Are you using AI the WRONG way in SEO?

Are you using AI the WRONG way in SEO? In this webinar, we’ll break down the biggest AI mistakes we see in SEO today, talk about accidentally producing content that doesn’t perform, and share practical strategies for you to use AI to create smarter workflows with better search results.

That’s what we’re discussing on October’s Majestic SEO podcast – joining our host David Bain are Giulia Panozzo, Rejoice Ojiaku, Eli Schwartz and Sante Achille.

Watch On-Demand

Listen to the Podcast

Transcript

David Bain  

Are you using AI the wrong way in SEO? If so, what precisely are you doing wrong, and how should you be using AI correctly? 

Hello, and welcome to the October 2026 edition of the Majestic SEO Panel, asking whether or not you’re using AI the wrong way in SEO. 

I’m your host David Bain, and joining me today are four wonderful guests. So let’s meet them, starting off with Rejoice.

Rejoice Ojiaku  

Hi, I’m Rejoice Ojiaku. I’m a senior content business specialist at Wise at the moment. So I’m really, really excited to be on this panel with everyone else. So thanks for having me.

David Bain  

Thanks so much for joining us, Rejoice. Also with us today is Giulia.

Giulia Panozzo  

Hi, my name is Giulia Panozzo. I’m the founder of Neuroscientive, a neuromarketing consultancy based in London, and I’m super excited to have this conversation today because it’s a topic I care deeply about.

David Bain  

Wonderful. Thanks so much for joining us, Giulia. And also with us is Eli.

Eli Schwartz  

Hey everyone, great to be here. Thank you for having me, David. And thank you all at Majestic. My name is Eli Schwartz. I’m the author of Product Led SEO, and I have been in SEO for about two decades. And for the last bunch of years, I’ve been working with companies and helping them cover and discover their SEO strategies.

David Bain  

Superb! Thanks for joining us, Eli. And last but not least is Sante. Thanks for joining us, Sante.

Sante Achille  

Good afternoon, everybody. Hi, everybody, I’m Sante Achille. It’s great being here. It’s a great pleasure to attend this panel and to have this open discussion with everybody. I’m a search marketing specialist. My specialisation is SEO, and I’ve been doing it for more than two decades, actually. And I’m looking forward to having a great conversation with you all here today.

David Bain  

Let’s not waste any time, and let’s go to Rejoice for the first question. We’ll try to ask all of our guests the same first question: What do SEOs typically get wrong when it comes to their use of AI?

Rejoice Ojiaku  

I think, currently, a lot of the biggest things a lot of SEOs are sort of getting wrong fall into two categories. 

The first mistake is sort of treating AI as a replacement for thinking rather than a tool that supports the thinking. So a lot of people are using AI to jump straight to the answer. So we’ll see them ask about keywords, content ideas, search intent, even competitor gaps, and I think that then they take that output at face value. But AI is really good at giving you something that sounds confident. 

Another mistake that they make is focusing too much on output and scale. I think that’s because AI makes it a lot easier to create a bunch of articles at a time, and suddenly the question becomes: how much should we produce using AI, rather than what should we actually produce? 

So, for me right now, AI in SEO is about helping you analyse faster and spot patterns, but we still need that human judgment, real search data, audience insights, etc. I think we shouldn’t get too focused on outsourcing the thinking to AI; instead, we should use AI to better support our thinking.

David Bain  

Wonderful starting point, Giulia. What do SEOs typically get wrong when it comes to their use of AI?

Giulia Panozzo  

I completely agree with Rejoice. I think where I currently struggle is conveying how important it is not to outsource our critical judgment to AI, even though, to a certain extent, AI has democratised knowledge. 

So now we can do a lot of things that were out of our depth a few years ago, but to Rejoice’s points as well, AI is very confident, and sometimes it’s very confidently wrong. So you need your own critical judgment to challenge the output, because otherwise it’s just like the same thing going over again. And a lot of that output sometimes can inform not necessarily better strategies, but rather actually very factually wrong tactics, wrong ways to reach your customer. 

And another way I think that SEOs are using AI wrong is treating it as an output rather than a means to an outcome. So, for example, I see some teams attaching KPIs to AI usage, which I find mad. And I understand the need behind the efficiency. We want to scale things, we want to make things faster, we want to be the first at doing something, but sometimes that’s not the right way. 

I feel like again, when you take away the human connection from not only your own use of AI and like your team’s knowledge, it’s dangerous for the way that you also might reach your customers, because they can tell whenever something is AI-driven, and I think it creates a massive divide between you and your customers.

David Bain  

An amazing couple of points there that I’d love to follow up on, but before we do that, let’s get Eli’s opinion on the initial question.

Eli Schwartz  

So I think that the biggest thing that everyone’s doing wrong when they’re approaching AI is they’re looking to find and replace the words that they used for SEO, and then replacing them with AI or AEO. So they’re building citations instead of links. They’re building prompts instead of keywords. They’re measuring visibility instead of rankings, and I think it’s a fundamental paradigm shift that’s very interesting. 

Whenever I talk to actual SEOs about how they’re approaching this, and then I ask them also how they’re using AI, they’re using AI as consumers, as we all do. But somehow, when it comes to thinking like a marketer, thinking like SEO, they think that it needs to fit into this other bucket. So, do we actually click links and citations when they’re mentioned in LLMs or not? It comes down to what your category is, who you are and what you’re clicking. 

I think the biggest thing that people are doing wrong, and it’s the thing they always do wrong with SEO, is they keep their marketer hat on, and their consumer hat is not tight enough on their head. So I think, when it comes to any marketing, really, you want to think like a consumer and ask: How would you sell to yourself, or how would you sell to your family, or how is someone you know going to interact with this new technology? And it is a fundamental shift from the way SEO worked and should have worked for the last 10 years.

David Bain  

Amazing! Thanks so much, Eli. Sante, what are your initial thoughts?

Sante Achille  

I think there is an excessive proliferation of tools and tooling across the entire SEO process, enabled by all the AI you’re using. And you can see this on LinkedIn, because when I go in and look, there is an amazing number of how-tos around things that are all AI-enabled. Sometimes with scripts, sometimes with prompting, etc. So, I think this is inflating and bringing to a new level of confusion the whole, let’s say, SEO process, which really hasn’t, and at the basics, at the fundamental, hasn’t really changed all that much, I think. 

For example, there’s a lot of fuss, I think useless fuss, around researching keywords or fan out queries, as they are called, and I think that’s a lot of wasted effort spinning your wheels really for no return because we all know that as as people move to natural language it’s basically limitless the ways that people are going to be expressing themselves so it’s really like trying to predict the lottery and so you just need to stick to your the basics, the fundamentals of things to actually move forward. 

Try to understand at a deeper level what you’re actually doing, because now you have an enabling technology that can allow you to go deeper into the things that you really need to know, rather than trying to speculate on things that you may have access to, but you don’t really have enough know-how around, and we really don’t have as most of us SEOs, we don’t have the you know the in-depth knowledge to actually use that data to call an accurate judgment.

David Bain  

Some incredible opening thoughts there. We had Rejoice saying don’t use AI for final strategy. Giulia talking about being careful about using it for critical judgment, and talking about how it was concerning that some teams were using KPIs for AI usage. Eli suggesting that it’s really important to think like a consumer, not like a marketer, and Sante talking about there being limitless expressions nowadays, and stop trying to predict the lottery when it comes to targeting, maybe certain keyword phrases. 

Giulia, I think I’d like to explore your comment a bit more regarding teams using KPIs for AI usage. What do you mean by that?

Giulia Panozzo  

You might have come across teams where if you don’t use ChatGPT enough in your work, you essentially lose your access because you’re not reaching the the quota to obtain access, which is very weird, I’ve come across other teams as well where AI is strongly encouraged, and I feel like it’s dangerous not only for critical thinking, but because paradoxically we are chasing more efficiency and more content, but we create just slop that then we have to go in and you know correct or evaluate because I think everyone right now can relate, hopefully. 

But essentially, when you get a lot of stuff done with AI, which is great, but sometimes I might get some reports that are like a 10-pager of Claude Slop, and I’ve got to pick out the main message. Then maybe I use AI myself to shorten it because I don’t have much time. Now we’ve shortened the time it takes to do something, so we are expected to produce more. 

And at that point, it becomes essentially agents trading information, and it’s no longer my specialism to communicate something to another person with a different background and a different set of specialities. 

So yes, I think my frustration with that is that we tend to confuse efficiency for content production or the amount of content we produce, and sometimes it’s just like removing the human layer, which is what we’re hired for as people who have been, you know, in SEO for so long.

David Bain  

Rejoice, let’s get to your follow-up on that, to a certain degree, because you were talking about a related subject: you shouldn’t use AI to do too much. I’m not sure if you use the phrase “critical thinking,” but you were certainly talking about the strategy side of it, as opposed to the implementation of items that had been agreed upon. So, what’s an example of an absolute no-no when it comes to utilising AI for?

Rejoice Ojiaku  

So, I think based on what Giulia said when it comes to the KPIs, I think for me it would be the KPI aspect of things. Nowadays, I think AI is really useful in helping us understand KPIs, but I just wouldn’t let it. 

People are trying to let AI decide what success looks like for them and their business, and the danger is that we can give AI this goal to improve SEO performance, for example, and it will give you a list of reasonable metrics like traffic rankings, impressions, and so on. But it doesn’t necessarily understand which of these those metrics absolutely matter to the business because your business will be measured by different business goals. So as Giulia said, it still requires judgment. 

My huge no-no is that when we are looking at data, we can’t let AI determine what’s important to us. And I think a useful way to utilize AI, especially when it comes to data, especially when it comes to setting the standards for KPI, is that the KPIs that really matter has to be agreed beforehand, agreed with the team, agreed with based on the business goals and so on, and then we can use AI to kind of look at how we manage large amounts of data because an AI can at least help us spot patterns or maybe flag unusual changes, and then we can then see and try to connect those different data points.

AI can do that for us, because it would take a lot of time to do it manually. So instead of just telling me that organic traffic has dropped by 20%, AI could help us investigate where that drop may have come from. So it quickens that process for us, but we still need to be able to measure it against our business benchmarks and goals. 

So I do think that AI can be useful, but it’s such a massive no-no to let AI determine what success looks like for us in the data sets we are feeding it.

David Bain  

Thanks so much, and Eli, I’d love to dig deeper into your piece of advice and thinking like a consumer. So, what’s the difference between thinking like a marketer and thinking like a consumer? And I guess, how can AI help with that?

Eli Schwartz  

So I don’t know if the group, or anyone watching or listening, has tried any of the AI assistants. So someone invited me to Instinct, the personal assistant. Someone invited me to Instinct a couple of weeks ago, and I’m absolutely hooked. So it’s in my WhatsApp. I send messages. I’m not even going to an LLM. 

I think if you go to any SEO conference, there are all these presentations about how to do AEO or GEO or whatever it’s called for ChatGPT, but how many of us are actually using ChatGPT versus speaking into Gemini, speaking into just Google Search, typing into Google Search, or using Google Lens? 

So that’s what I mean to really think about how someone in my industry is going to be doing something like this? In my newsletter that I’m publishing tomorrow, I talk about how important SEO is for what happens in personal assistants, and I think SEO is more important than ever. So with the example I use in my newsletter, and this is something that I’ve been doing a lot of with my personal assistant instinct, where I take a picture of something and I just say, “Find me someone that will solve this problem for me”, such as I’m looking for a vendor. I want to know how to get home. I want to know how to commute. What’s the best time to leave my house?

Everything these personal assistants find comes from websites, and they find these websites through SEO. So SEO becomes more important than ever to ensure that all these personal assistants get this information. That’s what I mean by thinking like a consumer. 

If you’re a plumber, that’s the example I use in my newsletter, or if you have a SaaS software, or if you are selling consulting, which a few of us here do. How is it that people are going to find you, and what is the medium they’re going to use? Like I know for myself, my SEO and my own personal consulting website is absolutely atrocious. But the way people are finding me is through word of mouth and through LinkedIn. So if I want to promote my own consulting, it’s not about how do I jam keywords into my website or how do I get citations for AEO? It’s really how do I double down on the things that are already driving things to me. The other thing that I might do around SEO is get people to read my book. Now I’m not going to get people to read my book just because they Google for product-led SEO. They need to hear that from somewhere. 

These are all the things I mean when I think about consumer best practices, and I’ve been pitched by many people on how to do better SEO for my book. Yes, there are best practices to do, but in reality, and David, you’ve published a number of books, so you know this more than anyone, and I’m honoured to participate in those books. The reality is the way people discover books is not because they Google for like, hey, I’d really love to read a book about you know SEO in 2027, let me find what David Bain has written with Majestic. They are hearing it from somewhere else. So when you think like a marketer or think like a user, how is it that people discover these books? How do people discover these things? And again, think like how you would do it, and then you can think about those inputs.

David Bain  

Is there as much data available when it comes to thinking like a consumer, because you could imagine that thinking like a marketer is data-led, and to a certain degree, thinking like a consumer is a little bit more feeling and conversation-oriented?

Eli Schwartz  

You have to find the right data. So we’re here with Majestic. So I want to say you should be using Majestic. So there are things you can do to get ideas about what people are searching for, but then you need to go search for them yourself and see what’s there. So I wouldn’t just build an SEO plan off a bunch of keywords. That’s data that might not support what the customers are doing. You want to have those feelings, then use them to validate the data. And if there is no data to validate what you think is happening, it doesn’t mean it’s wrong. I would say use data to validate, not data to build your strategy.

David Bain  

Sante, Eli mentioned keywords there. I’d love to get a few more thoughts on the limitless expressions you mentioned earlier. From a data perspective, is it possible? Is it worthwhile mining limitless expressions or keywords just not what they used to be, and you just need to write from a topic perspective?

Sante Achille  

Well, I’d like to continue on this discussion around AIO or any of the flavours, let’s say, of what is now the new standard that we find. It was the position zero, and how to get there. And fundamentally, things have not really changed all that much because if you go back to just well, you stick to the principles of creating a page that addresses generic needs. 

Recently, I have been doing more work on products, particularly Shopify websites, and I really haven’t fussed at all chasing any of the trends around AIO, GEO, and whatever several 100 flavours of that you can find. In LinkedIn and elsewhere on the web, I’ve been following the basic principles of natural language processing, setting up a framework around a piece of content, and then sort of like engineering it in a way that it reflects a certain structure, and then addresses what people actually need to know or are curious to know around a certain product or a family of products. 

Getting back to your question, I really tend to take, as an order of magnitude, what’s coming from my Google Search Console, and really elaborate any kind of strategy around what I find there, which really grounds my performance in what the work that has actually been done. And that’s where some AI can come in and help look at a set of 5,000 terms, and then give a picture, as Giulia rightfully pointed out. You need that 30,000-foot view on your data. That’s where the power of AI can actually come in.

I think I have a particular view on these things because of my engineering background. So this is all modelling. Marketers and whoever is not technical doesn’t really know much about the nuts and bolts that are under beneath all of this. So starting from natural language programming and moving on, and as it evolved into neural networks, etc. that kind of let’s see perspective and having the the background of somebody who actually needs to or had needed to predict the outcome of a certain type in an engineering way, so really designing things sheds a completely different light on all of this. 

I have a very puzzled and worried look on a set on a data set because I put some inputs in on one side and then come out on the other side, and the first question I ask myself is, how reliable is all of this? At most of what you get when you’re talking about keywords is useless. Okay, you’re chasing things that are completely ungrounded many times, not because of hallucinations or these kind of things. I think the recent models of the big LLMs are rather stable, so they don’t really take you out on a ride if you ground them properly and provide a very stable, robust context for the algorithm to work with. So I think there’s a lot of groundwork that maybe SEOs and in general people working with AI don’t really provide because they’re trying to use it as a tool to get back to what Rejoice was saying as a replacement for thinking. They want a quick answer. I think this is a big problem with society in general. You know, we want everything now. It has to be fast, and and unfortunately, things don’t work like that.

David Bain  

I guess we may though sometimes experience taking over a project when someone has been looking for the fast solution, and we have to fix a lot of things, or perhaps even backtrack a lot of things. Has anyone experienced working on a project like that, where perhaps someone, perhaps less experienced in the use of AI and SEO, has started to create or implement a strategy that’s just not beneficial at all, and they’ve had to rein that in or change it entirely?

Eli Schwartz  

I’ve worked on a number of projects like that as a consultant. It’s very difficult to come into something and be able to not scream out loud about how terrible the internal employees are doing things. So you need to like come in and and earn political capital, and slowly say, hey, maybe it’s not great that you have 10,000 pieces of AI content, or maybe share a link to something that Google said about AI content. 

I think that most companies I meet look at AI as some sort of magical silver bullet that gives them exactly what they want to get. So they feed all their data, and then they get confirmation bias that their data says exactly what they wanted to say, or they use AI to fire their entire content team and produce it all with AI. 

So I think most people that I meet have completely jumped the shark on doing everything the wrong way with AI.

David Bain  

It sounds like an SEO’s use of AI to use another analogy is what some people describe as as the the effect that money has. So money is a magnifier. You know, if you’re doing things wrong, then you’re probably going to do things more wrong, but if you’re doing things correctly, then hopefully it can actually have a positive impact in your life. And it sounds like AI is like that. 

And in your example, if you have like 1,000s of pieces of AI content, do you just get rid of it? Do you attempt to try and improve it, and and and still see some value there to to some degree?

Eli Schwartz  

I think that question is what’s the value of the content? We’ve had this issue exist with programmatic SEO before AI. When I started my SEO career, and Sante, you’ve been doing this longer than me, so you probably remember the verb spun content, where that was like the earlier version of AI, where you just kind of found and replaced like different words, and then you called it new, and Google thought it was new, and then you posted your you stolen content and then you spun it and then you posted it back on websites like Ezine articles. Then you got backlinks like so that was just earlier version of manual AI and and now we have actual AI. That content didn’t serve any purpose. It was completely useless. So this new version doesn’t serve any purpose. It’s completely useless.

Now if you found something that serves a purpose, like if you created, if you are an e-commerce site and you had no product descriptions and used AI content to create product descriptions, that’s not a bad use of AI. But if you are a, let’s say, a review website and you used AI to create reviews of stuff you don’t sell and things that you don’t have any business talking about, probably a good idea to get rid of it. It doesn’t serve any purpose, and it’ll likely get a site penalty. 

That’s the thing I’ve seen more and more of. I see a lot of sites actually getting penalized. I think Glenn Gabe was one who coined the term Mount AI, where you see this rapid rise in traffic and then a rapid descent. So a lot of sites are having that while they’re having the rapid rise. They think it’s they’re they’re very smart. They’ve created all this content that drives a lot of traffic, and then that becomes the problem for them. So most of the time, yes, delete it. But if it is serving a purpose, and you know the user actually would benefit from some of this AI content, then a decision has to be made.

David Bain  

Obviously, the AI that SEOs tend to use may be centered around ChatGPT, Claude, Gemini. Is there anything unique regarding those individual tools, or maybe another tool that you’d like to mention that SEOs get particularly wrong?

Giulia Panozzo  

I’ve got another use case, but it might not be related to one of the tools that you’ve mentioned. I was asked to weigh in on, and I was asked what my thoughts were about scaling fake influencers with AI. To and the the main goal that the you know the webmaster wanted to obtain was to obtain more authority, and I was like, “Yeah, no, that’s the wrong way to go about it”. People come to you because, or like, people look for review because they want a shred of human, you know, perspective and authenticity in a sea of slop. 

So if you feed them with something that’s fake. Not only don’t serve their purpose, but it’s also like you actively erode the trust they can have in you, and that’s a misconception. I think that is quite popular in SEO that the only danger is the manual penalty. 

But there is the danger that we don’t even see, which is erosion of trust. Like people will start to think that we’re not legit, therefore they won’t even look for us anymore, and that’s something that is hard to prove because there’s not much data that we can see other than like loss mainly in searches, impression, and so on. But it’s definitely a danger that’s there. So that’s my like the the closest I’ve been to witnessing something that could have gone wrong for AI.

David Bain  

I’m just going to ask a few questions rather than directing them. So please just jump in whenever you you feel that you’d like to add something. Another thought is in relation to automated publishing. Is it ever a good idea to automatically publish either content or code from AI that that hasn’t been reviewed by a human being?

Rejoice Ojiaku  

I think that’s a huge risk to do because I think with automation publishing, like an agent to sort of give access to an agent that can change internal links, can change canonicals, metadata, etc. It might be one mistake that might seem small, but if the agent is making hundreds of changes because you’re doing automation publishing or is just pushing out content, so that just tells me that there is no review within that workflow. There is no review to actually see if it’s being if it’s true to what you’re trying to sort of put out there.

In terms of automating publishing content, there has to be a brief. So if the agent is automatically just sending things through, I question what brief have you given it? I think automation for repetitive, low-risk work is fine, but for higher potential impact of change, I think the stronger the approval needs to be. So I would happily let an AI identify opportunities. I think that’s fine. I think AI can you know recommend those changes or even prepare those changes, but for any significant live site changes, I would always say a human needs to say yes that this makes sense, and then it can sort of go off because what it’s very hard to start recalling content back or having to go in and you know quickly make that change. So automation should remove unnecessary work. It shouldn’t sort of remove the accountability we have when we are pushing out that that sort of content.

David Bain  

Rejoice, what’s an example of a low risk activity where you wouldn’t mind automating?

Rejoice Ojiaku  

So I wouldn’t mind, for example, like if I had a particular page that needed different metadata, meta title, meta description. I wouldn’t mind necessarily feeding the agent a the guidelines around it, and maybe feeding it the keywords I’m trying to sort of generate that for, and for it to kind of automate and give me like a thread of descriptions that I can read through and then kind of review because the actual work of writing, the actual work of you know thinking about different ways to sort of wordy, especially a lot of SEOs we’re not necessarily copywriters at heart, so we might you know need a bit more boost. I’ll automate things like that. I’ll automate minor changes that is very sort of low risk. Whether it is a case of looking at you know spotting different canonical tags that are not self-referencing those tiny things, but anything that I think needs to be published live on the site, which shouldn’t be automated for the agent to push it through without it sort of having that final approval from a human. 

I think automate the process, automate the actual lead up to it, but never actually automate work that needs to sort of go live, and because then it just shows a bit more. You know that we’re callous and we’re not thinking about the user in mind. If we’re just constant, you know, trying to publish so many articles, and the agent is sort of doing it on, you know, on this wheel, we’re not thinking about the consumer at all. We’re just trying to push content for pushing content’s sake,

David Bain  

Has anyone seen an increase in grayer hat SEO powered by AI? So the creation of fake digital footprints, reviews, offsite citations-something intended to trick LLMs.

Giulia Panozzo  

I think you can tell that now everyone tries to leverage that because no one wants to be left behind, even though like we know that necessary is not the best way to to act. And to be honest, like if so, for example, I know like Eli, my site is atrocious for myself, being an SEO and all that. Like I, you know, like I’ve always been wanting to change it and to refresh it, and yesterday I was trying to get Claude to redo my site, and I couldn’t like I couldn’t connect with it because it just looks like everyone else’s, and I know that we’re trying to you know make sure that we we leverage AI to you know defeat the information about us, but sometimes it just flattens the experience and it makes us just more equal to everyone else. And we spent as marketers years trying to stand out in a sea of other choices, and now we just settle for mediocrity because it’s more efficient. Because we see everyone else as well, like producing more, and you know maybe like their site is more sleek. 

At the end of the day, though, I think yes, we’ve seen a rise in that. But I also think that we are slowly seeing not a rejection, but. People are being more careful now because they know that you can tell when a site is done with Claude. You can tell when something goes wrong because not necessarily all of the SEO guidelines are put in place. Sometimes the site breaks. Where does it come from? People don’t know. So yeah, but I’m interested in knowing what other the other panelists have seen.

Eli Schwartz  

I think that’s interesting point, Giulia. I actually think that there’s no problem with having Claude sites because what AI does is it raises the baseline, so everyone has to have something. So the same way LinkedIn raised the baseline that you don’t need like the most well-designed resume, let’s call it, or most well-designed personal brand online. Everyone get LinkedIn, and then you just put in whatever you want, and it may be a little content. It’s baseline. So I think having a Claude site, and you and I are probably guilty of this if we have bad sites. That would be the baseline. If you are a business or a consultant or whatever you are, you need to have a web presence under your name, and you need to have a site. And Claude would be the baseline. You can overdesign it or make it better designed, and that’s good. 

And I think the same applies to content. If you have a service website, you have to have reasonable content. There are so many examples of terrible websites with terrible content, and I think the future is that you can’t have terrible content. It’s fine for it to be AI. Again, we’re not talking SEO. We’re talking just to communicate with your readers and your users about what you do. It’s fine to be AI. You can go above and beyond by adding a voice to it. And I think Claude or ChatGPT or Gemini is going to give that baseline. 

From a content standpoint, I mean, I use Substack for my newsletter. If I wanted to make my life hellish, I could put my newsletter on Mailchimp, create my own HTML template and do all that. But Substack again provides a template that makes it very easy to manage everything. All I have to do is write and send. There isn’t much customisation I can do when it comes to images and styling, but it’s the same. So I think customisation is going above and beyond. But that’s probably the best thing that will come out of AI. You can do many of these basic things very easily with minimal effort.

Giulia Panozzo  

I’m interested in knowing now, and if you’ve seen or and that goes for the entire panel. But, for example, whenever I write content, it doesn’t necessarily come out the way a customer would, you know, like, pick the main headlines and appreciate it. So I’ve, you know, I’ve heard from other people. Well, just use AI to like, you know, make it more consistent. Sometimes I do, and then the output just doesn’t sound right in my voice. But it surely gets better, right? But I’m interested in knowing if anybody has done tests where they have put AI content and fully human content to the test, so AI-assisted versus human, and seeing the results not only on citations, prevention, and so on, but in actual traffic.

Eli Schwartz  

We have a different example. Maybe someone else has experienced this, but a client was delaying paying me, and I sent the appropriate emails, and they ghosted me. So what I did was I wrote a really nasty email in Gmail about what I thought about their delaying paying me, and how long it was taking them and why they were ghosting me, and then I told Gemini in Gmail to fix it, be appropriate and passive aggressive, and then end with like, and I told you so. Like this is what’s going to happen if you don’t pay me. And they paid the next day, right? So, like my wordy email would have like ended up in a Twitter thread or something like that. But here, like it, it accomplished what I wanted. I obviously edited what Gmail output. So I think that’s the thing, you can take this content. It is better. It’s probably much better than anything we could come up with on our own, unless we’re Shakespeare and amazing writers. So you take your terrible content, and you put it into AI, and it reframes your thoughts or reorders things. You can agree or disagree, but you’re no longer starting with blank comments or just emotion.

David Bain  

Yes, great point there. I’d love to just ask one quick follow-up point regarding what Giulia and Eli were saying there, and you were talking about Claude sites. Giulia, you were saying they’re easily identifiable. What aspects about these sites are easily identifiable, and are you talking about any type of sites? For instance, you can use Claude to create a WordPress theme or something else. What aspect about it makes it so easily identifiable, and what is negative about this?

Giulia Panozzo  

Yeah, I mean to be honest, there’s nothing wrong. Again, I’m not demonising Claude because most likely what I’ll need to do before my next conference, you know, like I want to have a nice site, which is not the one that I put on in 2023, so I might do it myself, but I think what makes me identify a Claude side is just that the fonts look the same. Even when I try to prompt it differently, it just looks like everything else I’ve seen normally. And there’s always been like the tiny dots to start the sessions. And again, you can prompt it differently. Right, so it’s not a matter of like, yeah, don’t use Claude because you can’t have your personality shine. Fine, it depends on prompting. 

I think what sounds wrong to me, for my own purposes, is that I don’t want to be like that again and I want my personality, my work to come out, and maybe that’s the way to have it done nicely and fast for sure. Because we don’t have like a lot of the time we spend on clients as consultants, so we don’t really have the time to go and you know fix all of the tiny things in our SEO. But yeah, I feel for me is more matter of perception. So, if I look and if I sound like everyone else, why should anyone decide to go with me instead of some other consultant? 

So that’s my challenge. Again, that’s something that I’ve noticed in how I approach websites. But I think to your point, if it’s something that’s only visual, and then there’s the experience and the background and the content to back up on the experience and the value that you can provide, then I think you know it might be a tiny friction for somebody that feels strongly about AI, but in the end, it can still serve the purpose.

David Bain  

Let’s finish off by asking all our panelists a closing question and get them to confirm who they are and where you can find them. So we’ll start off with rejoice. Rejoice. What’s one positive thing that SEOs need to start using AI for right now?

Rejoice Ojiaku  

I think it kind of varies. I would say for me, the main one positive thing would be to use it for pattern spotting across large amounts of data. I think that would be very, very helpful. So, for example, you know, you could give AI a lot of like your search console exports, keyword data, content performance, or core data, and just ask it to help identify patterns that would that would most likely take a longer time to do or spot manually. I think you know it can actually help answer things like which pages, which page types are losing visibility or which topics are growing.

So I think that is where AI is really useful because as SEOs we do deal with a lot of data, and sometimes it can be quite difficult and you know time-consuming to spot those patterns. So using AI to do that would be a really good use of AI to find the signal faster and not to make this decision for you. So that will be my one positive take. 

You can find me on LinkedIn, Rejoice Ojaku. Always happy to connect with other marketers and sort of speak about AI and any you know new ideas and new concepts that you know are sort of going around in our industry. So that’s where you can find me.

David Bain  

Superb, thank you, Rejoice and Giulia. What’s one positive thing that SEOs need to start using AI for right now?

Giulia Panozzo  

I think automating the tasks that we do that take a lot of time otherwise on the regular. So, for example, I use Claude to replicate a Netlify site that does causal impact analysis, which is something that I do very well. But it also takes a lot of time. So, anytime there is a project shipping, then it feeds into the causal impact analysis, and it has an archive of where the most impact was generated. So I think that’s a great use of AI. Again, my challenge with that is that I had to prompt it correctly, and I had to keep spotting errors. But that’s because I was outsourcing my knowledge, and you know, like maybe there were spots where Claude was getting wrong, but if you do master a process and it takes a long time to do on the regular, just use AI to automate that.

You can find me on LinkedIn and then at neuroscientive.com, which by tomorrow might look different thanks to Claude. Who knows?

David Bain  

Thanks so much, Eli. What’s one positive thing that SEOs need to start using for AI?

Eli Schwartz  

I think that AI is very helpful with pattern matching. So upload a bunch of your data and find a pattern. So find out if there’s a city like that most of your users are coming from, or more of your users are coming from. If there’s a keyword, a landing page. So I know all of us have done this, where we sit and stare at Google Search Console or stare at Analytics and try to discern a pattern, but quickly drop all of that in into any AI platform and find patterns, and then you’ll go down different rabbit holes and find other patterns and upload more data. So that would speak to what I mentioned earlier about understanding users. So find those patterns, you’ll have more dimensions on who your users are.

You can find me on LinkedIn, of course, and subscribe to my newsletter at productledseo.com.

David Bain  

Lovely, thanks, Eli, and also with us today, Sante. What’s one positive thing that SEOs need to start using AI for right now?

Sante Achille  

So my call is a little bit more technical, based on my experience that I’ve been having, and I have developed. Over the past couple of years, at least, I find that it can be a very powerful tool to actually increase productivity and efficiency of the entire, let’s say, SEO process by creating bespoke procedures that clients can actually use internally, so that they do the heavy lifting internally, and I, as a consultant from the outside, can look at the more high-level strategic stuff. 

And this sort of like blends into a trend that I was able to talk about, and people agreed with me in a meeting I had just a few days ago in Rome with IT companies that develop software, and there was a consensus with the the people that I actually spoke to who agreed on the fact that there is a market that is probably emerging where middlemen who know the company, who know the processes, who know the behaviors, who know the people within an organization, can actually create what I call procedures, and these procedures then can be industrialized once they have been consolidated by iterating with the company, and then they can be handed to companies who develop software to create a product out of that. 

This is now possible within SEO, and I can give you a concrete example as my last consideration. For example, Shopify. If you have a huge catalogue of 10s of 1,000s of products, with the help of AI and obviously some background knowledge in coding and the technical ins and outs of the platform, I was able to automate the full SEO of all of the products, and I did that based on the people that work internally within the organisation. So it was really bespoke to the person physically doing it, based on their level of knowledge and their ability to interact with the tool I created. Let’s call it that way. And now I have a full procedure that goes from automating the images all the way to the creation of the product description and the collection description that can be ultimately created into a product per se, that is bespoke to that particular organisation. 

So SEO can have a completely different look and feel right now, where you go from a let’s say an industrialised approach to the products that we now have on the market to a very micromanaged level, where people like us who do consulting can deploy very, very specific kinds of procedures that are a perfect match for each organisation.

For anybody interested in talking about this and anything else around AI related to SEO, I’m Sante Achille and the best way is to catch me on LinkedIn or metricdrivenmarketing.io.

David Bain  

I’ve been host, David Bain, and you’ve been listening to the Majestic SEO panel. If you want to join us live next time, sign up at majestic.com/webinars, and of course, check out SEOin2026.com too.

Previous Episodes

Follow our Twitter/X (@Majestic) and Bluesky (@majestic.com) accounts to hear about more upcoming episodes!

Or if you want to catch up with all of our previous episodes, check out the full list of our Majestic SEO Podcast episodes.

Back