How to Get ChatGPT to Recommend Your Brand | Ross Hudgens transcript

When a customer asks AI which company to buy from, your website may be only one part of the answer. Reviews, publishers, videos, and online discussions can all shape what they see. Today, I'm joined by Russ Hudgens, founder and CEO of Siege Media and author of the forthcoming WY book, Generative Engine Optimization: The Definitive Guide to AI SEO. We're going to explore what business can actually do to get recommended, how creators and affiliates fit into that process, and how to tell whether the work is producing customers.

Ross, welcome to the show. >> Great to be here, George. A lot of respect for you. So happy to chop it up. >> Fantastic. I love it. So, let's get into it. Um, first and foremost, what is the one thing you're doing for clients today that you would not have done two years ago and what result convinced you? >> Uh, a affiliate marketing is definitely one of them. Off-site mentions essentially is a big motion. Uh, we did digital PR for links primarily, but that wasn't on transactional bottomfunnel articles.

So, what convinced me of that is you just run enough clients through uh prompt trackers and put the right transactional prompts in there and then you look at what's influencing that and you just see so many third parties there >> that it just became a no-brainer at some point that we can obviously we can pretend just consult on people of how to do that, but so many clients obviously need the work done. So we uh decided to to add that service and and hopefully help them uh solve that problem. >> It seems to me that the biggest difference between traditional SEO and the new era of generative engine optimization is primarily third-party uh mentions and citations.

Do you see anything else besides that or is that the primary thing? I I think that's generally right. I mean, you could argue that this whole change has come with an evolution of less topfunnel content. That is in traditional search in a way as well with AI overviews. There's just less demand because AI summaries are doing that. So, you could call that it depends on who you ask. I I think it is still a lot of the same motion.

So, a heavier transactional emphasis seems seems larger. So that may also mean and it it's gone in out in popularity but still making a difference is first doing transactional content on your own site. So I'm sure you've seen George like the best X roundups that people do on their own sites too that uh are influential and uh the sometimes waning in popularity but still have a place in overall strategy in my opinion. >> Right. transactional content does it how much does it matter if you're doing it on your own site versus you know on a third party >> it depends on the industry so our most common framing is in consumer industries and then also consumer B2B which I know you touch in in the work you do there's more likely a probability that there's affiliate sites and affiliate creators and creators out there that are doing those third party reviews in like transactional niche B2B the long sales cycles generally don't support those affiliate motions.

Therefore, that audience doesn't exist. So, a lot of the gap of who are those third parties is not there. So, the solution is often created on your own site. So, for those niche B2B folks, we see more need for commercial intent on-site content uh all things equal. How does a brand go about researching these transactional prompts that their end users or their end customers are asking? Are they using just you know an AI prompt tracking tool or is it something else? >> Uh our way of doing it and suggestion is to use uh traditional search as a proxy.

So if you look in traditional search volume and see that people search at high volume for say project management software normally we'll suggest that you map that to a query like what is the best project management software and if you use exact framing that already clearly has demand on traditional search pretty strong argument that that will carry over in some form into AI of course it'll be longer tail >> but we'll say carry that for the core use cases of your product and then we also suggest doing something called product page mapping.

So think through the industries, the the roles, the business sizes that you you fit and have pages that support those too. And ideally they have search volume also like you could do SEO agency for healthcare that has search volume, right? So those things are are good proxies and you can ideally have a map of all those and then if you get that into your prompt trackers, normally you have a directional starting point that's that's helpful for for visibility.

[snorts] I want to talk to you about visibility because that's something that I'm curious about and I still after all this time after researching and you know talking to people I still don't quite get how do brands go about measuring their AI visibility given that these are generative um engines right so so they generate an answer every time and each time you enter a prompt it can be something different the answer will also depend on the user us history, their location, their time of the day.

I mean, there's so many variables about this, right? So, when when uh these these visibility companies say that, okay, we'll help you, we'll give you a score of how visible your brand is in AI, how do they go about this? And how really accurate is this information? >> It I mean, you're right. No one's going to replicate the exact same answer. But the analogy I like to give is these outputs are like a compass, not a GPS.

So you're not going exact, you're not an exact point, but with enough data points, daily tracking, enough sample size of prompts, you're probably directionally relatively accurate. So that kind of compass component of like we're going the right direction. It's directionally probably reflective of what the the average person is going to get for that thing. That feels reasonable to me. >> We we have started to try to >> add layers to to actual traffic tracking and stuff which we could get into at some point, but that's that's generally how I feel good about it anyways. >> Yeah. >> Okay.

So let's say that a brand comes to you and they're like, "Ross, we got ad dollars. uh we got you know budget we would like to but our visibility you know we did some analysis you know we're not showing up we used to be number one in Google and everything is great um our our rankings are still there but clicks have dropped and now we are getting extremely anxious about our um AI visibility with that case what would you do what's your strategy >> yeah uh our our so start with the prom tracking so we have a good sample size of data.

We can infer what's missing from the strategy. So, we'll open that up and we'll see what the diagnosis is of potential angles. So, there may be a high evidence of UGC content. That can be YouTube, that can be Reddit, which is going down a little bit recently. It could be uh editorial articles. Uh I'm referencing Peak. There's a lot of great trackers out there. Editorial is typically an affiliate play. There can be some non-affffiliates in in certain verticals that could also be referred to as reference URLs, but that will often be a signal that maybe we need a program of some kind or otherwise we need a motion to to generate those citations in particular.

And then we'll often see in say those niche B2B verticals, it'll be a lot of competitor listings and those will most often as of today be listicles that are showing up. there is more of a diversity of longtail pages, but that that's one core component. And then a final thing that I still don't think there's a perfect tracking device for is the brand comparison. So, we did an analysis that showed the people that had the most AI search traffic actually correlated most from a Spearman uh point of view with those who had the most versus articles on the site. >> I think that has something to do with like you're helping navigate that conversation for LLMs to pick you more often. you're giving them the guidance and also you're it's less competitive so you're owning that they may come to your article but it wasn't just for versus articles traffic it was just the entire site had more visibility so making sure you're really leaning into versus articles is a component that should be in trackers as well uh but in most industries having that I think ups the urgency of doing things like that >> you mentioned there in the beginning that Reddit is down a little bit lately Why is that?

And yeah, how are you seeing that? >> Yeah, so there there's studies that I think propwatch debuted the study, but they made saw that chatbt was now citing uh Reddit far less. So Google is still very very prominent all all things considered in its usage of Reddit. It's also had public conversations in in the news that they may not renew with Reddit. We still don't know the kind of finality of that. So there's kind of uncertainty around Reddit in regards to that.

This said when I do chat GBD prompts, I still see Reddit there. And one thing I've been saying recently is like if your audience is prominent on Reddit, be on Reddit. And you that's a place of influence. And that's something I think people kind of strategically stop who stop at 101 stuff miss with the 2011 of of if you're influencing if if you're where someone is influential that can be powerful. example I like to give is uh if you search best CRM software last time I did that a Reddit thread shows up number one.

So within that it's worth like a huge amount of money in terms of click value. But if someone's researching their own listical for best CRM software they're probably reading that a high frequency. So that's a place of influence even if it's not directly impacting chatt as frequently as it did. someone else is finding that article and referencing a huge percentage of the time in their creation. So, it's like a snowball thing that so much of this listical and off-site stuff >> contributes to is just get where the eyeballs are and eventually that'll snowball if you're actually good and worth being there. >> Yeah.

What came to my mind right now is let's say that a brand starts spending uh you know money on AI visibility and how do they go about measuring the effectiveness of you know their AI visibility strategy um how how can like let's say that they start working with third parties with YouTube creators they start working with um people who help them uh get organically uh mentioned in in Reddit, how do they measure the return on their investment on that? >> So, it starts with tracking.

So, there's a few ways to do that. We can go down a rabbit hole, but one of in B2B, one of the core ways we recommend is having a selfattribution form that says how did you hear about us? So, if the user indicates how they heard about you, that's a pretty good indication. So you can do that percentage split of true AI influence that you have because attribution and G4 and other tracking is just generally pretty challenged.

The alternate routes for that are do it post conversion if you're worried about conversion rate. You could have a multi-choice. The problem with multi-choice is people will just kind of pick an easy solution of the list generally the first one. So that's where selfattribution is is powerful because you don't get fake answers generally and they'll be more likely to also surface new channels that you didn't even know are influencing you.

Like we had Claude coming out of nowhere. That's the example I like to give and we would have even known that if we hadn't asked that. Other things that I've been thinking about recently, uh Eric Woo is came on our podcast and he opened my mind to some interesting stuff RAMP is doing. One of the things he mentioned that I think we're almost certainly going to bring to clients is if you carve out direct traffic to the homepage or sorry direct traffic to your site if you exclude the homepage most of those other URLs are from AI because who is going to type in random long string URLs that's unlikely to happen and you could look at that preach boom like uh late last year things of that nature and look at the before after to kind of have confirmation there.

There will always be some exceptions like you may have a careers page that gets bookmarked, maybe you have a calculator that gets bookmarked, things like that. But if you exclude everything and then everything else is like a decent chunk of AI traffic almost certainly. And then the the variable he added on top of that, which I don't think he's got precisely correct, but it was an interesting thought process was AI behavior to the homepage is different than our direct traffic behavior prior to LLMs.

So we could compare cohorts. We now hypothetically could say, "Oh, this type of user who normally engages in this way is actually from AI." And if I group that with AI search actual traffic and compare prior to chat GBT and after, we may actually be pretty close and like that decline in overall. And he mentioned they're actually up. They're ramp. They're super highly funded. So, I'm sure that's part of it. But that gave me some at least more clarity of how we could get a little more precise in this thing and get closer to measuring that ROI you're talking about. >> Cool.

So when you advise brands to work with creators and um uh outside websites, how do you decide which ones are worth pursuing? Do you start with something that is already cited or sources that actual buyers trust? like what's your decision matrix there? >> Yeah, good question. If there are LMS's within the prompt tracker, we'll see influential YouTube sources. So, it's possible we'll see age sources. So, maybe you did the best video of 2025 and we saw that you published that one year ago.

That is a reasonable thing that you may update in one year. That's a good person to reach out to maybe immediately start to build that relationship. That's one variable. The secondary one would be to look to tangental spaces. So like if we're in CRM software, maybe I'll look for someone who did CRM software for a certain vertical or maybe it's just a different kind of sales software like sales like Gong type reviews.

If someone had reviewed Gong and I'm a CRM software, that's pretty good chance they may review mine. And if I can tell that they're actually driving LM influence, I can measure that for that Gong review they did. That's a pretty good indication that if they build that CRM software YouTube video, I will will get that same outcome. >> And then of course just influence, right? So I can go to the channel, is it well done?

Is it quality like your your videos are? Those are of course more bullish things and generally as an SEO evaluating like how consistent are they and like how they define titles, those types of things that are likely to kind of drive the output we want without having to handhold them. Seems all things equal pretty pretty safe. >> Yeah, I love B2B because it's just so high value. Um, if you look at my channel from the outside, you know, you might look at the views and you'll be like, "Oh, okay.

Well, this guy is, you know, he's barely getting any views, right?" And to some extent that's true, but the value of the views is so not equal, right? I can be getting millions of views, but if I'd be doing, you know, gaming or entertaining entertainment content, I'd be getting RPM of a dollar or $2 or something. But because I'm doing something like, you know, best CRM for X vertical or whatever, every view that I get, because it's a long form, you know, quality traffic, uh those convert super well.

Sometimes my videos might get a few hundred views and then those convert to like a few thousand dollar worth of uh of a customer right away. And yeah, that's that's why I love it. That's why I stick in it because it's just >> you can get you can get very far by just having very little. And that's also I think what brands get wrong because sometimes brands reach out to me and then I I pitch them, you know, my price and how much it costs to work with me to create videos and then they're like, "Oh, no.

We don't want to do that because you you know, you don't get any views." And I'm like, okay, so you rather work with someone who gets way more views, but that is not targeted traffic. It's not, you know, not the audience you wants and pay them rather than work with someone like me who might not get so much views, but that's the your user like that's your ICP right there uh looking for it. And that's I think also the beauty of like search traffic as well on YouTube and across search engines as well. >> Yeah.

I mean the the whole thing we were talking about before call I mean seems like I mean YouTube with with Reddit declining a little bit YouTube is seemingly number one in most places. We'll see how that holds up. Of course I'm sure that's in uh Google's best interest for that to be case >> but uh it's definitely top of mind for for a lot of people to be there. Obviously, it's so hard to do well and you're talking about your your quality standards going up and it it makes sense as a byproduct of all this demand that's pouring in. >> Yeah, there is a lot of demand.

I tell you that because it feels to me that there's not so many like B2B creators. So, I get I'm not even going to lie 50 to 60 emails daily of outreach. And these brands are like aggressive. They they don't send just one email. No, no, no. They they can send sometimes like two three emails a day. And [laughter] they're like, "Hey, why aren't you replying? Like, let's work." And I'm like, "Hold up. I'm trying to record some videos." [laughter] >> Yeah.

It's wild west out there. It's impactful. I I I'm curious if you have any like unique, peculiar LLM influence things that you find people asking for you in those asking of you in those videos. >> Um, what do you mean? >> Yeah. So, one thing we've been recommending because if we assume Gemini is best at kind of rendering videos just because it's YouTube, the [clears throat] text is is is valuable, the description area.

So, when you could nudge an influencer and I I forget if you do this, I haven't looked at your descriptions recently, but say it was HubSpot and it was like best CRM for marketing and you put that co-itation text around the jump to where HubSpot best for marketing is seemingly that would be something that Elms reliably would pick up in that video with some confidence. It's stuff like that or is it like they need to be first with this exact sentence description in the audio?

Of course, you're editorially driven, so you're not gonna give them necessarily that discretion, but those are kind of things. >> Yeah, I'm curious about. Yeah, >> the video itself. So, there's I think there's a couple of components, right? Um the title, the video transcript itself, and then the the description, right? Um like you said, we obviously have the editorial uh discretion. So, so I do have brands telling me, "Hey, uh, please make description this." I'm like, "No, that was not in the agreement." Like, we we didn't we didn't agree to this, right?

Um, but yeah, to get it's very important to get the script right. Uh, because that's what obviously uh the LMS pick up. Um, talking by the way about LLMs as well, it's very interesting. When I was doing these podcasts in the beginning of 2026, I was talking to a lot of people and we were talking about like the evolution of the use of different models, right? I started off with Chad GPC, then I went to uh Gemini, then I discovered Claude and I went really deep into Claude and then my Claude started becoming annoying recently, like you know, maybe a month or a couple of months ago.

Every time I asked it something, it didn't just do it. It It started pushing me back a lot and hey, you should consider this before I answer D. And I'm like, come on. I started getting really frustrated. And then my graphic designer was like, you should try Chad GPT for image generation, for thumbnail generation. I tried it, Ross. I kid you not, it is amazing. like Chad Gibbt in my uh world made a huge comeback with the whole Astra thing as well.

They can now edit videos. They can the image generation is just topnotch. I generate my own like thumbnails with it now. And and it's it's really really good. And I think my hypothesis I'm curious what you think as well is the more people use the more feedback an LLM gets and the the better it can become. Is that is that a case you think? >> Yeah. Yeah, I think that's that's correct. An interesting thing I've been thinking about recently that's a sort of a byproduct kind of counterfactual on that is at some point does that defined excellence or framing start backfiring because we see it so much.

The example I like giving is we'll see you often see a eyebrow in on landing pages. >> Yeah. >> Or on decks or anything. That's a AI trope, right? like and normally we we we recommend that before AI but now we all see that we like is this made with AI even if it's technical best practice >> if we see that best practice now flood everything it may kind of backfire not not saying it's the case in what you're doing but that's that's only like counter consideration I've been thinking about >> when we create things is is does this some does this have the chance of backfiring because it has that characteristic >> I wonder how much do people really care about if something is created with AI?

So, my workflow currently is >> I have Whisper Flow. I don't know if you've heard about it. Um it's it's just >> it's the best thing that happened in my life. [laughter] So, I use Whisper Flow and then I use Chat GPT. And I don't like typing. I'm not good at it. But of course, I don't just want to be like, "Hey, reply or draft this post or something." I will talk my thoughts and my brain dump into it and then it will polish it up and then my thinking is okay well why would I spend so much time perfecting every word, every sentence?

Like English is not my first language, right? When I can just brain dump what a broken English into chat and it will beautifully write this post or this, you know, whatever and then I can just publish that. Wouldn't doesn't isn't the value for the end user higher that the post is higher quality even though you can see that it's AI or the same thing with like websites. What isn't it better to have an AI code your site uh rather than spend you know weeks doing it yourself or spending thousands of dollars doing it.

What do you think? >> Yeah, I I I agree with you. The framing I like to give and I and I said this in the book a little bit was if you give AI something great and have it do something with it. That's the best best way of using AI. If you have it simply like >> create a blog post on >> influencer marketing best practices from scratch without any context from you that's going to be terrible or at least redundant not very valuable.

So in your case there you're giving it your thought leadership your of course experience via voice and then it's it's essentially an editing function which is a great way to use AIS is as an editing function or ideation function >> um as well. So on the website my only kind of like thought process there depends on the product like if you're just emailing someone casually I don't think there's a lot of risk there but let's say you sell a $200,000 product like we do sometimes. uh if you feel that a lot of that's being outsourced to something that could be a click of a button one you may feel good about that in some ways if we can obviously communicate that balance which we still do we say we use AI because we do use AI but you still got to create the best thing to you're trying to sell that you have the best thing so a lot of time when you use AI for starting point you're the same thing so >> I like the thought process of giving it something great >> like It can be design, thought leadership, some kind of framing.

It could be the branding style guide. And I know to your point, George, I mean, that stuff is getting better and better where if you gave it a great branding style guide, it's like Apple's branding probably pretty solid that it's going to output. I don't think it'll be Apple yet, but it'll be pretty good. Yeah, it's getting so increasingly good. It's it's just this exponential. You really feel it, you know? It's like you're on this crazy wave going up and it's just Can you walk me through a case study with one of your clients?

Um kind of what was their starting point? What did you do? What did it cost maybe? And uh what commercial outcome changed? >> Yes. So one of them that's in the book, so we have three case studies in the book is is Zapier. They were kind enough to share that with us. the AI integration um tool. They're I think the fifth most cited domain in LLMs right now across transaction and bottom funnel and they have a core product advantage because of that uh because of their product type because they allow essentially they do a lot of new things now but you can zap between tools of course.

So their thought process is create transactional content and you can uh get front and center in that conversation when they're now thinking about using it with other tools and that's been fruitful for them. So what we've done is we did really high quality uh content uh around those kind of roundups the versus content they were very early in that in that world. We were pretty high quality not very pretty very very high quality in approach.

They have a very high editorial standard, which I have a lot of respect for. We'll create really nice images above the fold and happy to share links. I'm not sure exactly how many people will be listening to this versus YouTube, but um we'll we'll do really deep testing. So, it's it's not surface level. I'm using AI. We'll try to get actually access to the tools wherever possible. Test out those tools and do the comparison.

So some of the content we've done is digital PR content as well. So it's not just the transactional content which I talked about but within digital PR we'll do surveys that tie to their product offering and then we'll pitch these out and Zapier has some strong visibility. We'll we'll run this by around 500 to a thousand people and in doing this they'll get a lot of high quality placements and what this will also do is this will give them data that is uh fruitful for them from a repurposing standpoint.

So we can use this data across other content update that content more frequently with productled data and that will allow them to uh generate freshness that differentiates them in some way or at least that's the argument of it. So, it's kind of that twopart like transactional content that's really high quality and editorial versus content scaled out because of who they are. And they also do have a a a lens towards of course including themselves where it's appropriate and those are doing really well also.

So, they have I think to your point George like 78% visibility. What does that mean exactly? >> But but it's doing pretty pretty well in their market. They're obviously well known. I'm sure >> you know them well. >> And those and those comparison uh articles, they're posting them on their own site. >> Correct. Yes. >> Okay. >> Does it make sense for them to reach out to someone else and get, you know, this comparison article to be published somewhere where it's relevant for their uh niche? or does it is it enough that they're doing it on their own site? >> Yeah, I mean Zapier is highly authoritative just because they've been around so long and I think they would want users to drop on the main site.

They definitely get linked to at high frequency. >> Yeah. >> But um yeah, so far they they've only done it on on their own site and seen a lot of fruit that way. They also, the final point I'd make, we talked about Reddit. We don't help them with Reddit. We don't do all their content. So, we're only a small part of that pie. I can't take full credit. Our team can't take full credit. >> They have their own subreddit, uh, our Zapier, and they're really engaged there. >> And I think one thing they do that very directly influences LM visibility is one, they're they're building fans.

So, they're engaging with them. This is leading to downstream uh positive uh mentions on on Reddit for those kind of articles we talked about when people talk about automation platform. Zapier is always going to come up and the other thing they can do is on their own subreddit they will occasionally have these organic posts that come up where people actually talk talk about Zapier. And of course there's true organic pros and cons but when you own the community of course you have control about whether or not you want to keep that >> how you want it to be ideally framed.

You could actually or I'm not saying Zapier did this. They I don't know. But if I was a brand, I would go to my top customers be like, "Would you be willing to post a review of us on our own subreddit?" That would mean a lot, be appreciated. And these are the kind of like compounding things that probably uh really have added up for them. >> Nice. Wow, that's fantastic case study. Thank you for sharing that. [laughter] >> Cool.

Yeah. And one final point is to to your point, George, uh Dan Slagan, the CMO, did share on LinkedIn that 40% of their visitors come through LM LLM, which is wild for that's really high. Uh and then 25% of their signups are from LLM. So they have that attribution. >> I uh I can't speak to exactly how they do that internally, but that's pretty good indication. It's not just uh fluffy. >> I can I can totally see that happening.

Some someone is sitting there in the office and they're like, "God damn it, it's not working. Chad GBC, how do I connect this one thing to this other thing?" Well, you can totally use Zappier for that. [laughter] >> Exactly. Yeah, they they've carved out the market doing a great job obviously, >> but they got a lot of competitors now as well. They got Make.com and some other other ones as well. Pipe dream or something like this.

Yeah, I mean SAS everyone obviously it's becoming more and more competitive for everybody but um they're first so I'm sure they they're they're doing a decent job at hopefully staying ahead. >> What's your take on SAS palopsis? >> Is that a thing or is that another thing? Because we've heard a lot about it, right? Everybody's vibe coding their own tools now. Yeah, it's uh I I think anything complex it's it's just that like we have a head of auto automation at Siege and at different points I've considered one do we want to build competitor do we want to build LM tracking tool do we want to build a um content workflow tool and I I just this could evolve but I I haven't jumped to that one I think it just needs to be done really well these these companies are shipping features so fast and staying ahead of the market that's challenging to do.

I think a very simple vibecoded product, sure, but like anything that you're trusting your credit card to I I just struggle with that being the thing. So, I think upmarket complex SAS, multi-feature, elegant so far, I think a lot of people are reporting that we're hopefully on the crest of like the the negative perception of SAS. That doesn't mean it's going to be smooth sailing. For sure, some people will vibe code some new tools, but I think those are mostly like those s single feature tools that are that are most challenged. >> Yeah, I would agree with that.

Um, I think the people saying that probably [laughter] maybe they never tried doing it themselves because once you start doing it and then once you're like, "Okay, I'm gonna build my own project management tool." And then you just start to go through features and then debugging and then this and that and it's not working. You're like, "Oh no, I just rather pay 25 bucks per se for this thing." [laughter] >> Yeah, exactly.

The marginal value and we've seen this with our own tooling for what it's worth. Like we have vibe coded some stuff internal use and just our head of automation's time went from 100% build to slowly it's like 10% maintenance, 20% maintenance, 30% maintenance, 40% maintenance. >> Yeah. >> And uh that's kind of the unwritten cost that compounds for sure. >> Yep. I built um I built a tool uh it's called an invoice submission system.

I work with uh lots of freelancers and my whole problem was that I well first off I need to pay these people. I need to verify the tasks that they're submitting me uh submitting to the work, right? How much they're billing me? Um because we pay per minute. So are they billing me an accurate amount, an accurate number of minutes, etc., etc. What is their rate? So there's like bunch of complexity and and some formulas there.

And then one guy reached out to me in the team uh in my Slack and he was like, how are you um doing this like manually? He's like, oh, let's build a tool for that. And it was this in it's we built it. It took a long time. It took like three months to build. And guess what, Ross? Every month I have to message the guy and be like, "Hey, something is not working. Hey, someone is having a hard time submitting an invoice, right?" And this might be a minor thing, but you have you got to have that person there who goes in to the NAN or whatever it is, does some computer magic, and boom, you know, it's it's then it starts working again.

But it is helpful though. It is it's saving me a lot of money and time. So for sure uh but I wouldn't be going and and you know trying to build my own project management tool something like a ClickUp or whatever because yeah it's just it's it's not worth it. I'd rather pay for it. >> Yeah. [snorts] If it's it's a distraction from the core business model, I still think it's probably not worth doing in a lot of instances.

Uh but the people have developers on staff they know that that's like their core competency pro potentially worthwhile u for sure but um that's not necessarily us >> super laser focus is the key right from everything >> that I've studied and read so far. >> Agreed. >> Um so this is an interesting one. How do you choose prompts without accidentally creating a test your client is guaranteed to win? That's interesting. Yeah.

So, one way we've been evolving this, we talked about the versus prompts, wanting to monitor those. I think of those as like one, they should be a category. So, you can do most tools now will have topics or categories and then they'll have tags. So I would generally put all ser core products in a category. Say it's for us it could be like digital PR could be content creation it could be uh affiliate marketing that we talked about that could be each categories we have prompts around and then tags could be things like industry small business other variables that you may kind of think about um generally and one of the things I realized about core prompt tracking for like year one of doing it year two of doing it we were inclined to report on this aggregated metric of overall percentage score to clients.

And the challenge with this is that it disincentivized adding more prompts when it when you don't want to change that score and decline it and have a lot of questions from executives about why your score went down. You don't do that. So, we've eventually evolved now to categorizing reporting. So, we could report on digital PR specifically. And sure, we could add overall, but we need to be clear about the categorization sections.

And one other core area of that around prom tracking is the models you pick. >> So there's so many variables here that really swing things, but something I've been telling people to stop doing is stop tracking perplexity. >> So a lot of models will immediately track perplexity or or a lot of the trackers. And the reason for that is that a lot of the data similar web has a great X account. I love I love staying on top of of like change in model share.

It's less than 1%. It's continuing to decline. I don't see a clear route. It's going to come back suddenly. Of course, do the self attribution. If you see it in your tracking, you see it in referral data being above one to 5%, sure. But the challenge if you're using a model that's less than 1%. A lot of these trackers will equally weight the models. So it's basically saying perplexity is 25% as important as maybe the other three models.

So that's problematic where maybe one URL has a ton of influence. It's some like spammy AI site and that one makes you think you need to go out and pay that person $10,000 to get that listing because of the way it's it's modeled. >> They could solve this with better waiting. Maybe some some trackers are already doing this. Sure. But making sure you're picking the models that are helpful, which means you can segment it even by model too in your reporting.

We can go a lot of routes whether or not it's worth the energy. But uh that was the overall insight for me was like, oh man, we could be making completely the wrong decisions because we're tracking perplexity here and that's not a great thing. That's an interesting take. Do you What's your opinion? Do you think there's going to be one winning AI LM model aka the Google emerging and it's the monopoly of search or do you think there's going to be a sea of different LMS for different purposes? >> Uh it feels like a big three really.

Um I mean Anthropic's been a little shaky as you said earlier on this recording. We're on Anthropic and Claude. Uh still like it a lot for the business component. So a lot of B2B niche B2B claude is still very prominent. I think chat GBT is definitely not going away. It's continuing to to grow and grow gain mind share. So I think it'll be a triopoly of sorts. That doesn't mean there won't be some niche ones. I haven't really seen that emerge in our clients kind of referrals for what it's worth.

Uh Beyond Perplexity was there year one and now it's kind of not showing up anymore. So Google I think will maintain the market lead. That's my instinct, but there will still be a lot of chat GBT and cloud and if you if you're a developer tool, man, you could have outsized influence on AI. Uh, and then also what we see is if you're AI tool, if you're specifically AI software, by nature, people are doing AI queries on AI more often.

Therefore, those could actually be the majority share for you in terms of pipeline that they send. >> Interesting. I am gonna say controversial thing but I think that it's gonna move towards one winner. It's gonna be someone is going to emerge as the Google of AI basically. You know I don't know why but it seems to me that the the more users use a model the better it gets. The better it gets the more users use it and the more people attract to it.

So that's why there's this kind of AI arms race right now because people like I think these companies also understand that um there's so many different models but I think it's going to be a winner take most type situation almost like in a Google right nobody's been using Bing it's only recently kind of did a little of a comeback [laughter] but I think it's going to be the same thing I'm I don't know if it's going to be Chad GBT or Google or you know Anthropic but if I had to S this is my prediction for future for future me in in 20 years or whatever I think is going to be a winner take all or winner or take most let's say let's put this >> yeah it's interesting we did I mentioned the three top models literally today I got super excited about Grock I have a model Y and you can basically can talk to Grock I can sync it Gmail calendar drive and now I can be working in my car full self-driving >> oh my goodness >> and where my mind goes is whoever has the distribution so I I kind didn't think about Grock as like a potential, but now I'm like clearly Elon knows what he's doing from a distribution standpoint and they have some distribution.

Obviously Google has distribution. ChachiT seems the most challenged from like native distribution which maybe they're building hardware products to have that change. >> But it's interesting. Yeah. Who who knows? Maybe it'll be Grock of the four uh if we're all in space and he's owning SpaceX. So, uh, >> Elon is not allowing any other models to be in the in the Tesla. It's just it's going to be Grock and Grock only. >> Yeah.

Or the Cyber Cab. I mean, those are here in Austin and blowing people away. Uh, for sure. >> All right. Um, I want to ask you a bit of a selfish question for myself before we move on to talking about your book. So if you were advising my channel, my YouTube channel, what would you want me to test and what would you measure before asking me to make more of it? So um if I uh to be yeah to be basically more useful to both humans and AI models, right?

So if I want to be cited more um in LLMs, what would you be advising? >> Okay, so specifically we're talking about your SAS reviews channel just for the listener. >> Yeah. Is that correct? >> Yeah. >> How you get cited more in LMS? Yeah. An interesting thing I transparently, George, on this call, we could have this conversation. I was like, you SAS is still pretty broad, right? So, you're hitting a lot of great SAS tools.

It's upper market and generally apply applicable to most people, but I was like, could George or someone further stretch their influence by maybe going maybe niching down to sales or something, right? like spit picking a niche within these B2B products may help distribution. I don't I don't know that with certainty, but I'm guessing you've considered that. >> Yeah. Becoming more an authority in a specific niche. >> Yeah. >> And I've been struggling with that for my whole life, [laughter] >> especially doing YouTube.

It's hard because it's like for me, I don't want to uh box myself into one thing. Although rationally I understand that by focusing and having I guess I just haven't found yet my the one thing I guess and that's it's just like >> uh explorative process still. >> Yeah. Yeah. You're you're young. You'll figure it out. But one other thing I've been thinking about recently, I'm not sure if you're doing this, >> but when we ask a YouTube creator to create these videos, obviously there such great video content, >> a byproduct I see in LMS is a good amount of Instagram reels that are just snippets of that video distribution.

And obviously it has like the right description text which gets pulled into LMS, but thinking about clipping off channel on the other UGC platforms could be like a package deal that people find attractive. >> Okay. >> Um Yeah. If you're not doing that already, >> right? So, so hitting more organic uh distribution channels basically clipping that. >> Yes. Yeah. Specifically Instagram. I've seen it a decent amount surprisingly.

Um Okay. >> But I think it could I I don't know exactly how you would distribute on LinkedIn necessarily. Um but we'll I'm a little more bullish on LinkedIn recently since they cut down on AI comments. Have a setting now you can turn off non-connection comments. So that's we'll see how it goes. >> Interesting. Yeah. I don't know. I have I struggle because there's way too many platforms and I don't like any of them. Like to be honest, it it sounds so funny, right?

But I don't like any of them except for YouTube. I I was a kid. I grew up watching YouTube and it's just the platform. I don't even like I don't even like the short. I don't even like creating shorts. Think about that. Like people talk about get into social media. And I know I'm being foolish by saying that. But this game is so hard in itself and just cracking one format and one platform is already such a challenge in itself that then you know thing obviously you can put together a system where you have you know clippers and and and a a team that is going to distribute that for you and and all this and maybe I should do that.

I will think about that. Seriously, that's it's [laughter] it's it's an important thing to do. >> Yeah, you made me think uh yeah I I agree with you. I mean, I haven't done it with our YouTube interviews for what it's worth. It it can be challenging, but uh I could see with these bestex art videos you do. Obviously, you talk about six tools or whatever number of tools. Each clip even on YouTube could be isolated into a review of just that tool.

I know some of your clients will pay you just for a dedicated review, but even like a short snippet of like CRM software, best CRM software, you could do Pipe Drive, Salesforce, and just take that one minute that you talked about those tools or two minutes. That's one possibility for for a little bit more reach. I don't know. >> Shout out Pipe Drive. They're an Estonian company. I moved here now. So, I'm going to go visit their office.

We're hosting. >> Yeah, that's our CRM. That's >> nice. Yeah, I'm I'm uh I'm going to be hosting a webinar um in October with them together. So, that's going to be fun. >> Nice. >> Talking about AI visibility as well, how to win it. So, it's [laughter] good. It's a timely conversation. >> Nice. They show up. They're doing well. Yeah. >> Yeah. Please, Ross, tell me about your book. This is exciting, man. You've been writing this book and and tell I remember we talked about when we met at the Peak AI afterparty, you were like, "Yeah, I've been putting in hours into this every morning." You said something like three hours or something.

You've been writing it. So, this is exciting. Please tell us more about this. >> Yeah, thanks. Uh, coming out October 12th, uh, got a copy here. You probably can see it behind me as well, but, uh, yes, basically sprinted, um, really fast, uh, late last year when someone gave me the idea in my young president's organization group, business forum, super helpful. They were like, "What's a masogi challenge?" And the masogi challenge is, "What is a super hard thing you set out to do?" And I was like, "Oh, it's a perfect timing to write a book on on AI search." Obviously, it's challenging to put something out that doesn't feel dated.

And I I'm happy to report that I don't think that will be the case for this book because I specifically try to tailor it to strategies at a high level. And I think the universal things like get third-party reviews, get your transactional content right, that's going to be pretty evergreen in my in my opinion. Of course, some people may disagree with me, but uh that piece is something I'm pretty pretty stoked about. So, uh yeah, it's on Amazon.

We have if you go to siegemedia.com book with a dash in the middle, it's also in the topnav. Uh you can get it there, but would would really appreciate if people check it out. >> Love it. Well, you definitely have a very SEO optimized title. Should we I should say that it's [laughter] like the def the definitive guy for AI search. You know, you can't get wrong with that. >> That trust me, it was in my thinking of like I'm not attached to go as a term.

Maybe I curious your thought process. I was just like it has a search volume lead. I'm going to go where the data goes. I actually kind of prefer answer engine optimization honestly, but the data is telling me go and then of course, yeah, I was like, I'm not going to repeat AO in the in the subtitle, but AI search may clarify people don't know what the heck this thing is. And there's definitely search volume for AI SEO too.

So seemed like a smart thing to hit. >> Exact same thinking when I was uh naming when I was picking up the name for the podcast. I went I looked at the the demand of the data and go generated it was like much more than AO. So I'm like all right I'm not going to be a dummy right I'm just going to go for the thing that people are actually searching for. And then I was like hold up a second but what if someone is looking also for this for AI search?

And that's why I named it also boke. So we areite quite close. Oh, nice lock set. There you go. >> Yeah. Awesome. Well, um, yeah, check out Ross's book for sure. Um, I'm going to order it and I'm going to read it. I'm curious to to see what's inside. Thank you very much for your time, Ross. I appreciate it and I hope you guys enjoy this conversation. You're going to find all the links to everything down in the description in the show notes of this episode.

Thank you. Ciao. >> Thank you, George.