Companies are realizing that an ongoing, consistent PR program can help them show up in search, whether through SEO or GEO (= AI-driven search).
Meanwhile, PR pros are bombarded by information about how to optimize for GEO. As practitioners scramble to keep up with the latest updates, it can be confusing to determine what _really_ matters when advising our clients on which approaches work best.
Spoiler alert: The goal is not to panic, but rather to view GEO as a new audience segment.
My guest Katie Robbert, CEO of Trust Insights, joins me to discuss:
*While GEO is now here, SEO isn’t dead – what matters for both?
*What type of content do machines prefer?
*How to get your brand to show up in GEO results
Show summary:
In this episode of PR Explored, host Michelle Garrett, a PR consultant, author, and writer talks with her guest Katie Robbert, CEO of Trust Insights, about how SEO and GEO (generative AI engine optimization) affect PR work.
Robbert explains GEO as serving a new “business to machine” audience—LLMs like ChatGPT, Gemini, and Claude—which summarize information from sources such as listicles and quoted articles, making PR placements and strong SEO fundamentals still essential.
Rather than throwing out playbooks, she recommends reverse-engineering AI results by asking for sources, targeting high-quality publications (including trade media), and measuring impact via analytics segments and “how did you hear about us” form fields.
They discuss EEAT (experience, expertise, authoritativeness, trustworthiness), avoiding low-quality AI content, and making “machine-readable” content more detailed across owned and Google properties.
Robbert also notes using constrained-source tools like NotebookLM and improving accessibility to help machine readability.
00:00 Welcome and Guest Intro
01:16 Why GEO Matters Now
02:00 Defining GEO vs SEO
03:29 Trust Insights GEO Course
06:46 SEO Not Dead Explained
11:05 No Need to Overhaul PR
13:19 Paid Placement and Authority
16:18 Measuring LLM Discovery
20:01 Are LLM Leads Better
24:30 PR Priorities for SEO GEO
27:07 Avoiding AI Content Slop
29:21 Google AI Mode Question
29:42 Critical Thinking vs AI
30:27 Sniff Test and Skepticism
33:27 Ask for Sources
34:07 Control the Inputs
35:10 Notebook LM Workflows
37:11 Trade Media Strategy
40:49 Press Releases for Machines
43:24 Wire Services and GEO
46:16 Start Your Own Publication
47:42 Be Everywhere for LLMs
50:50 Accessibility as SEO
52:14 Final Optimization Takeaways
Show notes:
Katie Robbert on LinkedIn: https://www.linkedin.com/in/katierobbert/
Trust Insights GEO 101 course: https://academy.trustinsights.ai/courses/geo-101-for-marketers
Trust Insights site: https://www.trustinsights.ai/
Full transcript:
PR and Search – What Really Matters When Advising Clients
Michelle Garrett: [00:00:00] Hello everyone and welcome to another episode of PR Explored. Pr. Explored is the PR podcast where we delve into trends and topics related to public relations. I’m your host, Michelle Garrett, a PR consultant and writer, and my guest today is Katie Robbert. Hello, Katie.
Katie Robbert: Hi. Thank you for having me.
Michelle Garrett: I’m so excited you’re here.
Katie is the CEO of trust insights, but I know her from the, the content marketing world and marketing profs, events and things. And, I have followed you and, respected your advice for so many years now, and I’m so pleased to have you here.
Katie Robbert: thank you. I’m excited. I think that, it’s interesting.
I worked at a PR agency for a while. I have a lot of friends and peers who are in the PR world, but I myself, like it’s, if you asked me to sit down and do pr I’d be like. [00:01:00] I don’t know. but I respect it. I understand it, but it’s just, it’s, I’ve been so adjacent to it for so long that I’m actually really excited that I have the opportunity to talk with you about where my world and your world really converge.
I.
Michelle Garrett: Yeah. and I, I, like I said, we, I could have had you on at any point in time and you would’ve had, insightful things to say, but I attended your marketing profs, session. Think Marketing Profs hosted an event about SEOI forget was like the future of. XEO or so there was some, I don’t know what it was called, but, but you gave a very insightful session there and I took a lot of notes and I thought it would be really relevant to, what people that, that listen and watch, PR explored would be, the work that they’re doing. So thanks for being here.[00:02:00]
I wanna, we’re today, so we’re talking about, of course we’re talking about SEO, we’re talking about GEO, which we’re gonna use GEOI think as the acronym. I know every day it’s something different for AI driven search. and we’re gonna be talking about it in the context of how it fits in with the work of PR Pros, but we will probably touch on some other things and we’ll try to define anything that, you might not, know.
We’re gonna use some acronyms here and there, which I am not always a huge fan of, but we have to do it. Do you I is GEO the term right now?
Katie Robbert: I think so. people make, they start that tongue in cheek rattling off. G-E-O-A-E-O-A-I-E-O-X-E-O is what I think Marketing Pros was referring it to, but that was more of the cross section.
I feel like I need to like make this simple, like it was the cross section of, [00:03:00] a lot of things. But yeah, call it what you want. You can still call it SEO, It’s just now GEO generative AI engine optimization, but G-A-I-E-O doesn’t really have the same ring to it, but basically what we’re talking about is SEO plus how you show up in a large language model when someone does a search in a chat box.
Michelle Garrett: Okay. I’m putting a couple links in here too because Trust Insights has a fabulous course on GEO and maybe you wanna talk about that for just a second before we hop into our questions here.
Katie Robbert: Yeah, absolutely. So be between myself and my co-founder. we’ve given a lot of talks about what GEO is and what we realized was that there was a real opportunity to do some education around.
the basic tenets of GEO, which is really, I like to call it technical, SEO, on steroids, [00:04:00] which we’ll get into why, SEO is not dead. Thank you very much. but yeah, so the course is 90 minutes of education around the core tenets of GEO and how to get started with each of those. And the good news is if you’re already have, if you already have a solid SEO.
Strategy. So SEO typically the three tenants are what? Onsite, offsite and technical. offsite being things like PR coverage, onsite being what’s on your website, technical being, more of those keywords, meta descriptions, all the traditional things. Then you’re already in good shape, so we go over that, but where also generative AI fits in to each of those.
Michelle Garrett: Yeah. And that would be a really great starting place for people, who are trying to get, get their arms around this. Because I think I, I really, I believe in following the advice of people that you trust. [00:05:00] And you and, Christopher Penn, your co-founder are very trustworthy, people to follow your advice and, yeah, I think I see advice all over the map.
I know I follow a lot of SEO people. I follow a lot of PR people and I get really frustrated by some of the advice. And it’s hard to know because one person will tell you one thing and another person will tell you something else. And so it’s just you never know, what, what to really listen to.
Yeah. But I think that would be a great starting point. And, it’s, a reasonably I think, how long is it? is it, how long was is the course?
Katie Robbert: The course is 90 minutes.
Michelle Garrett: Oh, okay. That’s,
Katie Robbert: self-paced. You can, I think, I forget how many modules and chapters and all of those details, but 90 minutes self-paced.
You can start it today, put it down, pick it up next week, you know it’s not going to expire. Immediately. you can certainly take it as you have time to take it. We [00:06:00] design all of our courses that way. We wanna make sure it fits into people’s schedules because, we all have a lot going on.
The last thing we need is to add courses and education on top of it, except for the fact that we all need it. Myself included,
Michelle Garrett: my favorite thing is the seven day, free trial. It’s oh wow. Okay. Seven whole days, man. I’m gonna
Katie Robbert: seven whole days
Michelle Garrett: clear my calendar.
Katie Robbert: And on day six and a half, you’re like, oh, I gotta do that thing.
Michelle Garrett: That doesn’t work very well. Anyway, everyone should check that out. So let’s talk. So let’s, so we ha, I ha I have some questions and we don’t have to stick to these by any means. And, please ask questions. If you’re listening in and have a question, we would love to answer it, if we can.
So please feel free to put those in the comments. so the first question I’ll read, let’s set the stage. SEO isn’t dead, but now we also have GEO. Can you talk about the difference? And you did that a little bit, but let’s expand on that a little bit.
Katie Robbert: Yeah, [00:07:00] absolutely. you basically have a new audience to serve with GEO and that is a large language model.
So when we talk about things like machine readability, that has always existed. That’s how the Google search algorithm work works. It’s a machine and it’s looking for, certain markers in your content of, helpfulness, usefulness. They have their whole. algorithm in terms of how that works.
So your content has always needed to be machine readable, but now with a large language model, an LLM, which, so for those you know, who just want those definitions, a large language model is something like Open AI’s Chat, GPT or Google’s Gemini or Philanthropics clawed. So when you open up that chat box in chat, GBT.
You’re accessing a large language model. This is your new audience segment that you wanna be thinking, how do I serve this audience? and by that, [00:08:00] it’s the machine. So instead of so we talk about B2B, B2C, this is B two m, business to machine, which is just another segment to consider.
You don’t have to do anything, bonkers and rearrange everything. It’s just adding more detail the more. Information you can feed the machine, the better your chances are of being, shown in that large language model. ’cause a lot of people are wondering, how do I show up for search if someone looks for, the best PR firm in the Chicago area, or the best AI consulting firm in the United States?
I wanna show up for that. It’s the old, how do I show up in, spot one in Google search. for Google, you’d follow the tenants of really good SEO. So you know, onsite, offsite, technical, if you wanna show up in a large language model in search, now you have to do onsite, offsite, technical, [00:09:00] and then also.
Cater catering your content for how it’s found. So the way that a large language model, at a high level, I’m gonna keep everything, super high level. I’m not gonna get into like very deep technical details. We certainly can do that, like maybe at a different conversation. A large language model like a Google Gemini or an open AI’s chat, GBT, it’s doing, you ask it a question, it goes and does its own search.
So it’s acting like a search engine within a search engine.
It finds what it thinks are the best. Rated websites the best top quality, and then it is saying, okay, I found 99 resources that I think answer this question. Let me summarize from those 99 resources what the answer is for this question that you’re asking.
And so it’s very similar to. Getting really good coverage because that’s what it’s looking at. If you [00:10:00] look at the sources. ’cause you can always ask for sources. If you’re doing deep research in one of these models, it gives you the sources you’re seeing things like listicles, you’re seeing, articles that people were quoted in.
This should all sound very familiar to PR folks. These are the things that you need to be doing anyway, but this is the data that these large language models are using to serve up. Their answers. And so if you are not in those resources, then you’re not gonna show in the large language model. SEO isn’t dead.
PR isn’t dead. None of those things are dead. Everything is very much alive.
Michelle Garrett: And if you, we were talking a little bit before, if you haven’t been doing those things, this is the time to do them. And if you have been doing those things, just continue. Maybe double down, put a little more resources behind those, behind SEO and pr.
I think this works very well right now, right into the [00:11:00] next question. It dovetails right into this. So let me put it up here. I, this drives me crazy. So a lot of people that I see in the PR world are advising their clients to just completely throw out their playbooks, based on, AI driven search, GEO, whatever we’re calling it.
can we talk about how. This is not as crazy a shift as some might try to lead you to believe. ’cause I have a feeling they’re probably trying to sell you something. So that might be the reason
Katie Robbert: most likely. Yeah, it’s not a crazy shift. as I just mentioned, what if you look at the sources that a large language model is pulling from, it’s the sources that your PR team or your PR firm would be getting you placed in.
And so it’s gonna be. the large language model is gonna look for sources like top 10 consulting firms of 2025. Those are probably placements that you’ll be wanting to get anyway for [00:12:00] your consulting firm. And then the large language model is gonna seek that out provided, they look at, I don’t know if I apologize. I don’t know if they necessarily look at domain rating of a website. I can certainly find that out and follow up, but I, I don’t remember exactly how they gauge what a high quality source is, but it’s probably very similar to how if you were looking in a tool like an hre sword, SEMrush would say, here’s your domain rating score.
So if you think about it in those terms, what is a high quality? Resource for you, and how do I get placement in that so that when a large language model is trying to answer the question of talk consulting firms, it can find that resource easily because you’re in there. the other way to like.
Reverse engineer it is do a deep research, project in any one of those tools and say, tell me the top consulting firms in this area [00:13:00] and then list all the sources and you can look at the actual sources that it was quoting and say, are these on my pitch list? Are these people I have, contacts with?
Can I get in any of these publications so that the next time someone’s looking for this, I’m showing up?
Michelle Garrett: can we, can I ask you, and this is just an offshoot that it just popped into my head. Yeah. So when somebody pays to get into a publication, okay. So let’s just say they paid for a Forbes article and they try to Sure.
Play that off like it’s an earned media, article, so Which happens. is, does that count? Is that, does that. Is that weighed the same? in search?
Katie Robbert: I honestly don’t think so. I could be mistaken, but I don’t think so because I don’t think the language model is looking at whether, it’s not looking at how you manage to get into that publication, it’s just looking at the publication itself.
Michelle Garrett: Okay.
Katie Robbert: It may [00:14:00] have some, algorithms in the background that are like, is this, are we seeing sameness across this, resource? Or is this resource highly regarded? Is it authoritative? it makes those judgment calls, but overall I think you can pretty much. A high level make some assumptions about what is considered, a good high level, high domain rating resource that you would wanna be placed in.
But again, I’d say your best bet is to, do I call it quick and dirty? Do a quick and dirty research project on the question you would want to show up for and take a look at what the large language model considers to be those resources. And then that becomes your. I would say additional focus list.
I wouldn’t stop doing to your question about throwing out the playbook, don’t stop doing what you’re doing.
You may just be, amending what you’re doing, or this may just become adjacent to what you’re doing. You also need to ask, [00:15:00] the first question is, why do I need to show up in a large language model?
Is that important to me? That’s where I would start, is do I need to be showing up in a large language model?
I don’t know. It’s gotta be personal to everyone.
Michelle Garrett: I think most people though are when they’re looking to hire someone, maybe not as many as I think, but in my way of thinking, everybody is going on and, typing online to find, who to hire.
Or maybe they’re asking a friend, okay, or maybe they’re seeing something on social media or whatever. But I do think, a fair number of people are going in and searching.
Katie Robbert: There’s definitely about, quite a bit of search, and especially if you look at search engines like Google, they have what’s called AI mode, and so that sort of is their standin for something like a Gemini.
So it is important to factor in, do I show up in a large language model? Should it be your only priority? Absolutely not. There’s still billions of humans on the planet who are using [00:16:00] traditional sources to find you. But it is now just another segment of your audience to consider. So no, don’t throw out the playbook.
No, you don’t have to overhaul anything. It’s just one more, it’s. It’s just a brand new segment that you get to, cater to.
Michelle Garrett: Yeah. But
Katie Robbert: don’t forget about everyone else.
Michelle Garrett: Yeah. and a couple things I hadn’t heard of the business to machine term, so that was, that’s new to me. And then, are, are your customers using, a search?
Are they using search to find you? I think that’s the another thing to find out, right? Because some professions they might not do that as much, but I think, for what we do, it’s pretty common. And I’m even having people come to me and say, I found you. I was talking to chat GPT and I asked these questions and you came up.
So then I know that it’s working, but maybe it doesn’t matter as much, for everyone. I don’t know.
Katie Robbert: and there’s a couple of different ways to [00:17:00] approach that. That is, basic analytics and basic measurement. And so you can take a look at your referring domains in your web analytics, like a Google Analytics.
You can set up segments to, to specifically capture. Gemini or open AI or philanthropics clouds, we actually have an instant insights and that’s, I believe that’s also covered in the GEO course. That’s very easy to, just create those segments within your web analytics, so then you can see how much traffic is coming from those sources to my website.
So that’s one way to take a look. So you can see, is it, am I picking up, any sort of traction with these large language models? Step number one. The other thing, the super basic, I’m gonna call it old school, is you can either ask people or put, on the end of your contact forms, how did you hear about us?
Michelle Garrett: [00:18:00] Yes. I think you mentioned that at improv and I was like cheering for my seat. ’cause I started keeping a spreadsheet and I probably only a few years ago and I actually, it’s a, just an Excel, it’s like a Google sheet. yeah. and I just, I literally write down every time I get a lead, where did it come from?
How did they hear about me? and so I think that’s, I think that’s great advice. And obviously if you’re a giant company it’s gonna be a lot harder, but for smaller businesses, which is, that’s a lot of my clients. it’s much easier to do that. And it really gives you some, great data to take a look at.
And you can also almost, if see seasonal, did work, pick up, did leads pick up, so I think it’s, a great, tip. I love that tip.
Katie Robbert: and for an enterprise size company, they’re probably. They probably have a customer journey that includes something like request a demo, talk to a salesperson, this, that, the other, someone fills out a form if they do it, whether they’re at a live event and they have [00:19:00] the tablets, or they’re doing it on a website, it shouldn’t, they should already have, how did you hear about us on there?
And so you’re either just adding an open field like other, and then type in I did a chat CPT search, or you’re just adding in. Generative ai. And so it really just depends on how you wanna capture that information. But if you’re not currently capturing, how did you hear about us? How did you find us?
Whatever that looks like, whether it be, person to person or on your website, that’s a great place to start. We’ve actually seen. On our contact form when someone wants to hire us for speaking or consulting, we’ve started to see a pickup in the amount of people who are listing some sort of a large language model as their source of how they found out about us, which is fantastic to see ’cause we’re.
We’re trying to do all the things we’re educating about, just to demonstrate that it works,
Michelle Garrett: And that, I think that’s, how I view it too. It’s like I’m the Guinea pig, so I’m gonna do these things and see how they work out and then maybe I can implement ’em for clients. [00:20:00] along these lines, I wanna ask this question ’cause I’ve seen some back and forth about this.
Is, there a difference when someone finds you via a large language model platform? as in it’s a more qualified lead versus, just a search.
Katie Robbert: I really think that depends on what you think is a qualified lead.
Michelle Garrett: Okay.
Katie Robbert: it’s, the terrible answer to that question is it really depends on how you are scoring leads.
like for us, email is one of our stronger channels. Organic search is a stronger channel. we could set up a whole different, large language model channel. Say like in our, attribution analytics, like
if someone comes in through large language models that score’s higher than someone who comes in through [00:21:00] social media, for example.
But that’s personal to us. I don’t think it makes it a more qualified lead.
Michelle Garrett: Okay.
Katie Robbert: To you, but to me. It really depends on this is the thing that’s tough because what you consider a qualified lead is not what I would consider a qualified lead. ’cause we do different things.
Michelle Garrett: Okay.
Katie Robbert: And so if you are finding, that you wanna rate someone finding you through a large language model higher
Then that’s personal to you? I could. Okay. And for the nature of the work that we do at Trust Insights, it is actually something that we would consider because that means we’re doing our job. People are finding us through the things that we wanna educate and consult them about,
Michelle Garrett: right?
Katie Robbert: So for us, that would make it a higher qualified lead, but that doesn’t necessarily mean that they have, money to spend with us.
It’s just the source of where they came from.
Michelle Garrett: Yeah, I just feel and this is what I heard, somebody I respect talking about was when you have that conversation with a, [00:22:00] a Chad EPT or whatever, it’s, you’re getting to the point of, you’re getting a, probably a very short list or a couple of people or whatever it is, but, the time you’ve gotten down that far, you’ve asked a lot of questions that kind of qualified those people more than just Doing a search and a bunch of people came up and you had to do more work to figure out, I think that’s where I’m trying to understand and it makes sense to me. ’cause if you have a body of work out there and you know that’s what GEO is pulling from and the person has heard of you maybe through something you’ve done, I feel like there’s a couple of ways where it would make it more, a less of a.
Path to get to the sale? I don’t know.
Katie Robbert: I think, I definitely think that there’s merit in that. I think it’s, definitely true in some instances. The thing that we are blind to is what people are actually searching within a large language model. That’s not data that those companies are ever really gonna give up in the way that, like [00:23:00] Google gives us Google search console data so we can see what people are searching, how they’re searching.
Or like an SEO tool, like an hres where you can see the keywords. and like where they went when they searched, those keywords, we’re not gonna get that data from a large language model. So it’s harder for me to say definitively.
that scenario is always true. I do think it’s sometimes true.
there may be something else you wanna tease out of that conversation when someone does reach out to you and say, I found you through a large language model. Because it could be as simple as I wanted to know the top consultancies in the Boston area and your name came up.
to me, that doesn’t make that person qualified. That makes them just interested in learning more. So they’re still at that top of the funnel. Whereas in your scenario, someone’s going through and getting that short list. You’re right. That does put them like closer to middle or bottom of the funnel, but we don’t know that because we will never get that data.
Michelle Garrett: Okay.[00:24:00]
I appreciate it. It just makes my head spin.
Katie Robbert: Yeah, I was like, that’s a very unsatisfying answer, but that’s what the answer is.
Michelle Garrett: that’s the thing, and that’s, I have a very, a small handful of people that I really respect their opinion in. Somebody in that circle has talked about this, so I’m trying to understand if it’s.
Plausible ’cause it makes sense to me, but then some people will push back on it a little bit. anyway. Interesting. so the next question, and this is where we’re gonna talk about one of these acronyms, and I’ll put it up in a minute. what matters for, both SEO and GEO from a PR perspective?
Katie Robbert: Good quality content, end of sentence. so we know that Google is still like 95 to 99, 90 9% of the market share in search. so that’s just true. that hasn’t changed with large [00:25:00] language models. Google is still the dominant four. And so we always look to what Google is doing with search to set the standard so they have, their.
Google search quality rater guidelines, which is what we use. It’s all in their developer, notes. You can easily find that information, but basically it tells you what they look for in a piece of quality in a, sorry, in a piece of content to determine that it’s high quality. And at one point it was a little bit more simplistic of does it answer the end user’s questions?
Does it provide value? Is it helpful? And now what they have is, and I had to write this down ’cause I never remember what this stands for, is the EEAT, experience, expertise, authorit and trustworthy, which totally just rolls right off the tongue. And so basically, ideally the content you’re putting out answers.
Or ticks these boxes. [00:26:00] So it speaks to your experience. So think things like case studies, your expertise. How long have you been doing it? Can you demonstrate that you’ve seen the good, the bad, the ugly, the authorit. you’ve been on stages or you’ve, had really good coverage through your PR program because people actually want to associate your name and your company with their, brand, their articles or publications, and then trustworthiness.
do you have a really good brand reputation overall or are there. Google reviews that are all like one star that you’ve never addressed. But hey, still just go ahead and trust me. so you wanna make sure you’re still ticking those boxes. That’s still true for GEO because it’s really, again, I keep saying it’s just like SEO on steroids, so your content still has to hit those marks [00:27:00] from a PR perspective and from a general search perspective, you wanna make sure that you are putting out high quality content.
This sort of gets a little bit into that conversation of AI generated content, and I think what the term is, like the AI slop, which you know. Not a great term, but it’s very descriptive. you wanna make sure that your content doesn’t fall into that bucket of sameness or mediocrity where, if you, if your PR team doesn’t think it’s high enough quality to pitch it
Why would you put it out there for the machines to also read? I think that’s a really good way to think about it is if the humans don’t think it’s good quality, why would the machines think it’s good quality? Because at the end of the day, the humans are really still the end user. You’re just putting another machine in that customer journey.
Michelle Garrett: I wish we could shout that from the rooftops, because I feel like that some companies don’t get that and they’re just, they’re replacing [00:28:00] the human, writers and content creators with, the AI driven stuff. And it just, like I just don’t understand what they’re thinking. I feel like there’s gonna be, like, they’re gonna have some negative, Results from that
Katie Robbert: more content doesn’t mean better quality. It doesn’t. It just means that there’s more mediocrity out there. And so what ends up happening in that instance. So if you’re using generative AI to generate a lot of mediocre content, great. You have a high volume of content. That mediocre content is what a large language model is gonna pick up on, and that’s gonna start to, maybe, damage your brand reputation because you lose that authoritativeness, you lose that trustworthiness because, it doesn’t really, it doesn’t really give your point of view.
It doesn’t demonstrate your expertise. It doesn’t demonstrate your experience.
Michelle Garrett: No.
Katie Robbert: Or you’ve programmed it. So that 20 pieces of content are the same [00:29:00] thought leadership content with just a few words change, like the machines catch on pretty quickly, just the same way that humans do, if not faster.
Michelle Garrett: Yeah. And I think we see some backlash as far as the companies that publish the list of the 10 best and then they put themselves on the list.
I’ve seen Oh yeah. Some stuff about that. and we do have a question, let’s, I dunno if the whole thing’s gonna fit on the screen here, but we’ll see if we can read it. People are gradually replacing Google with AI for basic searches. Google has noticed and added an AI option at the top. How do you see that impacting where people end up with their searches?
Katie Robbert: here’s the thing. Humans have to use critical thinking. They have to use judgment. AI does not change that. It gives them more options to pick from. we talk about how AI is spreading misinformation. That’s not a new problem that we have with technology. you [00:30:00] can go on any social media network and get misinformation all the way back to things like a live journal or a friend stir.
you can go before that with. Zero technology and people spreading word of mouth rumors, that’s misinformation. It happens. AI is just doing it faster and I’m, guessing that’s what you’re getting at is that how do you see, impacting where people end up with their searches?
You, the humans still have to exercise your critical thinking to see, does this make sense?
Michelle Garrett: I
Katie Robbert: think my co-founder, Chris likes to say, does it pass the sniff test? And so think about like when you take some food out of your fridge and you open up the Tupperware and you’re like, Nope, smells good.
It’s good for another day. Or Ooh, yeah, no, that’s gone rancid. that’s what he means when he says, does it pass the sniff test? You have to do that. And my recommendation. Is to try to look at multiple sources. So you know, if you’re getting an [00:31:00] AI search result, question it ask, is this factual?
You can ask the large language model. You can ask in the chat, how much of this did you hallucinate? how much of this is made up? How much is this a factual? And it will give you a response. It might not be the response you want, but you can challenge these large language models to say. Did you give me the correct information?
And sometimes it’s gonna say, no, you’re right. I totally hallucinated 90% of that. You can ask for sources. It has to give you the sources. If you ask for them, they may not be sources that you think are reputable. So my recommendation for humans. Challenge everything, go in with a healthy skepticism. Don’t just accept what the AI is telling you because it gets it wrong a lot because AI is programmed by humans who are also flawed.
Michelle Garrett: There’s just so much to this.
Katie Robbert: I told you, Michelle, like I was, I warned Michelle pre-show. I tend to go on these rants and I’m [00:32:00] trying really hard not to, but. the thing that people need to remember about ai, it’s just another piece of software and we need to treat it as such.
It’s not like a magic wand that’s going to it does a lot of stuff, don’t get me wrong. It’s very cool. I have a lot of like very cool Claude Cowork projects happening with autonomous ai. But it all still has human in the loop. So me being the human, so much supervision, I am not in a place where I feel like AI can just go ahead and, take over without any sort of human intervention.
That is just a terrible idea for a lot of reasons. And again, sorry, I’ll stop ranting.
Michelle Garrett: No, I, it just always makes my, brain hurt. Because you think, I think it’s like, it’s like you decide what you’re gonna believe or go by and then you get new, more information and then you’re like, wait a second.
But it’s good. it means that we are thinking critically and that’s, dying out. So we really, the more we can encourage. [00:33:00] people to really take, time to think through this. and, also when you get new information, it’s okay to reevaluate and rethink it again, and it’s evolving all the time.
yeah. ’cause AI is not new, the way everybody talks about it 24 7, it’s just, it’s wow, okay. I wanna know. If we can determine, what sources AI is citing,
Katie Robbert: you can, I will go on record to say I’m still skeptical that it’s giving you all of the information.
But you can absolutely ask for, and share your sources. And it will give you a list of sources. Just like in an academic paper, it has all of its citation, it has all of its reference material. You can ask generative AI to give you that information. and then you can decide whether or not.[00:34:00]
It’s citing sources that you are aligned with or not, which may change the way in which you use it. The way to get around, this is like a part two of the question that you didn’t ask is if I don’t like the sources That. Generative AI is pulling from you. The human are responsible for giving it the sources that you want to pull from and saying, only use these.
A really great way to handle that is like on a local machine that can’t do a web search. Claude Cowork is really good at working locally and constraining where it can go. You can turn off web search. A notebook, lm, for example, you can turn off web search and just give it the sources that you know.
Let’s say you have a bunch of academic papers that you wanna summarize. There is ways to handle the sources that AI is citing. If you don’t like the way that an open model like Aachi, PT or a Gemini is pulling the information, it’s an opportunity for you, the human to do the research and say, only use these [00:35:00] sources.
Michelle Garrett: Okay, so you can tell it what sources.
Katie Robbert: Oh yeah, a hundred percent. And there’s, notebook LM is a really good tool for that. so for those who don’t know Notebook, lm, part of the Google product family, you can basically use it like a really sophisticated research tool. So one of the ways that we use it, a use case, is we have, oh gosh.
’cause we just did an update. It’s 164 page sales playbook for our company. I’m not gonna read that. I’m absolutely not gonna read. I Okay, I lied. I did read it because that’s my job, but overall I’m not gonna keep rereading it. So I gave it to Notebook lm, and I can just ask specific questions and be like, what did we say we were gonna price this service at?
And it will pull up just that information or gimme a little bit more information about this particular. A deal, customer profile, and it will give me just that information. Another really good use case for [00:36:00] Notebook LM that I used was, I was working with a nonprofit, a Humane Society, and they were doing.
Interviews with staff and volunteers around what works and what doesn’t work with the overall process of the animal shelter. And I was able to give all of those transcripts to Notebook, LM and come up with a concise, here’s what everyone is saying. The problem is, here’s what everyone is saying is working.
And at no point is it pulling from resources outside of what I gave the notebook, which I find to be really helpful because I didn’t need outside resources. I just needed. Within that particular instance, what people were saying,
Michelle Garrett: whoa. Okay. No, that’s good. and if you have Google, then you already have access to the tool.
If you have Google Work Suite or whatever they call it now.
Katie Robbert: Yep. You can use it for free within Google Worksheet. I think. I would have to look, but I think you can access it if you don’t have Google Work Suite, you just, you may have a more limited version of it, but yeah, I would say [00:37:00] definitely.
So to your question, how do we determine which sources AI is citing? Ask it. Ask it for a source list. And if you don’t like the sources, take control over them.
Michelle Garrett: here’s a little offshoot question. So I work a lot with B2B clients, trade media. Is that, is trade media a good place for, companies to wanna show up in articles?
Katie Robbert: Yeah, absolutely. Because if it’s relevant to the kinds of things that their customers care about, absolutely. Because you still wanna be relevant to your customer, which is a human. And so you, your strategy is always gonna be personal to you. I keep saying, but what works for me isn’t gonna work for you.
So you have to determine your coverage strategy, your PR plan, your content strategy, your technical SEO, personal to you, not what the guy down the street is doing.
Michelle Garrett: I seem to recall at one point when I was talking to [00:38:00] Christopher, your co-founder, that he mentioned trade media is a good target because that’s the kind of place where LLMs do wanna pull from.
And he gave me a lot of reasons Yeah. Why that was. And so that’s been stuck in my head. And that’s, I just wanna check in on that question and make sure, ’cause it’s been a little while ago, but,
Katie Robbert: and if you think about trade media and how it works in traditional pr, for example, it’s more.
I always get this word wrong. Niche. Niche. it’s more narrow in focus in terms of the kind of content that it would cover. Therefore, it tends to be more highly authoritative because it’s not also gonna take in, Information from all of these other industries. It’s very focused in, the topic area.
So therefore it’s easier to have it higher rated. And then if you look at, your end users, your customers. So if you’re a B2B, I think he likes to use, my co-founder likes to use the example of ball bearings. [00:39:00] You’re going to very specific places to find out information about, ball bearings.
You can’t get that information everywhere, right? So those sources are higher rated and trusted. With the latest about whatever’s going on with ball bearings.
Michelle Garrett: Yeah, no, it makes sense to me. But I, I sometimes I see, because I think, again, in pr. There, and this is what we call the Wall Street Journal problem.
the CEO wants to be in the Wall Street Journal, but is that really where your customers are looking for information? And so that’s where I’m trying to draw a line where is, AI based search, where is it looking for the information? And you’re what you’re saying though, it is different based on every.
industry probably.
Katie Robbert: and if you think about your current strategy, I think that’s a really great example like Wall Street Journal versus like the industry, news. There’s nothing wrong with the general awareness of people just knowing you. You never [00:40:00] know who’s gonna refer someone to someone else who’s gonna know someone.
So you can still hit those, general awareness publications or general awareness top of funnel searches. But to then, to your question from earlier about qualified leads, you wanna really figure out where your customer is getting information, what they care about, and cater to that so you can, so both strategies can exist.
Michelle Garrett: Okay. Yep. Just curious. ’cause I just feel like, and also we see a lot of publications that are folding and I feel like trade journals have been one that I feel like is still a viable option as far as pitching in place in client news and stories. And we see reporters being laid off and whole.
Publication shuttering, but trade pubs are still viable, I feel. So that’s another reason that I, I the trade pubs, but, let’s talk about press releases. we’ve got, [00:41:00] about, I don’t know, 10, 10 minutes here left. And if anyone has a question, please ask. But we are gonna, talk about press releases a little bit about how we should really be thinking about press releases.
How do these play into the search? SEO and GEO.
Katie Robbert: So press releases are an interesting beast. It’s another piece of content. Again, this is, and please ’cause it’s not my world. Correct me if I’m wrong. A lot of times a press release is only used for announcing, a new thing. But it’s, we are so pleased, excited, thrilled, whatever we, whatever emotion we’re feeling that conveys joy about our latest product launch.
And then you the CEO says this is gonna be the greatest product launch in the whole world. find out more information, contact Michelle Garrett, hot Super. You never
Michelle Garrett: say that. They don’t say that
Katie Robbert: the diff. So those still have a place [00:42:00] in the strategy. The difference between that is that it’s for a human to read.
The machine learning version of a press release is. It is just what it is. It’s for a machine, so you’re gonna have more run-on sentences. You’re gonna have more keywords stacked inside that press release. So you can still say, we are thrilled to announce our latest. product release, but what you would wanna do in the machine learning version is get into more detail about what that is.
What the product is, who it’s for, who it serves, how it works. You don’t have to get into all of the ip, but like a little bit deeper detail, the way you would in maybe release notes for a piece of software so that people understand like what’s new, what’s changed. Whereas in a press release, people don’t necessarily look for that information.
So you just wanna make sure if. If a machine is picking up a press release, that it has enough information to answer the question of what’s new with Michelle’s [00:43:00] product? What’s changed? And so you may not get that from a traditional press release for a human. Because the human doesn’t necessarily care, but a machine does care about those details.
So it’s gonna be a longer, are less structured, makes all of our, writing skills cringe go out the window be, but it’s not for us. It’s not written for us.
Michelle Garrett: Yeah. Yeah. Do you have any thoughts about wire services?
Katie Robbert: We use them. We absolutely use them.
Michelle Garrett: Okay.
when you have one, I saw some data and I, wish I could find the source, I’m gonna check myself on this later, but Globe Newswire was cited, as the wire service that was being picked up more often in, GEO and I. What do you, do you see any difference in news wire services or can you speak to that at all?
Katie Robbert: again, not my world. Yeah. I don’t necessarily, I know what a, wire service is. Yeah. It’s still important. It’s still, in [00:44:00] terms of distribution, generative AI is worldwide.
And so it really depends. Again, it depends on who your customers are and where they’re getting their information, please correct me, if I’m misspeaking, I don’t know if there’s a, like local newswires versus domestic newswires versus global news wires, right? I think those things are really going to, those are the decisions you need to make in terms of where your customers are versus general awareness.
Michelle Garrett: Yeah. my own personal feeling about wire services is they’re expensive. First of all. They’re overkill sometimes, but now I feel like, yeah, maybe they’re important for certain, yeah. Press releases because of search. And I think the SEO value of releases was bandied about for a long time before the GEO came into the picture.
But now I feel like we’re back to Yes. You wanna use a wire selectively, and then I’m curious, and I’m gonna dig into which wire [00:45:00] services are, cited more often. I’m really curious about that. ’cause I know I saw that somewhere. yeah,
Katie Robbert: but this goes back to your question about can we tell what sources an AI is citing?
So go back and see which kinds of press releases might be showing up in the sources. And to your point, be selective. If a wire service is not budget friendly, then be really picky about it, just like you would anything else. Yeah, you don’t wanna just put everything everywhere.
Michelle Garrett: Yep.
Katie Robbert: Just be really selective.
It’s an opportunity to refine your strategy.
Michelle Garrett: Yeah. and that’s the thing, I still think there’s a value in a writing a press release and pitching it. Yes. But you don’t necessarily need to put every single press release on the wire.
And that can get really expensive. And again, you can pick different levels of the wire service, but that’s not what we’re talking about today.
Rights for another conversation. But anyway, yeah, I was just curious if they play a role in search, more so now than they were maybe a couple years ago. And I feel like [00:46:00] the answer is yes. So
Katie Robbert: yeah, the answer is yes.
Michelle Garrett: I don’t wanna rush through here and I know, I know Christopher has a question, but
Katie Robbert: it’s actually, he’s got a good question.
We can certainly talk
Michelle Garrett: about that one. Okay, let’s do that one and then we’ll, go on, we’ll try to fit these last two in. so Christopher says, given the importance of publications and the rise of sub stacks, what do you think about starting your own trade publication?
Katie Robbert: I think, If I’m approaching this from a very pragmatic stance, it would be what already.
So do a little bit competitive analysis. What already exists out there? What are you competing against? So you know, Chris, to borrow your ball bearings. example is there already a lot of. Content and publications that cover ball bearings. So how is yours gonna be different?
What is the differentiator in your content?
What is the gap that you’re filling in terms of, nobody’s talking about BA ball bearings in terms of their. Float [00:47:00] ability or their shininess, and those are questions that people are asking. So therefore, I can answer those questions in my trade publication. so I think it’s definitely worth exploring.
Yeah. If you use your substack in that way, if it’s just a pontification and naval gazing about ball bearings, probably not the best use of your time.
Michelle Garrett: Okay. Yeah. I think it, it’s a big undertaking, right? Yeah. But, obviously people are going crazy with the Substack. I don’t know how that’s gonna work out.
That’s my personal opinion. I’m like watching, sitting back a little bit watching. I wanna get to two more questions. So this question, when I heard you speak at the Marketing Profs event, you talked about how brands need to be everywhere now, and can you just tell us what you mean when you say that?
Katie Robbert: Yeah, absolutely. So one of the things that we advise and this, you [00:48:00] can learn more about in the GEO course, but at a high level. You wanna make sure that you have enough content for a large language model to scrape. So if you’re looking at your homepage and there’s two sentences and just a few links, a large language model is gonna struggle with that a bit because there’s really nothing to pull from that you wanna have information about.
Who we are, what we do, and it goes back into that machine version of a press release. It’s not for the human. You can even put the disclaimer, the following, block of content is for machine readable, audiences only, or something along those lines. But that’s where. one of the things we’ve done on the Trust Insights website is every page is now just the content is so much longer now.
It gives more information. It’s more of the story. It’s more of the, and here’s what we do and here’s anything that we can possibly answer about. What it is. So it’s getting those blocks of content, putting them, get one really good [00:49:00] about us and put it everywhere. And when we say put it everywhere, remember Google still owns the market share for search.
So focus on Google. Products and properties first. So for example, YouTube, we talked about YouTube as part of being everywhere. One of the things that you would want to do with YouTube is if you’re posting YouTube videos, make sure you have that you know about Trust Insights or about Michelle Garrett or about, this show.
So you had read at the top of this episode, a statement. That you now read for everything. It’s that consistency and putting it everywhere. Use the description boxes as often as you can in things like your business’s LinkedIn profile or your YouTube profile, and the descriptions for every single video.
Create that machine version of content. And just stick it everywhere. That’s what we mean by be everywhere. Like you can’t [00:50:00] actually be everywhere. So prioritize. But make sure you have those really solid about us. Here’s what we do. Blocks that are like, just stick all your keywords in that make it as readable as possible, but don’t fuss too much about the grammar.
Michelle Garrett: It’s so hard to get, people on a consistent, using a consistent boilerplate like in a press release when we have the about paragraph. So I can only imagine how hard this would be if you had a large company trying to get everybody to use that same block of text. yeah. But yeah, that’s what it takes, I think, and it,
but
Katie Robbert: that’s a human problem.
Michelle Garrett: Yeah.
Katie Robbert: That’s not a machine problem.
Michelle Garrett: But you can see, and when you do it successfully, I think you can see the results. it has, you can see the impact. Sometimes when you search on yourself, you could see that you’re coming up for certain things and maybe because you tweaked something or It’s interesting.
Katie Robbert: and I know that, we only have a few minutes left, but, something I wanna sort of address, because I know it was gonna be a question, is like the accessibility
Piece in [00:51:00] terms of your content. So a lot of times what we try to do, like for, accessibility for people who have lower vision, is we have accessibility readers. And so a really great way to determine whether or not your website is gonna be. Readable by a machine. Is to run your website through an accessibility reader because the machines are looking at it very similar to how someone. Would be using an accessibility reader.
So if the first, two thirds of what comes up in the results with an accessibility reader is just like images and links and jargon, machine doesn’t care. It’s gonna move on because it’s not finding what it needs.
So that’s a good way to approach is my website, is my content readable.
Michelle Garrett: Yeah.
Katie Robbert: By a machine. Accessibility readers are gonna be your best friend. you should be doing it anyway, however, It’s a really good sort of like double check. So now you’re hitting those two marks for [00:52:00] the people who need it and then also for the machines. So treat the machines like a person who needs an accessibility reader.
Michelle Garrett: No, that’s a great, I skipped it because I knew we were gonna run outta time.
Katie Robbert: I know. It’s, there’s so much to cover.
Michelle Garrett: but yeah, quickly, I mean if you, if as we close, ’cause we really are gonna try to wrap this up, if you’re trying to optimize your own intermedia for search, for both S-E-O-G-E-O, what would you do right now?
What are a couple takeaways
Katie Robbert: I would do owned first, because if you don’t have good owned content, what are you pitching? that, I think that’s like basics. so I would optimize my own content first. Again, look at an accessibility reader. Use your SEO tools. Look at, hf, SEMrush, whatever SEO tool you’re looking at.
Optimize your content, but then add more to it. So those pages that have less than 300 words of content. And then, the little toggle in your WordPress side is that’s great, but it’s not that long. Consider what it would look like to make that content [00:53:00] longer in a meaningful way. That’s a big statement, but there’s, again, there’s sort of those about us blocks.
So if you don’t have more content to add, consider repeating that you know about trust, insights about Michelle Garrett, about what we do, those kinds of things. Those are great places, especially if you don’t have a lot of content. So you’re optimizing for the machines. And then don’t skip your SEO basics.
don’t trade good SEO for GEO because GEO is good. SEO.
Michelle Garrett: Oh, I love that. That’s perfect. On that note, we could talk all day and I appreciate you being here so much. Katie, thank you so much. please follow Katie. Check out Trust Insights course. Look at their newsletters, sign up for their newsletters.
Both she and Christopher have a great newsletter and, they, they are just very, wise and trusted sources, so thank you so, much.
Katie Robbert: Thank you for having me. Yeah, we might have to do a part two.
Michelle Garrett: I would love that. [00:54:00] Thanks so much.
Katie Robbert: Thanks.
Michelle Garrett: Bye.
About the host: Michelle Garrett is a B2B PR consultant, media relations consultant, writer and author of B2B PR That Gets Results, an Amazon Best Seller. She helps companies create content, earn media coverage, and position themselves as thought leaders in their industry. Michelle’s articles have been featured by Entrepreneur, Content Marketing Institute, Muck Rack, and Ragan’s PR Daily, among others. She’s a frequent speaker on public relations and content. Michelle has been repeatedly ranked among the top ten most influential PR professionals.
Learn more about Michelle’s freelance PR consulting services here. Book a no-obligation call to talk about your needs here. Buy Michelle’s book here.
