Product of the week
Recorded May 20
(For the summary, please click here.)
John Eischeid: All right. Thank you for taking a few minutes to talk with me, Heather. To start, can you just um tell us briefly about the work you’ve done in the past, what you’re doing now, and where you think things are going for you?
Heather Lloyd-Martin: Awesome! Well, first thank you for for this this interview. This is really fun. So I have been in the SEO world for over 25 years. So I was in it back when there was just a few core people. We were all figuring out SEO best practices and the conferences were like super small because nobody knew what SEO was back then. And so since then, I’ve watched this whole transformation happen within the content realm. Um and I throughout the years I write content but primarily I teach people how to do it. I do a lot of consulting and a lot of training and it’s been fascinating to watch sort of the same conversations come up just in different ways. So, you know today we’re talking about how AI and LLMs and everything have transformed how we write and how we’re doing things, and years ago it was content spinning or it was something else. So there’s always something a little bit different in the world. And that’s why I love doing what I do because there’s new things to learn and new opportunities to leverage.
JE: Okay. Having said that, the latest thing to come along in this general realm, there is SEO, but people are starting to use terms like GEO which is generated engine optimization and AEO which is agent engine optimization. How do you feel like about those terms? Do you think that they are entirely new things? Do you think that to a large part those are more like sub-disciplines that are still built on SEO? How do you think that those terms fit into the overall picture? If indeed you think that they do, because some don’t.
HLM: Yeah, exactly. And uh it could just could be that the cranky – I’ve seen this happen so many times in the world type of of opinion coming from me – but I like the terms because they do describe that things are shifting. However, when we’re getting into the whole is SEO dead thing, then that’s a whole different conversation that I don’t think is relevant today. So, what we are seeing is the same things that we do for the normal “How you write content to be found” and to engage your readers and help inform them and move them along your sales process. That has stayed fairly constant and I would say that has stayed fairly constant over time. So these new terms that are coming in are reflecting that people’s behaviors are shifting and I find that fascinating on how people are starting to integrate say ChatGPT into their lives and not just how they’re searching for information anymore. But it will be interesting to see later on how – and if – a term shakes out to be more encompassing of everything. For right now what I’m seeing more of is that people are moving away from “I must create content just for Google because this is my main marketing play and I must optimize for every key phrase under the sun even if it’s not really great content” and looking at “Okay things have shifted and the rules have changed let’s go back to traditional marketing and PR and make sure we get interviews and citations and all these other things because that will help us with with getting citations and AI.” So, it’s an interesting evolutionary time right now and it’s fun to watch things shift and change.
JE: Yes. and what you touched on with the citations and that kind of thing, that’s one of the reasons why I prefer to publish both the summary and the full text of these. In the hopes that’ll increase visibility, but more importantly, it puts a more representative amount of what you have to say out there in the public domain. So, and on a related question or kind of moving along, what are your thoughts about using LLMs to generate volumes of SEO-optimized content? It’s kind of like letting machines write for machines. Do you think that’s a good way of using LLMs?
HLM: Oh wow. This is — it’s such a loaded question because uh when you say volume that kind of changes the game for me of how I would look at it.
JE: Okay. So, the word “volume” loads the question. That’s what you’re saying?
HLM: The word loads the question because what it reminds me of and again – this is coming from all the years of seeing things change and more – is the bad old days of like content spinning, right? So you would write an article, you would have the key phrases, you’d shift out the keyphrases and have basically the same article. You’d send those everywhere to all sorts of sites. It was a mess and that’s where I see a lot of companies – especially back in the early days where I saw them looking at these opportunities to write quick fast content and slice the writer out of it and thinking, “Kaching! Look at us! Suddenly we can do all these great things in Google. We can be found everywhere and we’re just paying like 20 bucks a month for ChatGPT.” And that drives me nuts and there actually I’ve seen a big company did do that – which would have all the brand recognition – that fired all the writers back when ChatGPT first came on board thinking, “Woohoo! Savings!” – realized that all of their content was not performing like it used to and it was not the magic bullet that they thought it was going to be. And so their next step was to rehire a bunch of writers because they realized a writer needed to be part of that process. So can it help with workflow? Heck yeah. I mean, so much. Is it the be-all-end-all of, “You hit a button and push publish and it spits out a bunch of content?” I wouldn’t recommend that in any way for any company because it might create content, but unless you’ve really trained it and have a writer looking at it and knowing that it’s fitting your brand voice, you just thrown good time after a bad result.
JE: Okay. That brings up two follow-up questions. I’ll start with the first one. First, you said that it wasn’t performing well. How much of a change was there?
HLM: They didn’t disclose that in the call that the call that I was on. What they did disclose is it was big enough to where the company was surprised that they were looking to hire writers because they had shifted away from that in the past. And that was considering the move they made to get rid of the writers the fact that they were bringing them back as a goal was a big shift.
JE: Okay. One related question. How many writers did they hire back or did they not disclose that as well?
HLM: They didn’t disclose that, but considering the the size of the company, I would imagine that they had over 50 people for sure.
JE: Okay. And my second followup question to your initial question– sorry if that’s confusing, but, um . . .
HLM: I love that.
JE: Uh, or your initial answer. See, I’ve confused myself. Um, so you mentioned they needed a human in the loop. So are they kind of moving over to the model now that I see a lot of people using where they kind of use an AI to do a general first draft and then they go through and revise it themselves. Because it sounds to me like what you’ said is they basically just were using they were just copying pasting from ChatGPT onto their website and that didn’t perform well.
HLM: Exactly.
JE: Now, I mean are the writers are is it a situation which the writers are once again writing from the ground up or are they working more in tandem with ChatGPT and other LLMs?
HLM: Uh, for this particular company, I don’t know what the eventual outcome is, but what I see with other companies – um, from small companies to large – is the realities are they’re looking to save money, be more more efficient, all of those things. But they also know that pulling a human out of the loop or they suspect that pulling the writer out of the loop is not going to help them. So, they’re looking for efficiencies of how to build that into their workflow to where a writer can write prompts and have something that will give them ideas or set up – uh, train – ChatGPT to where it can write in the brand voice so the draft that comes out of it is pretty good, or upload persona information so it understands how to write for that particular audience. Now having said that, I do that myself with with my own writing and the reason why I feel that more companies are bringing writers back into the loop is if you’re not a writer ChatGPT can give you some really good advice that’s not correct and you think, “Oh this is awesome it loves this headline!” And then you realize it pulled it from somewhere else or it’s citing links that don’t exist, or it is it might mention something that’s a benefit but it’s a really soft benefit. benefit. And as a copywriter, as if you would work with an intern, you might say, “Hey, I question this. Why do you have this as one of the benefits?” And then you start to build that dialogue with ChatGPT in order to help that content get better as if you were able to do that with a person. And for many writers, they work by themselves or even if they work in-house, they’re still creating content with headphones on doing their own thing. So, they have a lot of of conversations around how ChatGPT can work as a cybernetic teammate and actually help increase creativity, because it’s allowing us to focus on what we do really well. And ChatGPT is kind of doing the stuff that drives us nuts. That has always kind of been a – uh something that has gotten in the way of how we produce. So, the writers that get it and tune into it, they really dig it because it allows them to do more uh in a way that works for their workflow, not in a way that is taking over and they feel like they have no control.
JE: Yes, I use it for, um, mostly in that kind of way. You did mention just now though that you know training ChatGPT or another LLM to write like you. There are a few different ways of doing that. I know that Gemini has Gems which is good for specific tasks. I found that Claude is very good at matching tone and style for specific documents if you tell it to. Could you just flesh out when you say training it – like, I mean I’ve also worked on the other end of training an LLM – like one of the main models and that training process I was one of probably hundreds of trainers in that. So, I’m just kind of curious as when you say like training LLM, there are a number of different ways that you can do that and there are a number of different difficulty levels. So, could you just flesh that out a little bit more?
JE: Um, actually you said, “Write like you.” So when you said persona, I was thinking more along the lines of a writer uploading kind of their own persona as a writer. And that’s probably why I wanted you to flesh that out a little bit. But, I do understand the notion. Okay. But I understand you were talking about a different aspect, more of writing towards a specific consumer and saying this is the persona we’re targeting via the guidelines for our own writing. Okay. That makes sense.
HLM: And those are relevant, because certainly yeah, you want to make sure that you can have that information in there. Uh, and what’s cool about another thing that I’ll use with Chat GBT is I’ll say I’ll have the post or the blog or whatever and say, um, this is what I wrote for this persona based on what we know. How would you rate this post? I mean, and granted, you’re talking to a machine asking it to rate it. Uh, but it does give me ideas I might not have thought about before of how to make the writing slightly more specific and better for that audience. So, there’s that. Um, but there’s also just getting it used to you and how you react and how you talk and how you write. Uh, and so for say the LinkedIn one, I uploaded I don’t know how many posts, probably at least 50 of just Heather posts and I also have asked ChatGPT “Based on these posts, what would you say about my writing? How would you analyze it? If I were to just — if you were to describe my writing style to someone else, what would it be?” And then it will spit out that information that is a little bit consultative, a little bit supportive. And it has all of those aspects that truly is my writing, and it can also set up blind spots, uh or mentioned blindspots, which I thought was very interesting. Which is, “You provide a lot of good informative information and you’re a very good educator. However, when it comes to the call to action, you tend to pull it.” Like, “Oh! You don’t do it with client information, client copy, but you do it with your own.”
JE: Interesting.
HLM: Right?
JE: Yes, I understand that. And you’ve mentioned blind spots in some of your posts before, so that is a mental note I’ve made for myself.
HLM: [Laughs.] It’s a little humbling, right?
JE: Yeah, it is. So this kind of dovetales reasonably well into another thing that you mentioned in some of your posts and there’s kind of like you call it “The Gen X Anthem of do-it-yourself.” So do you think that the adoption of LLMs and AI tools is sometimes evidence of a generational divide as is often the case with many new technologies.
HLM: Yeah, exactly. And this is – a lot of this is anecdotal. I’ve read a study on this recently, but this is also kind of what I’ve seen and what I’ve read out there. So, um for Gen X, uh we we cut our teeth on this, right? So, the founders of the people who were set up SEO best practices, we’re all Gen X. So, we’re a lot of us aren’t afraid of technology, but we might use it as just like a tool. So what I’ve read is like people that are they talked about boomers they completely forgot about Gen X in the study were used to that never mind. But for boomers, they were saying that they used it as a supplement to Google. So they might do a Google search, there might be sometimes they use like ChatGPT or Perplexity or something else, but it’s not – they haven’t fully adopted it into their workflow or into their lives because it’s just a little bit different. They’ve been using Google for the past, jeez, how many years? And so that transition is not as seamless. Now we start looking at my Gen Z friends and they’re talking to Chat GBT as if it’s a life coach and asking it like, “How do I do this? I want a divorce. How do I get a divorce? What are things I want? I want to go on vacation. Where can I go that are cheap and I can have a lot of fun? Uh, this is going on in my business life. I don’t know how to deal with it. What are some suggest suggestions?” And that is where I feel like this technology is going even more. We’ve talked in search for years about conversational search, right? And this is conversational search! Truly conversational search! Because we’re having conversations with something who’s giving us input back. I mean, granted, it’s not necessarily, unless we’re asking for it, going out and finding information for us, but we’re having conversations where with Google, it was, do a search, refine the search, look at the results, and come back. And that’s what I find is fascinating over how people are using it now.
JE: Yes.
HLM: And as more generations come on, it’s I mean, who knows what we’re going to be doing next, you know?
 JE: Yes. And do you think the speed with which you can use the speed with which you can refine your search or get specific answers and that conversational aspect – do you think those are probably at least some of the driving forces behind AI’s quick adoption?
HLM: Yeah, I do.
JE: I like it because you don’t have to go through 10 links. It just gives you the answer. If you really want to do some background research into it, you can. But does simplify things a lot.
HLM: It does. And what’s funny is that if we would have had this conversation say a year and a half ago, or, you know, whenever whenever AI overviews was truly becoming prime time and it was no longer in beta, I would rail against it because I still feel like at least with Google – and it’s weird I feel this way with Google – that one of the joys of Google is being able to do that deep dive of research. So when I saw AI overviews like I don’t want to be fed this information up top. Let me do my own stuff. That’s what I’m to do my own research. Now with ChatGPT, I mean basically it’s doing the same thing, but because of that conversational aspect that I have with it, for some reason I feel differently about it. And I don’t know why that is. And I don’t know if it’s just a me thing, but it’s just that there’s such a different vibe coming from Google now as opposed to what we’re seeing, you know, even Perplexity, the others as well.
JE: Do you feel more comfortable with – yeah, you mentioned perpetuity and it’s deep research model. I think ChatGPT. So I’m assuming that you feel much more comfortable with those because you can go back and check the sources even though at times it hallucinates even the sources.
HLM: Oh, big time. Yeah, it’s funny how it’ll come up with stuff and even on my own site and I’ll think, “Oh, that’s a brilliant article that I wrote. No, wait. I never wrote that.” But it’s a good idea.
JE: Yes. I actually talked I did one interview in which they found out that you know there was a hallucinated link and they actually went out and said hey – they actually were putting, you know, some people are putting links in or taking links out depending on what results they get from an LLM.
HLM: Yep.
JE: That’s a slightly different topic. So you feel that LLMs are very good. They’re the conversational search that you’ve been talking about for decades or thinking that was always on the horizon. So you’ve established that as a strength. Are there any other strengths that you find that you think are worth mentioning now?
HLM: Well, a lot of it – um I’m kind of looking at from the standpoint of the content, right? And as I mentioned earlier, the ability for a writer or anybody, even if they don’t consider themselves a writer, to be able to tap into something that helps them brainstorm and get over the the fear of a blank page and produce content where they might not have had time for before or it didn’t work with with what they were doing – that is that to me is an amazing transformative thing. And I feel a lot that way because the first time I trained someone to use ChatGPT – a small business owner, typical small business owner issues – they didn’t have a lot of time. They knew they needed to create a newsletter because they made money every time they created a newsletter but there was no bandwidth. When I showed her how to do it and pulled that information together and come up with a rough draft, she sat there for about five minutes and she kept repeating, “This is lifechanging.” And I really think it is life-changing for people who are able to do that. And even for writers that realize, you know, we have these voices that we love to use and and that we enjoy writing and why not make it a little bit easier on ourselves to be able to still do the writing that we want, but maybe do it a little faster and be able to write about different things as well because we have the mental bandwidth and can have and really dive into our creativity that way.
JE: Okay. Now, are there anything are there any specific tasks or projects or things that you think LLMs are particularly bad at and that we should avoid using them for?
HLM: Yeah, I don’t dig the SEO recommendations. Like, I’m part of a group that somebody was saying, “Hey, I’ve got a brand new site and I want to SEO it. Uh, what prompt do I write that can diagnose my site and help me fix it?” And the people who knew a little bit about SEO were like, “Dude, don’t don’t do that. There’s a whole other thing to learn about SEO that ChatGPT cannot diagnose for you and do. And I know for a lot of people they like to use it that way because I think it saves them in their heads money on keyphrase research. They might not know that there are other tools out there, whatever, but I don’t dig it for that. I mean, it can help provide general keyphrase ideas. Like, I had a client send me some earlier today where most of them were pretty good. Some of them were really bad. And again, if you’re someone who knows and you’ve got the human in the loop, you can say, “This is great. This isn’t. Here’s why.” Um, but for someone who doesn’t know and just takes that information verbatim and does all that stuff, they’d be creating an extremely keyphrase-stuffed site full of words that didn’t flow in their copy. So, in that way, I don’t really dig it.
JE: Okay. What about just using the phrase like “SEO-optimized” generally and not stuffing it with key phrases but just saying, “Okay make sure that this is good for SEO but also fits for these other criteria.”
HLM: I don’t use it for that.
JE: Okay.
HLM: But the reason I don’t is because I can naturally do that myself. But, taking me out of the loop and looking at it for someone else – yeah. If you were to say, “Here are the key phrases I’m targeting for the page. Are there opportunities to use this? Or use these differently?” – I’d write a better prompt for that – “Are they represented on the page?” That could be relevant. With that in mind though, there are just like any other SEO tool that writes content – um, sometimes they become a little bit heavy-handed on it or they miss opportunities. So, having that human in the loop again will be able to say, “Yes, this is this is a good idea or oh no, I definitely would not do that here. I’m going to choose to do something else instead.”
JE: Okay – um, and you’ve kind of touched on this a little bit, but maybe it’s worth, like, fleshing out a little bit more – now with SEO just generating, generating like a bunch of optimized content for the site and just getting it out there – given like the amount of content that people already have coming at them and — you know, I can’t remember which study it is but it’s pretty well documented – that people are overwhelmed by the amount of information that we’re getting and now we have AI and LLMs coming along and they can generate just that much more that much more quickly. Do you think we’re going to get to the point where people are just so overwhelmed that they just kind of shut down and stop engaging at all? Are we getting to the point where we’re at our breaking point?
HLM: That . . . . [Laughs.]
JE: Um, you laughed a little bit. [Laughs.]
HLM: It’s an interesting uh conversation because it’s one that’s come up multiple times over the years, of “Wow we’ve got these many videos uploaded to YouTube! How are people engaging with them all? Is this the end of YouTube because people – we just can’t.” And this is where – again I’m going to bring in the human element back into play, right, because especially say for B2B companies where they’re not going to be chasing an AI overview necessarily. They’re looking for people that will come in and know, like, and trust them and purchase from them down the line. What they might want to create and what I recommend they create are things like thought leadership content, things that people aren’t going to be able to find anywhere else and AI isn’t going to be able to just spin it together from a bunch of other stuff. And you can use ChatGBT to start creating it and create outlines or a draft or whatever. But that kind of heavy duty thought leadership content that people will know, like, and trust and and make purchase decisions on often has the majority of it is human-in-the-loop, looking at everything, rewriting the content, smoothing it out, making sure that it fits what people want to read. And ChatGPT, again, can give you that good double check. But I would not at this point – like if a brand new company came to me and said, “Heather, what should I do in terms of content?” I would not tell them, “Create a whole bunch of content that’s keyword-based and upload it to your website.” Because that no longer works anymore and will just drive them nuts. I’d be like, “Hey, let’s look at that thought leadership content, that juicy stuff that takes a lot of time, case studies, you know, opinions, whatever.” But that’s what people want to read from you because the other stuff is sort of like a listicle. They’ll – it’s quick information. They can get it from a lot of people, but what they can’t get from a lot of people are the subject matter experts brains that people have in their companies of what they bring to the table and how they help.
JE: The other aspect of it too is that these LLMs and AIs, they’re just they’re not addressing new problems. They’re basically telling you how people have addressed problems in the past and they’re synthesizing a lot of that information. But, you know, if you’re a business owner and you run into something that you don’t know how to deal with your own on your own, um, you’ve probably done some research to see if anybody else has at this point. Or if you’re running a business rather and you see a problem in the industry that no one has addressed, then you put put out the thought leadership piece and at that point you should be addressing things that LLM simply can’t because there’s no content out there yet to do it. I guess what I’m getting at here is you’re kind of touching on the idea that the thought leadership stuff is what needs to go out there first. You need to have a human start addressing these problems before LLMs can synthesize the answers.
HLM: Well, exactly. Um, yes.
JE: Okay.
HLM: And it and it’s part of a branding thing too for companies as well. It’s like they talk about now that the behavior is different of how people are accessing information is there’s a that more of a shift to what we knew from traditional PR is like get on podcasts, you know, get cited in in other publications, write articles that appear on other than your own site. All of these things help to build trust. All of these things are thought leadership. None of them necessarily go back to the old SEO strategies back in the day of “Write a bunch of articles that are keyword-based and compete that way.”
JE: Okay. And a related question, um, we’re kind of getting to the point already where it’s very difficult to distinguish what’s real and what’s fake. Um, this goes not only obvious obviously for textbased content, but there are incredibly convincing video avatars and things like that out there. Do you think that this inability to tell the difference between what’s a real actual person and what was AI generated, even when it comes to video, is going to undermine, uh, the trust that people have in what they do find online? And if so, like how do they mitigate against that or how would you mitigate?
HLM: Yeah. Um, for things like video, I mean, if you’re if you’re coming up with, “Hey, I’m Barb and I am the face of the company,” and people find out that Barb’s not a real person, never has been a real person, and that’s not really disclosed. Um, depending on like how people are connecting with Barb and that AI avatar, I could see that as being a major branding weirdness. I mean, how many times have you talked to a chatbot, realized it’s a chatbot, and got really frustrated with the company because you’re not allowed to talk to a real person.
JE: That’s ever happened to me, but I don’t – surprisingly — I don’t spend a lot of time conversing with chatbots, per se.
HLM: Oh, like customer service.
JE: Like, for customer service kind of stuff?
LLM: Yeah, exactly. And then you, like, start asking questions and then realize, no, wait, they’re not responding the right way. And sometimes you’ll even ask, are you a bot? And they’ll say no, but they’re a bot. [Laughs.]
JE: Right. I’ve heard many horror stories about that. Luckily, it’s not something I’ve had to go through myself, yet. Yet. Yet.
HLM: But with the trust thing, you know people are – people trust what they want to trust from the people they want to trust, right?
JE: Yes.
HLM: So, from a business perspective, I think building building the business brand as well as the personal brand of the main people within the company, I think is even more important now because we want to connect with those people. That’s why we feel so let down when somebody is doing something from a business perspective that doesn’t fit with our values because we connected with them and maybe that’s why we purchased from them. So it may be with a company that has, say, a YouTube presence and their CEO is publishing a lot and they have the thought leadership articles and and all of that against another company that has articles but there’s nothing forward-facing from the people who are making those decisions. Um, it will be interesting to see how people make that trust decision, knowing that if we’ve connected with that person, we may be more apt to just trust them anyway, even if we may have gotten information from these other sources.
JE: Okay. Interesting. Um, and I do have one last question. We’re a little bit over, so if you have a few more minutes.
HLM: Oh, yeah. Cool.
JE: Um, yeah, just tell us a little bit more about Writers + Robots and what that is, what it does, what the intention is.
HLM: Um, well, thank you for asking. That’s really cool. Uh, so Writers – um, to back up, I ran the SEO copywriting certification training for over 10 years. So it was an online training where I had both in-house and freelance writers go in and learn everything about how to set up an SEO content strategy. And I loved it and it was so much fun. And then things started changing so much with how AI added and folded into the mix that I thought, “Well, wouldn’t it be cool to to start a community which is a bunch of writers learning AI together, of how it can help with their workflow and what they can learn from it and integrating in the SEO parts, so they have like an all-in-one “Here is my community of how I learned this step solution.” And it’s been really gratifying and fun because like in our world AI is like almost last year, right? We’ve been talking about this for a while and to us is not necessarily new, but for a lot of people it is like the new thing that they’re trying to figure out and once they do of like, “Hey I can suddenly generate a prospect list in seconds using deep research and a really good prompt!” That is really eye-opening to them. So this group, I made it super inexpensive for folks to just come in and learn with trainings from me and other industry experts and a whole lot of fun community support.
JE: Okay. All right. Um, I think that kind of does it. I don’t have any more questions. Do you have anything that you would like to add?
HLM: Uh, let’s see you asked about Writers + Robots. So that was really cool. Um, I think I am good. Um yeah, I think I’m good. Thank you.
Â
 JE: All right. Well, thank you so much for taking the time to talk. Um, yeah, it was definitely some good insights on using AI and SEO together. Thanks.
HLM: Hey, thank you. It was fun. Awesome.
