from a gamer, right? You know, this is my youth. It's the most fatiguing thing trying to work in VR because everything's not quite the right resolution. Things kind of flicker in and out. It's just the worst. So I'm not surprised many of their initiatives haven't kind of spiraled out. I have been using their model, their LLaMA model, like locally for some stuff, which is decent. But in the AI search space where I'm in, I just can't see them getting their kind of shit together to make it because these other players are like focused and moving fast on all this. And it just seems that Meta is just swimming through their absolute fortunes of ad revenue, which is great because they own such a big chunk of that, what I would refer to really as the social web rather than the open web. Because I'm primarily dealing with, you know, from the organic side of things, the open web and the indexing surfacing of that. Whereas Meta to me are really kind of the lord protectors of, like you said, the Facebook, the Instagrams, the WhatsApp, which are all ring-fenced communities outside really of the open web. They let people in and look through the windows and stuff in terms of Instagram made that change like last year about having posts indexed by default. But for what we're talking about, i.e., I think a future of agentic search, AI assistance, Meta will have their part in that. And with respect to if anyone Ever listens to this, I always feel like they'll be more like, they'll have their toys in that area, but they're not going to be driving the kind of commercial side.

Yeah, it's just interesting, like Apple as well, you know, Apple got their own plans for AI, it seems. And they've got some things, they've signed some agreements with different people at the moment. And it seems like because of this whole craze that's taken off, especially in our like tech bubble, people are using their devices like the MacBooks to kind of launch all these sort of, um, you know, tablets and stuff to like use with these AI agents. So yeah, they're taking advantage of, from it, from like a hardware side of it. So yeah, it's just really—

Siri to use Gemini, if I remember rightly.

I believe they signed a deal recently.

Yeah. And to me, that was like the big competition for Google then is just putting their hands up and being like, we'll just go with you, you know, let's not fight. So that to me was when I was like, yeah, Google is going to win this now. If they— because there must be some agreement there, or it's either going to be a huge kind of cash agreement, or if that model can do anything with those devices in terms of training, they've basically got between Android and Apple, that's like, what, 99% of the phone market covered?

That's what I mean. Yeah. Like you say, Apple, I think initially, and People can correct me, but initially signed like a very short-term deal because Apple said they had their own plans to do their own sort of AI, you know, stuff. But then I think then they just signed a longer contract because they were like, I think like you say, they probably just want to focus on the hardware, you know, selling the phones, selling the tablets, selling the MacBooks and all that. And they can just lean on like to Google to do the stuff on the search AI side, you know.

I think we'll see that with a couple of big companies. I think Microsoft's a really great example of this. So Microsoft just used to be about Windows, right? That's what Microsoft was. It was like, well, you know, Windows 95, 98, whatever, this is Microsoft. And then they got a little bit of a bad rep over the years, you know, that they went through the villain arc of everyone hates Microsoft and Windows is rubbish. But then they diversified their revenue so much. Microsoft is, I'd say, kind of popular again in some ways, but they've moved way back into a lot of infrastructure support stuff like Azure and all this stuff and making huge amounts of money. And it almost seems like Windows is just like an extra thing now. And I wonder, like you said, if you have these huge companies like Apple and Google kind of end up deciding where they're going to sit in this ecosystem because it Probably isn't the best game plan for both of them to necessarily fight each other on all fronts, especially if one's got, you know, a clear advantage over the other.

Yeah, it's going to be very interesting. And in terms of, uh, we'll move it on. In terms of the SEO strategies these days, like, it seems like people are in a bit of a limbo between doing the traditional stuff we've always been doing Versus trying to do stuff we've been told perhaps works for like large language models and that kind of thing, AI overviews and all this new stuff that's come across. So you're building that SEO strategy now. What would you be focusing on? What do you focus on yourself now? Um, has anything really changed or are you kind of just doing the same stuff? And yeah, interested to hear your thoughts.

This is actually, I find this topic so interesting and this is what I'm going to do some of my conference talks on this year, which is about decision-making in SEO and how you decide what route to take with your strategy. And I'm— I think it's worth the kind of caveat that I'm always talking from a perspective of someone working at an agency who is working with companies that want to be around for a long time. And With that in mind, everything we do is focused on revenue, basically, because that's why pretty much all organizations exist. Even if they're like charities, they want to talk about how can we get donations or how can we raise money. So the strategies that we plan, we're trying to get the highest expected value for the client, which includes a time period. So to take 2 extreme examples, there are completely black hat things that I know definitely work. And you can rank very well in weeks and make lots of money doing that. But you will get stomped out, you know, within a month or 2 with some of these techniques, right? So the best, say, expected value over 10 years is one where you are having incremental kind of value. And I had, I saw a really famous gambler called Daniel Negreanu. He's a famous poker player. And he was talking about if you played paper, scissors, rock with someone, and you knew that your opponent's strategy was they were just always going to do rock, that's what they were going to come in with. And they're like, I'm just gonna do rock every time. And his question was, what's the optimal way if you were to play them forever? What should you do? And the most common answer people give is, well, I do paper every time because they're going to do rock every time. And he's like, no, because they would notice after 10 or 20 times that, okay, it's not working. So I have to change what I'm doing. He said the optimal thing to do would be to do all 3, but you do paper just slightly more, imperceptibly more so they don't kind of notice that they're slowly losing more and more over time. And then that's the way to get the best value over time. And you can apply that to the strategies you're using with Google and the tactics really, which is that there are things you can do to kind of step on the scale with Google to give you an advantage. There are specific ways you can do things, you know, in terms of on-page optimization, internal linking, external linking, You know, all these little things, because that's what optimization is, is the sum of all these things. So I think the best strategies are ones where you identify the things you can do that are going to last and also things that build equity for your organization if search didn't exist. So my kind of golden rule for SEO is there's nothing you should do for SEO or geo or whatever you want to call it. that harms the user experience. Okay, there's, and there's no exceptions to that, in my opinion. Anything else you do ideally should have, should be building equity for the company. So if you're producing content, if search engines didn't exist tomorrow, or LLMs didn't exist tomorrow, that content is still helpful for your users, you've built equity. If you make site performance improvements, that's good for the users, you know, on your site. If you've done digital PR, And everything stopped looking at digital PR mentions, it's still good because you're in the right place where your potential customers are reading and they're finding out about you. And the reason why I like those kind of— that strategy and those kinds of tactics is generally they align with the goal of the system you're optimizing for. And by that, I mean, if we're talking about Google, if we're talking about Perplexity, ChatGPT, all of those things themselves are companies that exist and have goals and targets. And if we take Google as a nice easy example, Google wants to rank good pages or show good answers, basically because it helps them sell more ads because more people use their platform. So if you were doing things that in Google's eyes take away from that, you eventually will be sorted out by their algorithm. And there's other things in there as well, you have to consider, such as if your business strategy gets in the way of Google. So a really obvious example, a painful one, is maybe things like even flight comparisons. Google's trying to grab that away from certain sites. Certainly affiliates have definitely been hit over the years because Google's basically like, well, actually, I can make more money doing this myself. So I don't care that you're doing it. So making a strategy now is very much for me about long-term alignment with the platforms you're on, making sure you have a strategy that builds equity for your organization online, digitally, however you're doing that. So you're less exposed to changes in algorithms, to changes in user behavior, because all of these systems are ultimately trying to measure the human satisfaction with the thing you're building. So if you're trying to build to satisfy the machine instead of the, the end result, the human, you'll likely end off at some point hitting a— hitting some rocks. And it's so tempting to do some of these tactics that do work well. And it's one thing I've noticed, if you look at people in the industry, the ones giving the advice, about these longer-term strategies are normally people that have been in the industry longer, because we've all had to try and recover from websites that come to us with all these crashes. We've got another wave of them at the moment, which is the— I've had 4 or 5 organizations reach out over the last month about essentially losing, in some cases, over half their traffic. And I'm talking big websites here, because they had lots of programmatic, like listicle type content that was very self-promotional, you know, where they're putting themselves at the top of the list, because it seemed like such a great idea. But now Google has, you know, rightly decided, actually, that's not really what people want. We're not aligned here. So they've just been rug pulled. And now they're trying to figure out what they should have been doing in the first place.

Yeah, I completely agree with all what you said. And I think, I don't know if you agree, but I feel like the way we're going now with There are some bumps in the road, let's say, like listicles and things like that, which I have tried to correct recently. But I did feel like this whole AI integration side of it, things like EEAT coming into play, I feel like we're gonna get closer to where it should be. In fact, in terms of quality content, if things work the way that they could work, I feel like we could be closer to actually better quality content being surfaced more than it was before. I feel like perhaps it could be less gamified than what it has been in the past, but at the moment we're in a bit of a, like, an area where, because it's so new and I don't even think the platforms really know what's going on with it that much and they're working it out, there's still a lot of room there to gamify it. But I do believe if it goes in the right direction, we could be in a better place for users, you know?

I mean, that's the hope. I think the newer platforms on the block are in for a rude awakening because I don't think they realize quite how ferociously SEOs and adjacents will go after flaws in their platform. Google's been pretty chill about the whole thing. I think it was Liz Reid was doing a talk the other day talking about AI-produced content and basically saying, yeah, we've been dealing with you know, spammy produced content, spammy, spammy. I love that word I just made up. Kind of spammy produced content for 2 decades now. And she was basically saying, this is no different. It's just the way it's being produced has evolved and we'll cat and mouse it. And I've had the opportunity as well, sometimes to speak to Googlers about some of the engineering teams that they work with. And we used to, maybe for some of your listeners that are newer in SEO, so we used to have a thing called Toolbar PageRank, which Google provided us, which would basically give us a 0 to 10 score of the PageRank of any page you were on. And it produced this nightmarish hellscape where everyone was buying and selling links because you could you know, objectively see what PageRank Google had assigned it. And it had, you know, we created whole link marketplaces and stuff. And obviously, they decided that wasn't a good thing, because it was giving SEOs too much information, and it was encouraging them to do things they didn't want them to do. But speaking to some of these Googlers, the people I think sometimes who work on the engineering side of Google who are making, you know, these algorithms and stuff, Are very, very, very far removed from the reality of people building websites, running businesses online from SEOs. And they underestimate the depths SEOs will go to to get an advantage. Because I've been told there have been conversations where the engineers have been like, oh, well, we can just give them this score, like in Search Console. And then the Googlers nearer the front line are like, no, no, no, no, no, don't give them that. Please don't give them that. It will cause chaos. So I think that interests me as well in terms of the LLM-first platforms we're seeing. I don't think they've really kind of got that spam team in place, that all the people who have been in that position before. And it's obvious because there have been several pretty embarrassing examples of people, you know, injecting stuff into ChatGPT and stuff. So I'm super interested how that's going to evolve. And I agree with you. I think personally it's going to get worse before it gets better because the technology normally gets this flow of you get the new tech and it's all great. And then people are hammering it with different, you know, death by a thousand cuts of how can we optimize for this? And then they start finding ways through and then you get that wave of, okay, we need to update the algorithm to harden it to this kind of attack or this abuse. So I think we've still got that first big wave Um, to come.

Yeah, I agree. And, uh, yeah, it'd be very interesting. And it's because there's so many sources of information as well. Like, they gotta contend with— they got a Google Discover that's got its own problems. Then you got the AI overviews, uh, AI mode that, you know, all these things, and they all got their own sort of problems that they're trying to resolve.

It's—

yeah, yeah, yeah. I wouldn't wanna be in their position. And obviously, You know, I've had my own rants on here about AI abuse and health information and stuff. So we won't go into that one again. But in terms of misconceptions that you see like around at the moment, there's a lot of guidance and stuff on LinkedIn. Actually, I saw today that LinkedIn became like the biggest influencer for LLMs, or it's like grown hugely like in recent weeks in terms of influencing large language models. It was obviously Reddit, then they had a bit of a nosedive, and then YouTube became very influential, probably still is now LinkedIn. Um, but there's a lot of crap on LinkedIn, as we know, just as much as great stuff. So what kind of misconceptions are you seeing the most?

I think that whole, that whole point's quite interesting because after Google's helpful content update, I mean, LinkedIn became huge even just for Google. We had loans sites, or just AI loans content on there ranking for tens of thousands of really expensive keywords, just because it was generated content on LinkedIn and Google loved the domain. I feel there's definitely some manual twiddling happening on these LLMs in that they're obviously taking big bites out of Reddit, YouTube, now LinkedIn, and someone's kind of at the top desperately trying to sculpt this into something that's not quite leaning heavily so one way. And I think that's happening because they're still reliant on traditional search indexes to get this information, right? And these sites do surface a lot in traditional search. So it's, it, It's just a cause-effect kind of loop and they're trying to fix it after the results are coming out. That's what I feel is happening. In answer to your question about kind of misconceptions and stuff, again, this is probably going to, because I'm, although my first talk is in like 12 days, I haven't started my deck yet. So I'm still very much spitballing ideas, but I think there's a lot of what I'd maybe rudely describe as faulty thinking in terms of tests and conclusions that happen. And while we are doing business, we're not doing science, as Will Critchlow always so brilliantly puts it. I do think there are a lot of things that people want to believe are true, will do an experiment or a test which will not disprove what they're saying. And then, but it's taken as evidence. That's definitely a thing. So the examples that I've played around with and tests are really, I would just say tests are really hard to do, um, to get any kind of solid conclusion for a whole bunch of reasons. So I very broadly try and test concepts about how search engines are working rather than things like what's a ranking factor for, because it changes on every single site, every niche, you know, it's almost impossible, I think, to do that unless you've got access to huge, huge amounts of data like the main platforms. But to give you some examples, the ones I've done most recently, you may have seen, was I did a test with schema, on-page schema, JSON-LD schema and LLMs, because the reason is I'd seen a lot of advice saying, okay, schema is like a geo thing because it's machine readable, it's structured, therefore it's really important for LLMs to understand your content. Like, okay, fine. And then as evidence of this, I saw someone had put schema on a page. I think it was like a business address they did, and they didn't publish the actual address on the page. And they asked ChatGPT, GPT about it and it managed to get it. And they're like, see, it uses schema. This is proof. So I did another test, which was essentially recreating it, but I'm, I made up some schema about ducks that didn't exist, you know, like waddle style and read number and all this and put it on the page and then asked the LLMs about the address for this company. And again, it was just like, Yeah, this is the address, despite it not being valid schema, which again doesn't disprove that LLMs use schema. Because again, I had people saying, well, this doesn't disprove this, da da da. What it does show is that it's just, to me, what I think is happening there is it's just reading in that case, the HTML content. Like any other text. It doesn't particularly care that it's JSON-LD. It's certainly not doing what schema is designed to do, which is to remove ambiguity by explicitly stating this thing has this value. I don't think that's what's happening. So I was just trying to demonstrate, look, you can't just say a thing and then prove it with a test that doesn't prove you wrong, because I can just make ridiculous tests that do that and then make claims. The other one that got popular because I've done a bunch of these was the cats.txt thing, right? Which was again, so there's llms.txt and like, I'm not even that against using llms.txt because it's not a huge time investment. So again, from the expected value point of view, you're putting in very little for potential gain. Like, fine. Like, knock yourself out. I don't, you know, I'm not worried about that. What I am worried about is I have seen lots of people say it is a geo thing, it is used, and you know, you need to do this. And the evidence of it being used was that it was crawled by LLM platforms and Google had indexed it. So again, I was like, well, I'll just make a cats.txt file that describes Office Cats. And then on the tracking, you can see it gets crawled by OpenAI, it gets crawled by Perplexity. Google indexes it like any other file. So again, it doesn't disprove LLMs.txt is used or isn't used, despite there being, as far as I can see, no actual evidence. But it's just showing that you can't use that as evidence that it's a thing. Because I can just be like, well, by the same standard you're using, cats.txt is a thing. And, you know, this, this went to the point where I had people saying, well, I asked ChatGPT. what llms.txt was and if it uses it, and it says it did. And I'm like, yeah, because it's returning text that everyone's written online saying it's a thing. And again, I published a few posts on like LinkedIn about cats.txt and then showed if you asked ChatGPT and even Google's AIOs about cats.txt protocol, it was like, yeah, it's a protocol that's being newly adopted for cats. I was like, it's not, but it's because they're text predictors and it's just feeding on what people say. So there is that really frustrating, I think, loop of misinformation because once something, you know, there's that, what's that expression about mistruths going around the world 3 times before the truth's got its shoes on kind of thing? You know, that's happening with LLMs. You get a mistruth, it goes kind of viral because it's people want to believe it. So then it gets republished by everyone who, you know, is like, I'm a thought leader, look, I read this thing. And then the LLMs hoover it up. And then kind of all the massive people that want to learn about it are like, oh, LLMs are a great tool for learning. And then they get the misinformation. So I think there's a lot of that happening at the moment. And I'm not saying necessarily right now it's causing huge problems, I'm just saying that way of thinking will lead to lots of wasted time, effort, money, lost revenue, lost jobs. Um, and it just makes the industry look a bit silly in my opinion.

Yeah, you gotta be careful with like, you know, we've got so much to do. You gotta be careful where you spend your time and prioritize and stuff. Yeah, it's a tricky one. And I just wanna get through a few more topics while I've got you here, 'cause this one really interesting 'cause you had this great Spotify series. And I don't know what happened to that because I know Spotify did a bit of a, even with some of mine, it got rid of some episodes. But obviously some of you're really interested in the patterns side of it. I'd been reading recently about some of the recent ones that Microsoft, Apple, Google are all doing very interesting stuff about emotions and how they're gonna try and read a lot more into people's actual You know, how their tone of voice or how aggressive they're typing or their facial expressions and everything like that is going to be very interesting. What ones are like catching your eye at the moment and what people may not be aware of?

So the paying podcast, I just, I haven't added to that, I think since like May last year, just because I'm trying to juggle about 200 balls at once. So some of them sometimes have to hit the floor. So I was using NotebookLM to make little podcasts about them because I personally struggle to sit down and read patents cover to cover. So what I was doing was getting NotebookLM to have like a 10-minute chat about them, which I'd listen to on my commute. And then based on how interested I was, that would determine if I actually bothered reading the whole thing. Patents is an interesting one because I use them Essentially to reinforce my decision-making, because just because a patent exists doesn't mean it's kind of deployed, doesn't mean it's active in search, doesn't mean it's integrated in that particular way. And again, you have the danger of making these kind of faulty decisions if you take them literally. So I always use them as a, this is another piece of evidence. So when someone maybe says to me, oh, I think search engines are using this thing as a ranking factor. I'm always like, what primary evidence do you have of that? Has a Googler spoken about this? You know, has— is there documentation on it? Is there a patent about it? Because all these things point towards that. And it was super helpful when we found our Google exploit, trying to kind of trace back on things that we saw mentioned in different bits of kind of variable names to try and track paintings down. So that was how I discovered the site quality painting, and that got me on a whole different rabbit hole. I guess the one I've seen recently that has got me thinking, and I love the point you made, by the way, about search engines trying to get more, or these companies trying to get more into the human side of things, because again, this is what I was talking about earlier, which is that people overlook the fact that they are human and you intrinsically understand a lot of the things that machines are desperately trying to understand and very smart people are trying to make them understand. So I feel some people sometimes dumb down their strategies for machines and are un-future-proofing themselves because they're not aiming for where the machines are going. But I keep digressing here. To answer your question, I think the most interesting patent that I've seen recently is this one. I can't remember who covered it first. I think it might've been Joshua from Amsiv, which is the patent around Google using AI to create landing pages. Because I think this is highly likely to happen because we've seen the similar things happening in their ads platform over the years, which is You know, don't worry, we'll write the ad title for you. Don't worry, we'll write the ad for you. Don't worry, we'll make the creative for you. Don't worry, we'll decide what keywords to bid on for you. You know, it's more and more of the decision-making and power is going into this big machine learning model for better or definitely for Google for the better, arguable for advertisers. But in terms of what I said, alignment, to Google's own ambition and model that sits very firmly, I think, in that lane, i.e., we will make more money if we determine what is said on the landing page because we have all the data, which I don't necessarily like. But when it comes to placing future bets on what's going to happen, I think that to me feels likely because historically, A lot of the changes Google have made have been about keeping people in their ecosystem. Featured snippets, PAAs, AI overviews being the obvious one. Even the stuff we had 10, 15 years ago, which the first things which were, if you Googled the weather, you didn't have to go to a weather site anymore. It just showed you the nice little panel of where you were and the next 7-day forecast. So Yeah, that, I think that's really cool that people are as a community looking out for these patents. I think if anyone wants a vibe-coded AI project, I think that would be a super good idea for yourself to have a little agent that's constantly combing through and downloading patents and looking for stuff specific to search for SEO and then summarizing and slacking you about them because there's hundreds and hundreds of these obviously, and they're pretty heavy reading. So that's a great job for an AI assistant to go and maybe help you with your fortune telling.

Yeah, I love that. Yeah, that's the thing. It's not like you say necessarily actually going to be acted upon or in the way that they've put the patent down. It's more about what I find interesting is like a big part of our job is mainly like reverse engineering things. Like we're trying to work out how Google makes their decisions and trying to please them and please other engines that we're trying to show up in and stuff. And I just find it fascinating trying to get in their head, you know, because you are kind of getting into their mindset, you know, whether that's Apple or Microsoft or Google. Like they obviously thought of this and thought it was a good idea that they need to have it secured. So for some reason. So yeah, it gets in their head a bit. In terms of agencies, obviously you run one. It's gonna be very tricky, I guess, in the agency world. It always has been, but going forward now with this, the whole new AI world that we're seeing, and you know, especially if SEO's like your main kind of product, what makes a good agency these days and how do you aim to kind of stay at the top or like stay in the game as long as you can, I guess, going forward?

Well, that feels like a loaded question. Um, so I think what makes a good agency is the same thing that's always made it a good agency, which is looking after your team and having a happy team. Because an agency is just a collection of people that work together. It's very rare you have a massive advantage with you know, specific bespoke tooling or data. You know, every agency's got their own way of doing things. Some build their own models, some have their own SOPs, but it primarily comes down to having quality people. So that's— if ever I'm talking to other people in other agencies, or if I want to figure out how another agency is doing, I always ask about their staff turnover. Because to me, it's the biggest indicator on how things are going if people are leaving. And that's something I'm really proud of at CANDOR. So we just, we don't enter awards, but we got a certification, really, the Great Places to Work, which is like an independent survey of all the team and asking them a whole bunch of, I think it's like 60 questions about how satisfied they are with like working conditions, equality, pay, all that kind of stuff. And as an agency, we've got really low turnover, which means we've got I'd say the mean average experience here for SEO is probably approaching a decade. So if you picked a random person in the team, that's kind of the experience level you're getting. And they can work with clients over multiple years because that level, I think, of understanding is more important than ever because it's not just like SEO is not just now, okay, we'll optimize some keywords and we'll chuck a load of links at it. There's a lot more involved strategic processes that include understanding the brand and the customer and the messaging to plan for that long term. I have noticed, because I deal with a lot of the kind of incoming leads that we get, a lot of them are asking about AI and geo. And I've been a little bit stubborn, kind of cut nose off to spite face in terms of When I've had people be like, I want— we know what we're doing with SEO, we want a geo strategy. I've kind of been like, in my opinion, that's not particularly a thing. If anything, it's a subset of SEO. There are, you know, it's another layer, but they're primarily using search indexes to get this information. And, you know, the strategy is slightly different. So rather than just trying to rank your site, it might be identifying fan-out queries and appearing across those sets of sites. So there are differences. But to me, that's no different from people that came to us because they want more exposure in, you know, Google Local boxes, or in feature snippets, or in PAAs, or in shopping results. It's just another way that you tune what you're doing. Or Discover is another great example. We have specific Discover strategies for clients. You know, there's technical requirements for Discover, certainly kind of content ways you can tune stuff for Discover. So, you know, I think you need to embrace AI in terms of, certainly in terms of tooling. There's a whole bunch of stuff we use AI for now internally to help with processes, but I'm not going to pretend it's something it's not. You know, our agency name is literally Canva. And the premise has been, you know, if we are always very honest and open about the state of things, what's possible, what's not, what's realistic, that will help us in the long term. And that's certainly for us how it's played out. We've built that trust over many years now with lots and lots of clients. And as a result, we still get 90-odd percent of our business from recommendation and people talking about us.

Yeah. Yeah, it's interesting. I only ask because, you know, some have predicted that some agencies just aren't gonna make it in the next few years if they don't perhaps change how they're doing things and update their processes and stuff like that. But it makes a lot of sense that, you know, if you're actually good with people first off, that's the huge, like, main thing. And I just was thinking about if you're worried about obviously with AI now with Companies can spin out a website quite quickly, potentially be optimized already, and maybe what is going to be the future of websites. I know we're not quite there yet, but people obviously don't have the knowledge of search like perhaps you do, see the website traffic drop from organic and start thinking, what do we need to invest in that anymore? And then, yeah, so I guess there's that kind of thing.

Yeah, so I saw a comment on LinkedIn yesterday. And someone was saying, it was about the growth of zero-click or LLMs or something. And they were like, oh, and SEO agencies are saying SEO isn't dead, you know, this, you know, look at this. And my point to them was the same amount of money, if actually not more, is still flowing through the web. So someone wants their running shoes, whether they Google it first or use ChatGPT first, That money's still being spent.

Okay.

Now, yes, someone's moving the cheese, but if anything, that means you need help more than ever to make sure that you are there. 'Cause it's not like the money's gone. They're not now shopping on the high street. If anything, it's becoming more centralized. It's becoming more machine-led. So in my view, there's more space than ever for optimization. I think the keeping up, you know, kind of thing of agencies has just always been true. We tend to go through, you know, I've been in agencies now for 18, 19 years. We do tend to go through like many industries, these waves and cycles of, yeah, there'll be, you know, fear or there'll be new technology and agencies that were already flying a bit close to the sun don't make it. Or ones that have got very rigid processes and don't want to change won't make it. But that's always going to be the case. You know, to me, the point of an agency existing is you're meant to be the specialist in the know at the forefront, and you're meant to have a team that knows each other and can work well together. And you have a process for these things. So people can give you agency to strategize and execute on their behalf. So to me, this is table stakes, you know, when it comes to keeping up with what's effective. And again, I think there's a— I'll probably have to put this as a caveat in my talk that I do about this subject, in that I think there's an assumption that because I'm like a longtime SEO person, that I'm defensive about SEO changing because I have vested interest in it. The fact is actually, One, I don't care what happens because I will keep up with it. That's what I have done for 20 years. I don't care. 2, if anything, if things do change and technology shifts quite quickly and rapidly, I kind of like that because I'm really good at keeping up and it will make it harder for my competitors to keep up. So bring it on. But my honest opinion is a lot of the foundational stuff hasn't changed. It's more around user behavior attribution and how we tie it all together and the journey's got a lot more complex. So yeah, I think things are changing, but I honestly, I'm not massively worried about it.

Yeah, I guess like you said it and it's like in-house, it's also the way you're messaging things to clients or stakeholders, that's gonna be key. And if you're good at doing that, If you've always been good at that, then you shouldn't have any problems. And like you say, it's not like, I mean, that people do have less money in their pocket these days for other reasons, but there's still money being spent. And there's also, it's not like search is going away, it's just being searched in different ways. And in fact, we've seen like there's probably more search than ever in different places, or many different ways, you know, methods. So.

Your point on websites actually was quite interesting. So about people like vibe coding, you know, small sites, pages. I mean, again, that's something I think that's been around for a long while with platforms. For instance, like Wix has been an easy way. Like I've got a friend about 4 or 5 years ago who's a roofer who was like, oh, you do websites, can you build me a website? So of course I was like, no, leave me alone. I'm not getting involved in this. go and use this thing called Wix and they actually made themselves, and they're a completely non-technical person, like an okay-looking website. So Kanda, even before this, like 7, 8 years ago, we've always positioned ourselves to build what we call hairy, scary websites. So the things that need various business integrations where you need middleware to talk to your system that was built in 1989, You know, we've done a lot of these kinds of projects where they're more business systems tied to websites or, you know, things that at the moment, at least, yeah, would be an absolute nightmare to try and vibe code because we saw the bottom end of the market getting eaten up by the Squarespaces, the Wixes, the kind of WordPressy, there's, you know, there's a lot of those no-code type platforms, and they're great, you know, brilliant, make the web more accessible to everyone, like, you know, bring it on. There's always going to be space for specialism because that's— because knowledge asymmetry is where you can find, you know, advantage and profit.

Yeah, is it? No, it's true. Even my site's on Wix, and as an SEO, I'm a bit ashamed of it because I don't actually spend enough time actually optimizing it. I could probably do so much more in it if I hired someone like you. I could probably be ranking really high for the stuff, but I just got— that's like just a thing I stick things on, but I'm with you on that.

I mean, I feel the same about Also Asked and I've Got a Blog. They're both, you know, I'm embarrassed when I talk to people that work in like SaaS SEO and they're telling me all the things they do and they're like, do you do that? I'm like, no, no, no.

Yeah, there's not enough hours in the day. In terms of migrations, because I know you've said some stuff about it recently and obviously it happens a lot and there's still so many mistakes that can happen. What are the Kind of common mistakes you're seeing and how would you go about doing it properly, like around migrations? Yeah.

So migrations, if it's a, generally for the larger, higher traffic sites, we actually do a risk assessment process first, which is working out essentially how much money we could potentially lose doing a migration badly. So we look at things like. What's the traffic spread across the site? So if 90% of your traffic is landing on the homepage, tends to be a fairly simple migration. If you've got 50,000 pages, and 90% of traffic is spread over 5,000 of them, it's potentially more difficult. Things like branded and unbranded search as well. So again, assuming you've got a fairly unique brand, like most people, if 90% of your traffic is branded, It's pretty hard to lose branded traffic, like you have to really cock things up to do that. Whereas again, if you've got sites that are like, you know, 70% unbranded traffic, you're constantly competing for that. So any disruption of that could result in traffic loss. So we use this process to essentially work out what is the dollar value of the risk of something going wrong. And then that's applied to how much time we'll actually invest in the migration. Because there's no point spending $10,000 on a migration for a site that's transacting $200 a month, because you could blow the site up and you wouldn't get that value in years, right? So we have a whole process around identifying where the risk is, and then a plan for kind of managing that. In terms of things that go wrong, I guess from a strategic point of view, I think people sometimes use migrations as a time to also change lots of things, you know, and I've certainly had clients where we recently did a migration for a company called TravelPerk who moved to perk.com. And at the same time, they had purchased 2 other separate branded companies and wanted to rebrand those companies under perk.com and roll their different services into the same website and also improve the Travel Perks site and change the structure all at once and change platform, which was like, okay, this is the business imperative. This is, you're basically telling me this is what you're going to do. So we have to have a plan to mitigate some of the risk because technically a migration should be a one-to-one. You are maybe moving domain. That's a type of migration, or you're staying on the same domain and you're changing CMS. That's a migration. You can have a design migration. You can have a server migration. The biggest, I think, avoidable mistake sometimes I see is when people do like 4 different types of migration at once. And then if something does go not to plan, it's very hard to then unpick what has caused that. In most cases, not in all, there's some way to stagger these approaches. So again, I've done migrations where we've done like 20 country websites, and we've started with some of the smaller ones when we've been making bigger changes to see if everything went to plan. And then we slowly do the other countries when we're sure that we're getting the results that we expected rather than just chucking it all over YOLO and see what happens. So I think there's that strategic element. On the tactical side, on the actual nuts and bolts, I think there's things that go amiss. Like if you're doing a crawl of a website, you might miss a whole bunch of like non-canonical URLs that have got important links from other sites. So common examples might be where you've got really great links. Say you did a, say you've got a mailing list that goes out to 50,000 people. And you've been running that for a few years and your kind of fans or other sites have been linking to your resources and they've got the links from that email. It may be that they have your kind of UTM tracking code stuff on. And I've seen sites that have had redirects set up where, because it's got a UTM code on the old URL, it doesn't get migrated. So you're essentially losing a load of links, which we know are still super important to ranking. And then they see a drop after the migration because they've lost 500 decent links and everyone's scratching their head looking at the redirects or something, or, you know, not where the problem is. So things like that, I think, are an issue. And as well, painfully common is moving from fairly ironically fast old school websites to very JavaScript heavy frameworks. And that tends to be more, unless it's really bad, tends to be a more subtle impact. Literally, I've seen sites go to a worse user experience. So auditing the new site kind of on staging and comparing the things that you know will affect search and users, those things go hand in hand and making sure you're in at least the same position, ideally better. Before you go live. And the last one, which is a big one, is the looking at the internal link structure. So if you've got pages that are really important on your current site that rank really well, a lot of the time when a site's redesigned, the internal link structure changes. I like to do a crawl to essentially check that the important pages that are ranking are internally linked to with the same prominence as they were previously. So if you have a page that ranks really well and it's in your main menu, it's probably linked then to from every single page on your site. If something gets redesigned and it ends up in a submenu, like a click down, you could potentially lose thousands of internal links from pages themselves, which have links from the web. And again, that has a knock-on effect of suddenly a page that was driving significant revenue isn't ranking anymore. So that it's normally the small things that trip you up on migration. So yeah, keeping it simple, modular, phasing it, and trying to keep as much the same as possible. And then if you have improvements, again, trying to do them post-launch once you're happy everything's kind of gone as you would like.

Yeah, it all makes sense. I mean, the only thing I would add was if like there's company owners or whatever that watch this or devs, but making sure you get the right people in the room at the start, not just like, oh, SEO at the end and fix all the crap that's gone on because we've not made it.

Just put a video on the end. Yeah.

You know, we just messed everything up. Can you fix it? And what are the redirects that need to be in place and stuff? It's much easier, like you say, to do that risk assessment and the plans and the strategy At the start with all the right people in the room. And even like you say, like—

I wanna give a shout if I can quickly while we're on the subject of Sitebulb, because we've got a couple of webinars planned with them where we're going to talk about the ideal way a site migration should do, but then we're gonna go downstream to you've just been parachuted in a week before launch and everything's on fire. What do you prioritize? And I think it's going to be a really fun kind of series we're going to do with Jojo and everyone from Sitebulb. And if you haven't seen, your listeners, they've got a whole great kind of set of different courses on their site. Their resources section is really, really good.

Yeah, we've had Patrick on from Sitebulb and, uh, yeah, they do some great stuff. You know, we do our weekly news episodes and they've been featured a few times on those. So yeah, definitely have a look out for those and we'll probably feature them on the show as well at some point. Uh, sounds really interesting and sounds like something we've all been there and, uh, would like some advice on. Um, but yeah, we're finished off now. I've kept you longer than I promised and, uh, I really appreciate your time as usual.

No worries. Thank you for having me again. It's been a pleasure. I had a lot of fun.