How Brands Build Trust & Visibility in AI Search

0:00The Voices of Search podcast is a proud member of the I Hear Everything podcast network. Looking to launch or scale your podcast? I Hear Everything delivers podcast production, growth, and monetization solutions that transform your words into profit. Ready to give your brand a voice? Then visit IHearEverything.com. Welcome to the Voices of Search podcast and I Hear Everything production. In this podcast, we'll share the news, knowledge, and strategies you need to navigate the ever-changing world of SEO. Ready to expedite your company's organic growth efforts? Sit back, relax, and get ready for your daily dose of search engine optimization wisdom.

0:41Here's today's host of the Voices of Search podcast, Tyson Stockton. My name's Tyson, and joining me today is Alex German, co-founder and CEO at Bluefish AI. Welcome to the podcast, Alex. Thanks. Thank you for having me. Good to be here. Now, when we're chatting beforehand too, like we covered obviously like a wide range of things. And I think like you're in a unique and interesting position to be able to like shed some additional light into like where some of these like inaccuracies are. And like, what are kind of like some of the more like tactical elements that brands should be focusing on doing.

1:20a lot of times you're hearing about just kind of like this is the state of the world but it's like left us with a bit of like well where do we go from there next and before we get into kind of like those inaccuracies and that other piece we'd love to just kind of like hear your take on like how we are fitting into this new kind of like era in search what stage do you kind of feel like we're at as we get into kind of like what to do with it? Yeah, I think that's a really good starting question because as marketers, we're pretty accustomed to new channels on the internet. There's sort of a new channel that emerges every five or six years, whether it's search, whether it's social, whether it's mobile, now AI.

2:08And there is a bit of a pattern in how those new channels deploy. So, you know, across all the channels I mentioned, there are sort of these, these, you know, these kind of stages, these discrete stages where, you know, a new technology emerges, changes how consumers interact with information on the internet. Marketers initially probably don't quite know what to make of this new channel. Like you'll remember when social came out, when mobile came out, marketers like, absolutely not, this is not for us. But over time, it becomes clear that that's where the eyeballs are going. And so marketers start to take it seriously as a channel.

2:44It reaches kind of critical mass. And at that point, marketers start to test and they start to explore. They recognize that this is where the eyeballs have gone. And so they need to, you know, they need to test and learn into the new category. Then you get into kind of a Wild West period where there's sort of a, you know, a gap between like hype cycle and reality, lots of opinions, lots of acronyms. We're kind of in that zone with AI. But over time, you know, marketers are really good at sort of, you know, making up their own mind about things and doing their own testing and learning and looking at their own data sets.

3:18Marketers tend to assume that, you know, everyone is lying to them. And so they make up their own minds, which is actually a really healthy dynamic. And so as marketers start to test, they start to figure out what works for them. What are the best practices? What is the blueprint for managing the channel? And then you end up with a more mature kind of marketing channel with best practices, with standards, with an established value chain of different tool providers, and with kind of a formalized structure within the marketing org where there's a clear budget, there's a clear mandate, there's a clear relationship with other marketing teams.

3:53And so for AI, we're still kind of in that Wild West period. I think, you know, what we've seen with with agentic marketing is that, you know, search, of course, those teams were kind of the canary in the coal mines. They were they were the teams that that discovered, hey, search behavior is changing. AI is shifting how customers are searching for brands. They're shifting discovery. They're shifting transaction like the whole customer journey is starting to evolve. We got to pay attention to this as a brand, as a marketing team. What's happened since then, though, is that it's become clear that the rest of the marketing org kind of has a role to play, too.

4:31So we know that the models learn from all of these different touch points. They learn from content. They learn from, you know, data. They learn from social. They learn from commerce. And so now we're in this sort of this chapter of this categories development where we're trying to figure out who plays what position on the field. if you know search team has one role okay what is the content team's role and what is the PR teams what is the commerce teams paid media and and I think you know we should expect our industry to kind of digest that over the next you know six to twelve months and I suspect we'll kind of graduate to the next stage of maturity sometime in 2027.

5:09Which on on that point though because I obviously there's a lot of similarities between the cross-functional dependencies of traditional SEO as well as what we're seeing the needs today. From your perspective, and we've seen a variety of approaches to this when you're working with different businesses or clients, do you see the direction that AI search is going, that it is a channel, a discipline to be managed within, or is it more of the spread out across the organization, each person's playing a different role in a component of it? So it's like how much of it is diffused versus concentrated into like one resource group?

6:00Yeah, that's a, that's another good question, I think, because, you know, look, We think of both things as being true a little bit in that certainly, you know, AI is kind of a channel that's eating all the other channels, right? So it's not like it's kind of siloed. There are multiple AI surfaces. So, of course, there's kind of the AI native platforms like ChatGPT. There's also generative search engines like, you know, Google Gemini. There's Enterprise, Microsoft Copilot. There's Social AI, Meta AI. there's the retailer agents like you know amazon rufus walmart sparky and so there's been this proliferation across different media surfaces so this sort of like distinction for the consumer of like ai internet versus non-ai internet those lines are blurring certainly um and and i think you know increasingly consumers don't really care about like is this an ai experience or not ai experience their focus is look you know ai is here to help me have sort of a better faster smarter or online experience.

7:03And so I think the lines are blurring in terms of the surfaces. The lines are blurring in terms of the functional practice. So as you said, there's kind of this, you know, this kind of team sport dynamic where you need all of these different functional teams from content to commerce to paid media to kind of play a role. No one has a marketing team, you know, and even if they say they do, that team, I promise you, is less than 12 months old. So, you know, everyone is kind of looking at the teams they do have and kind of up leveling those teams and evolving those teams. And I mean, and that has happened organically, right?

7:41Like if you are a content marketer and you are building first party content that goes onto a brand dot com and you don't think that that content is going to be consumed by a model and make its way into an AI response. Like, you know, you got some problems there. So I think all of those teams have started to evolve. I think where we do think of it as distinct, though, to be clear, is that ultimately they're solving a new problem. All of those teams are solving for a pretty fundamentally different stakeholder than they were solving for in the previous iterations of their roles. I think, you know, one of the really interesting examples that we see today is on commerce teams, where, you know, up until now, commerce teams have been sort of, they've been playing, you know, largely a, you know, they've been serving sort of two stakeholders, they've been serving e commerce search engines, and they've been solving for human shoppers.

8:37And so, you know, if you think about like a product detail page, a PDP is like one of the primary tools in the kit there. A PDP has like 300 words on it, maybe 400, 500 words if you're lucky. A model probably needs, you know, 10 to 100x that to actually train on the product, on its applications, on its benefits, how it fits different customers differently, etc, etc. Plus, you need all of the customer reviews, you need, you know, you need a really rich data set in order to solve for that. And so what we've seen over the last six months as retailers have rolled out all of their shopping agents is a lot of commerce teams are saying, OK, like this is a fundamentally different game that we need to play.

9:19How do we play it? How do we measure it? How do we optimize it? How does this fit in? And so I would say there is sort of a, you know, long-winded answer to your question, but I think that ultimately it is sort of a holistic kind of thing that needs to be solved for. But within each of those teams, there are fundamentally different things that are happening. Yeah. And for me, it's like, again, like similar elements that we've been operating within, but it's just like amplified or on steroids because it's like you do need that like holistic, like this is like almost from like the executive perspective of like this is where our brand is going and how we're strategizing to compete within this world.

10:04But at the same time, you need those specialized focuses and those individual players, which probably fit better in those other teams. And I feel like a lot of times when you come across the organization, the businesses where someone has tried to take that walled garden approach of isolating SEO and I'm going to have my own development team, I'm going to have my own content team, and I'm going to like get budget and then just insource it all to within it's like those ones tend to always be a little short-lived um which it in some ways it's like great you have the control but at the same time like the effectiveness i'm always like you're always then trying to like i don't know work go a different direction than potentially the rest of the org i i yeah i think that's i think that's true And it is sort of the, you know, the brick wall that a lot of, you know, early AEO pilot programs are running into today where there was kind of this like top down mandate.

11:09Like, hey, AI search is a thing. You know, let's let's let's spin up kind of a synthetic V team to go figure it out. It's usually search led. There's kind of a limited mandate, but there's a lot of urgency. That team goes and does their job and they sort of map out like, hey, here's how we're doing. Here are the areas of opportunity. Here's how we need to start to manage this channel. These are the five other marketing teams that need to play a role. And then it stops there because those other marketing teams are like, who are you? You know, like we have our own marketing calendar and objectives.

11:40And and so I think the you know, the really. I think maybe unsexy, but really critically important part of this chapter in the industry is, you know, maturity is change management and figuring out like, OK, like how do we build connective tissue between these teams? What is the process? What is the mandate? Like, you know, a good example that we see today is, you know, everyone is racing to build first party content. Like, OK, the models need more content. We've got some gaps that we need to close. Cool. Let's update some first party content. If you're an SMB, straightforward. If you're even a middle market company, kind of straightforward.

12:23If you are like, you know, a pharmaceutical giant or a CPG giant or a large financial services company with like a 10 ,000 person, you know, marketing team and, you know, a thousand of those people are compliance people. Like, I'm sorry, but putting a new page up on your website and then changing it every 30 days, that is an incredibly tall order. And I know that, you know, Silicon Valley is a little bit like drunk on the Kool-Aid right now. Like, just put an agent in front of it. The agent will fix it. You know, the agent's not going to fix your compliance team. Like, maybe someday I look forward to it.

12:59But, you know, in big pharma, like, absolutely not. And so there are these like these enterprise realities that these teams are contending with. And so I would say right now, you know, at Bluefish, we only work with enterprise brands. We have a unique vantage point. I think the teams that are winning are the teams that are figuring out how to work together within their marketing work, within sort of a matrix marketing organization. If we're going to go create some content on our website, what are the five teams that need to be involved? And how do we streamline all of the processes that connect those teams?

13:32That's how you win at AI search right now, because there are a lot of teams that are really good at figuring out the problem, but really bad at mapping the solution. And the ones that win are the ones that can move from insight to action and execution to measurement over and over and over again at high velocity. Like that's what winning looks like these days. totally agree with that and it's like knowing what the problem is just seems to be more of like the table stakes and it's like being able to create that alignment having like a shared vision roadmap and then the execution front seems to be where they're like the greatest risk of falling short tends to be now with the piece on kind of like enterprise like i know you're in kind of a unique position where you have, you know, broad set of clients, great data vantage point, especially within these enterprises to see kind of like where some of the inaccuracies.

14:36And I feel like that to me is also feeling like one of those pillars or points that can help with the stakeholder alignment, where typically if a C-suite is seeing misinformation from AI search, it's going to tend to be heard to like across the organization, which gives you some headwind in getting that alignment. But like, from what you're seeing, like, what are some of those kind of like areas where should people be concerned with on accuracies? Yeah, first of all, people should be concerned about inaccuracy. 100%. I mean, you know, I think AI developed this very, undeserved reputation as being factually accurate because it was somehow trained on the internet and the internet is just this bastion of truth.

15:29So I think step one is just understand that there is a problem and a lot of companies, I think, are familiar with that in that they've done some ad hoc kind of research and they're like, oh, that response looks a little bit off. Doing it in a systematic way. Like at Bluefish, we have an accuracy product that is 24-7, always on tracking in great detail and depth. So doing it in like a systematic way so you actually understand and can quantify the extent and nature of the problem is sort of step one. Step two is figuring out, okay, what is the root cause? So actually doing the diagnostic to exercise to figure out, you know, not just are we being portrayed accurately or not, but why are we being portrayed inaccurately?

16:22How are the models learning? How is that sort of factoring into the way that we're being portrayed? And I think like on that front, the answer is more complex than it might seem. The models are, I would say, changing very quickly. They are relying more or less on various sources from week to week. Like we saw the Reddit drop off, for example, two weeks ago. We're seeing listicles drop off. Obviously, like brand verified content is on its way up as brands start to create higher quality content. You're also seeing brands start to change the nature of their first party content. So for example, if you're an electronics company, If you make a smartphone and you release a new model of your smartphone every year, which most of them do, the models really struggle to differentiate between this year's model and last year's.

17:20And so, you know, if a consumer is asking about the like battery life on the latest smartphone, does that refer to this smartphone, last year's model, the model before it? Same thing with we see problems with drug dosages, right? So the dosage of a particular drug really depends on any number of circumstances. So the dosage for one patient with certain preconditions might be very different than the dosage for another patient. So when someone asks ChatGPT, what is the dosage for XYZ drug? You could get any number of responses. And so the brands are realizing that they're starting to structure their data in different ways in order to help manage that.

18:02So overall, it's a big problem. We see about 20 % of responses today are inaccurate in one way or another. Those numbers tend to be higher in certain industries. So like pharma is more like 25%, 26%. Financial services is higher. CBG tends to be higher. So like ingredients are really tough. So like if you go to like, if you go to an AI agent and you say, hey, give me all the ingredients in, you know, a particular brand of dog food. or whatever, you might get like 50 ingredients and 49 of them are correct. But like the 50th is just like totally made up and totally synthetic and like doesn't actually exist.

18:46So yeah, it's a problem. And, you know, I think now to your point, I think the alarm bells have kind of gone off and agentic marketing teams, AEO teams are starting to use accuracy as kind of a rallying cry within the organization. Like, hey, folks, like this is a huge problem. We're taking these actions not just because we want to show up in chat GPT, but because there are major brand safety issues here that we need to manage. And it has been a helpful kind of, you know, change agent within these orgs. That's super fascinating. I mean, one on like the inaccuracies is like, yeah, the most critical core industries like health and finance are the ones that you're seeing with some of the biggest concerns on government actually that one would be pretty interesting too because especially knowing how like dated some of those sites are and like the amount of information too that's just living out there but also yeah i mean it's also like super uh you know super political and so you know like if you if you ask chat gbt like hey should i fly to like Qatar right now?

19:58Like, you know, is it safe to fly to Dubai? You know, like there's anything could happen. Like what does the US State Department have to say? Like those sources, sometimes more volatile than, you know, angry guy on Reddit. That's interesting. It's also like the way you're describing it to me, it's like, it's very reminiscent to like the SEO industry over the last 10, 15 years or so, but it just feels like the evolution period is like compact down to it's like what before might have been like a year two years of evolution is like happening every other month kind of thing and also and i remember this like i'm listening to study like from probably like three years ago or something it was like one of um within kind of google's patents and it was basically looking at like some of the strengths of semantic search versus LLM search and or LLM driven search.

20:55And it's like, yeah, you have like the huge advantages in like language and context and nuance. But then a lot of the things that you're kind of like mentioning too is like the exact advantages from like more of a knowledge graph semantic search area where it's like the accuracy is a little more stable, a little more consistent there. And I feel like it's been flagged is something that's of a concern. And maybe just as consumers, we're like, we don't want to believe, we want it to be accurate. So we don't want to accept that there's a 20 % question mark in there. What's been interesting is to sort of track the gradual alignment of incentives between the sort of, you know, frontier models, the open AIs of the world, commercial brands and consumers.

21:51And what I mean by that is, is that, you know, when, when Chachabuti sort of, you know, started to scale in terms of its popularity at that point, like the models were trained on, you know, essentially a download of the internet and the internet was a messy place, but it was, you know, that was largely seen as fine. The use cases and applications for the ChatGPTs of the world were pretty limited in the early days. And, you know, that was fine. It was mostly a magic trick. As it kind of started to become a little bit more, you know, mature and commercial use cases and applications started to become more important.

22:29We, you know, when ChatGPT launched shopping, consumers, of course, are like, you know, mostly using the internet for commercial purposes at this point, so it makes sense that that happened really, really quickly. And AI is just a better, faster, smarter way to shop online. It became clear to the open AIs of the world, maybe we actually need higher quality data sources to inform this sort of commercial behavior. How do we get that? And so there was this alignment between, you know, open AI, you know, the Samsungs of the world and consumers that all wanted an AI response that was going to give you an accurate, high-quality experience when you were shopping for an HDTV.

23:10And I think that has sort of accelerated, to your point, a lot of the underlying infrastructure that you need to make that happen. Like you saw a lot of the models announce new sort of data protocols and data standards that essentially enabled brands to pass data into those models directly. And of course, they're using kind of the product feed rails that already existed, but you can already see how those rails are evolving. You can see all of those additional parameters that are being added in where it's like, you know, like add an FAQ. Like what is an FAQ doing in a product feed? And so all of a sudden you're starting to get like these richer data transactions between brands and models.

23:55And I think that kind of shines a light on where the industry is going, where a lot of the ways that the early models were trained, it was trained off of what was there, which was like the brand.com. But the brand.com was mostly built for like human consumption. It just wasn't really designed to be a data delivery mechanism for a large language model. And so it makes sense that the brands and the models are trying to find new pathways. Like let's, let's not try to like overload this kind of legacy vehicle. It's probably not going to be the best way for us to get information from the brand. So let's find sort of a new, a new pathway.

24:32And to your point, the same thing happened in the early days of search where it was sort of, it was messy in the beginning and folks were like, all right, like, how do we, how do we organize this and what, what are we actually solving for here? And let's, let's optimize for that. Which, and I think the, the product feed piece is such a good call out, man. It's also like, it's one of those like painfully simple answers too. If you're from the, like, you know, the lm provider side where it's like yeah the infrastructure the cost like the investment to stay accurate on that is so drastic and if it's just like yeah no you just send it to us in this format that's easy for us to work with and it's like it's a win-win and i feel like there's yeah probably some like long-term tenure people at google that are just wishing that it was that easy back in the early 2000s of just being like, wait, you can just open the door and let them send it to you in a structured format.

25:35Yeah. I mean, this is why I think like the, the kind of existential crisis that we've seen play out on the Google SERP has been a little bit annoying for the category because, you know, at the end of the day, like what, what Google says matters. And especially folks in our industry really pay attention to that as sort of, you know, you guys have 80 % market share. Like, tell us what, you know, what is the protocol? And I think, like, you know, Google was kind of dragged kicking and screaming into the sort of AI space race, largely by ChatGPT and others. And you see that, like, there was a little bit of this innovator's dilemma on the SERP where you had like, you know, AI mode and AI overviews and 10 blue links and kind of all of this stuff kind of mismatched.

26:25And, and of course, you know, public investors were sort of looking for like, where's the ad format and trying to solve this like math problem of like, you used to monetize your SERP in 10 ways. If ultimately the consumer is just spending time in this one little box, do you have to sort of do with one what you previously did with 10? Like, how are you going to solve that? And but so I have empathy for that. Like that is a that is a challenge for any organization, large or small. But you have like, you know, really important folks on this on the search teams at Google saying like nothing has changed to like change nothing.

27:03Like keep keep doing things the way you've been doing them. And yeah, that is that that sort of creates a little bit more friction than I think is super helpful at this point. And it's like Google from their perspective too, it's like, and I think the way you phrased it was being drug into the race. It does feel like more of like, isn't it like Google wouldn't want to compete on it? It's just more of like, oh, they would have preferred the race to start like two years later. Kind of. So it was more of like someone just fired the starting gun a little early. And it's like, like, we actually have to run.

27:37Like that's how it feels working in AI these days though. That is kind of the pace of this thing. But you're right. The deployment cycles of every other channel, if you look at, there's this great, it might have been a Mary Meeker kind of chart, but it basically traced out sort of speed to a billion users across previous channels. So it was like search, mobile, video, programmatic, AI. And if you look at the deployment periods for all of those previous channels to get to a billion users, it was like three years, four years, five years, six years. And then AI, it was 18 months. Time for a one minute break to hear from our sponsor, Previsible.

28:21So you're looking for SEO help and you got a couple of options. You could start replying to spam from agencies that claim they can get you to rank number one on Google. You can pay an hourly rate for a consultant who will inevitably nickel and dime you with hourly charges. or you can work with a cookie cutter agency to quickly launch a strategy-less project with low success rate. None of those sound very good now, do they? Well, that's where Previsible's integrated consulting model comes in. Previsible draws from a collective 40 years of SEO and digital marketing experience to unlock your organic growth opportunities.

28:55They build custom solutions that combine strategy, technical expertise, content, and reporting to effectively operationalize SEO for your business. Prefizzable's four-stage approach ensures that your SEO programs thrive by starting off with a strategy-first approach. Then they support you in your efforts to create quality content, help you identify technical issues, and most importantly, they'll work with your cross-functional teams to integrate your SEO strategies to make sure that your SEO budget actually drives results, not just your agency's bottom line. So join brands like Yelp, eBay, canva atlassian square all who rely on the seo consultants at previsible for more information go to previsible.io that's previsible p-r-e-v-i-s-i-b-l-e.io

29:45and so to your point yeah i think this the the starting gun went off and everyone sort of scrambled to figure out like okay like now what and that's true of marketers right like i you know I say this as a guy who runs an AEO company, but we saw the Fortune 500 really quickly mobilize for this channel in a way that we've never seen before. At my last company, we sold into the Fortune 500 and we had maybe a 14-month sales cycle with large enterprise brands, which was seen at that point as a pretty normal cycle to land a new customer. At Bluefish, it's like 90 days. And so we've seen these really large brands scramble to move really, really quickly to get into this space, not because they want to or because that was a natural motion, but because that's what's required to sort of compete on the agentic internet.

30:44You have to move fast. Velocity is such an important aspect of it. going back to where we started the conversation though yes you have to move quickly to get started but then once you once you've started you have to scale the damn thing and that's that is a different sort of challenge yeah and it's like you're balancing that like early pace to like have a seat at the table but then it's like depending on how well you set the framework up is going to determine that scalability in some situations. I mean, most of our customers have anywhere from, you know, 30 to, you know, 100 different business units within their holding company.

31:25And so, and each of those business units has their own geographic market footprint. They have their own marketing teams and functions and customer segments that they target and channels that make sense for them. And so, you know, doing AEO as an SMB or even as a middle market firm, billion dollar company, you still have a relatively small, agile marketing team. You can do stuff quickly. As you get into the enterprise, it is challenging. And so part of why we exist is that we are there to be the sharp end of the stick and really help those teams move faster and create that kind of orchestration layer around helping them manage these programs and scale them out and kind of streamline operations.

32:09Because going back to what you said at the beginning, this isn't really a siloed channel anymore. It is a holistic kind of reevaluation of what is your marketing stack? Like, how do you as a brand, you know, reach your customers on the agentic intramet and what needs to be true to do that? yeah and then it is a big challenge too especially on the enterprise level and it's like it's one element that's kind of exciting about the industry right now is because it's like as you said name of the games move quickly you're a massive organization you're at a serious disadvantage like maybe you have the funds and the resources to fund it but sometimes just getting that alignment becomes such an overhead in the overall objective where if it's a small nimble company they need to convince three people to start acting like yeah totally also it's like production like barriers have dropped too so it's like it is kind of interesting that you have these like inside lanes that you know be interesting if people can take further advantage of it but i think you're right like weird incentive there are these these kind of weird incentive structure things that are happening right now like like for example like you just mentioned the cost of production has gone down.

33:28Totally. Like at this point, you know, I could, I could go to, you know, any sort of AI platform on this call and say, Hey, go make me a website for brand X, Y, and Z, optimize it for AEO, create, you know, content about X, Y, and Z. And like 30 seconds later, like that exists. But I think a lot of those teams are also realizing like garbage in, garbage out. Like if the cost of production and actually creating the content is dropping to zero, then the signal that informs it becomes that much more important. And so there are new things to be solved for here. I also think that the models are changing very quickly in interesting ways.

34:08Like 12 months ago, you could create some AI sop on your website, and that probably wasn't going to hurt you too much. Maybe it hurts the consumer experience with humans, but the models would sort of hoover it up. You saw all of these nonsense tables and FAQs on websites that were clearly like made for AI. Now I think that has changed and you're seeing more of a flight to quality. And there's this increasing recognition that, you know, do you need to create more content and data for models? Yes. But if that quality is poor, then the models will sort of move past it. And it won't have the sort of influence that you were hoping it might.

34:46And so that's part of the like, you know, healthy, you know, growing pains. I think it's fundamentally a positive trajectory but it is changing how these teams actually do the thing well i think we i want to get to the like the lightning round questions but i feel like we have one more area that i'd hate if i forgot or like didn't get to asking and and i know a lot of the listeners are probably going like okay great i get the importance the magnitude of like inaccuracies maybe starting to like click as far as that also being you know the the tip of the spear to get the alignment with the inaccuracy so what advice would you have as far as like great we found them we took like you said a systematic approach to it so we know like where it's happening what it is how do you then take that into action to then make change to it.

35:49Yeah. So once you've tracked accuracy, you've done your sort of diagnostic to figure out the underlying cause, probably the biggest surprise that brands have these days is how often they are actually the culprit and that ultimately changing first-party content and data can go much further than you would think. Like, obviously, if there is an offending third party blog or some some, you know, third party, you know, journal or community forum that is, you know, slandering the brand or, you know, stating inaccuracies like there are tools in the kit to report that to take action. There's there's, you know, paid editorial and PR teams are very good at that.

36:36But I would say the the you know, the interesting thing that we're seeing is is how much you can do with first party content. in terms of, you know, A, closing gaps. So, for example, you can see, hey, this inaccuracy was actually sourced from content on our website. Okay, well, let's go look at that URL and see what is the offending piece of content that's misinforming the model. That solves a lot of your problems. I would say, too, there is, you know, increasing understanding that a lot of times the models, when they are inaccurate or they hallucinate, it's usually because there is an absence of information about that particular topic.

37:19And so the model has sort of come up short in terms of high-quality training data to support its response. And it's sort of, you know, grabbed it at whatever else it can in order to close that gap. So sometimes it's just going from zero to one in a particular topic area. I think the most important thing, though, once you've done that, and I guess for the sake of clarity, influencing models is a relatively straightforward exercise. I think at this point in the AEO sort of category, you know, you can sort of change the inputs and that will change the outputs. And I think there's pretty broad recognition of that.

37:57The hard part is how do you do that in a sustainable way? How do you do that in a scalable way? How do you do that in a way that is actually, you know, sustained, you know, with your with your your key customer segments across models, et cetera, et cetera. And so to do that, I think the piece that a lot of people don't spend enough time with is measurement and making sure that you actually have clear trackers to measure. OK, we made a change. Did it work? How do we actually know? And so tracking kind of post-optimization accuracy improvement is something that everyone should be doing. It should be table stakes for any agentic marketing team.

38:38And we see very little teams doing that at the level that they probably should at this point. and especially like coming back around full circle to like the beginning of the conversation it's like when you're trying to align so many different parties in an enterprise having those like sources of truth and north stars to point to not just to like create the alignment but to keep the alignment is so critical and it's like if you don't have that measurement piece being able to keep the focus of a large organization becomes very, very difficult. And it's like, you need to be able to have that to just continue to keep people rowing in the same direction.

39:24Otherwise, you might achieve it for a month and then it's just going to go back to wild west. Yeah, I totally agree with that. I think there was a rush to report out in sort of early days with these AEO pilot programs. And there was a lot of senior executive visibility, like, okay, what have we done? Did it work? But that needs to be sustained. There needs to be sort of a QBR style check-ins to make sure that things are moving in the right direction and that executive attention is held. I totally agree with that. All right, so coming up now, I want to hit on kind of some quick form, short form questions.

40:11So we are lighting round. And the first question I want to throw at you is, what is the most common misinformation in AI search? And what are some of the more common contributing sources to that information? The biggest thing that we see is just a little bit of like reductive thinking about what AI search is. So for example, like we hear a lot, you know, CMO just wants to make sure we're showing up in chat GPT. Visibility is obviously critical. You need to be in the response. But the nature, like the way that you actually show up in that response matters just as much. So what is your favorability?

40:52What is your accuracy? Is the brand being represented in the way that is actually, true to what the brand cares about, what the product is about, et cetera, et cetera. And so I would say like the biggest, the biggest, you know, challenge that, that I think, or mistake that we see made is just treating the channel in an, in an overly simplistic way. It is, it is pretty nuanced. And ultimately if you're a brand and, you know, an AI model is going to be representing your products and services to a consumer, it's not, you know, a list of links. You're not it's not just, you know, where are you at the top or at the bottom of the list?

41:30What's actually being presented to the consumer is this long form qualitative description of your brand, of your category, of the merits of your products, the strengths and weaknesses. And so there's a lot of like qualitative detail that is actually really important. And of course, you want to manage your visibility. But for all of these other dimensions that we think are just as important. And, you know, we spend a lot of time on sort of education on that front, like helping brands move beyond just their rank to, okay, what is the actual quality of your representation? And one thing that like jumped out to is something you said just a little bit earlier in the conversation around like a lot of times your own content being one of those contributors to kind of harming the quality or the integrity of the information.

42:21and if i was an in-house like if i was on the outside i think a lot of people should be really excited and happy about that because that should be almost lower hanging fruit from a how do you drive impact and like how what do you do with this next it's like you are now in the driver's seat and it's like you have that opportunity to make make an impact without yeah being as reliant on externals. 100%. Next one I have for you is what's the biggest mistake brands make when trying to correct inaccurate information in the AI search? I think exclusively focusing on like third-party sources. I think, you know, very often there's a lot you can do with first-party content, first-party data.

43:11There's just a lot of first-party tools to close that gap. We're seeing direct data feeds to models start to play a major role on the accuracy front where the models are increasingly looking for high quality brand verified information that they are much more likely to treat as sort of the gospel truth versus angry guy on Reddit. So I think there's a lot that brands can do to take control of their AI accuracy once they have the right diagnostic. so essentially is saying clean your own house first and then worry about what else is going on outside of it yeah and i i look i think more more more broadly if you if you as a brand are investing in in really high quality first party content you're kind of on the right side of history like i wouldn't i wouldn't try to like compete with all of the garbage on the internet There's an unlimited amount of low quality content.

44:15Ultimately, what you want is you want the models to see your brand as a credible source of truth. And if you can close that gap, that goes a long way. And the reality is like a lot of first party brand content wasn't built for the agentic internet. It wasn't built for agents. It wasn't built for large language models. It was built for humans with a 20 second attention span and maybe search algorithms. And so, you know, I think there is a new stakeholder to be managed. And so there is this, you know, there is this sort of first party content mountain that a lot of brands are climbing today. And that's not to make it sound easy.

44:52It isn't. It is a lot of work and it is complex. But yeah, we are big proponents of, you know, brands as their own source of truth and using first party tooling to get there. well and we've also we've seen like we've seen already how fast like these models are shifting from like the dominant sources and there was like the reddit wave now we're in kind of this like youtube wave i feel like your own website and that firsthand information like that does not feel like something that is going to be an easy replaceable and so it's like if you're placing bets, it's like that just feels also like a much safer bet and more controlled, consistent.

45:40Like, yeah, it just makes a lot of sense in my eyes. Yeah. I, you know, look, I also think it's been sort of interesting. So because Bluefish really only works with kind of Fortune 500 brands, most of our clients, they spend, you know, billions in marketing on an annual basis each. These are the largest, most sophisticated marketers in the world. And I think in the early days of AI, everybody was a little bit reactive. They were a little bit on the back foot. There was some apprehension around what is this channel? This is, you know, this is unknown. This is, you know, uncertain. What we're seeing now, certainly with Bluefish customers, is they're playing offense.

46:23They're sort of remembering like, oh, like we spent$8 billion a year on marketing. Like this is just another channel. We have the best marketing teams in the world, super sophisticated search and content and paid media folks. This is ultimately a new way for us to engage with our customers. We just have to figure out what's the what's the right way to do it. And I think, you know, creating first party content and investing in first party content ultimately just helps build your brand equity. And and that's that's always that's always a good thing. So, yeah, I would say in general, you know, our guidance for our customers is always to sort of stay focused.

47:04And, you know, in these new categories, there's this like whack-a-mole phenomenon that you can really get pulled into with these new channels where, you know, like any idiot with a soapbox on LinkedIn can like proclaim themselves a thought leader. I would say for most of our clients, they kind of see past the noise and they recognize this channel for what it is. And so they're not going to erode their brand safety. They're going to invest in it. They're going to build it out. They want to make sure that if a model is going to represent them, it's going to be accurate. And, you know, and with that will come favorability.

47:39Absolutely. Next one I have is which information source do you think will become more important in AI search over the next year? So we talked a lot about first party content, but I think what is what is accelerating is brand data. And what I mean by that is like actual data that is made for large language models that is structured, is syndicated and is done so in a, you know, in a sophisticated, measured way. Right now, there are these new data pathways that exist between brands and models, but they are not really being used at scale yet. And we think that's going to increase quite a bit. And so, especially going into the holiday season, we think that brands are going to really invest in, okay, what is our agentic customer journey?

48:39What are the tools in the kit to reach our customers as they move through that? And I think you're going to see data start to be a much more powerful tool than it was in 2025 and for part of 2026. And do you see that going beyond the product feeds? I think in e-commerce, it's like, great, we have kind of like the framework almost for it with product feeds going into AdWords and everything else. But it's like, do you see that kind of spanning across other segments and industries? Yeah, I think that, look, certainly I think the highways are going to be expanded. So I think the definition of like what is a product feed is going to change.

49:25And I think you're already seeing that there's like new parameters that are being added into the product feeds. There are new ways that models and brands are using those rails to do stuff that hasn't classically been part of a product feed. Usually a product feed is skew, price, availability, some product specs. It's structured data, but it's pretty pared down. You're going to see that expand. There's going to be all sorts of additional sort of pieces of content that can be made available through those through those rails. And to your point, I think that there are there are likely to be new standards that exist between different kind of AI surfaces and the brands themselves.

50:05So, you know, open evidence, I think, is a really good, really good example. Right. like a verticalized AI application, you know, used for a specific purpose, they have a really clear incentive to go to all of the large, you know, drug manufacturers, for example, and make sure that, hey, we're getting the right data from you guys. We look at your websites. There are these like gaps in information. We need something that's more structured, something that's more exhaustive, more comprehensive. What's the right way for us to represent your products to, you know, Yeah, I think you're going to get into kind of the next level of the game where there will be new data standards and sort of data, you know, interchange formats.

50:51Interesting. Now, I guess for the listeners, a little peek behind the curtains, I kind of ruined the next question with that little follow up that I did unknowingly. And so the following question was going to be right around the product feeds. But I feel like you already kind of hit on that with with my little tag along. So I'm going to, I want to make a pivot on it. And so I guess I'm just caveating that being that it's, you know, just being thrown at you in real time. But from that, I guess like the question I would have is, do you see like the optimization of strictly structured data or I mean, product feeds being one facet, maybe within a larger topic of structured data?

51:34Do you think that that would start to emerge as like a specialization or almost like its own discipline and focus within our industry? Yeah. So usually when there's a new category, what happens in the beginning is that everybody kind of does everything. So in the early days of like GEO, AEO, there were, you know, everybody was sort of an AEO platform. everybody was basically trying to be everything to everyone. So, you know, we know that like a lot of our, you know, competitor companies, their products are designed for SMBs and they're designed for middle market and they're designed for enterprise.

52:18So it's sort of, you know, splitting the baby. What usually happens is that doesn't work. Like if you solve for everyone, you kind of solve for no one. And so there will be more specialization by customer segment. So like Bluefish, we're committed. We only work with a Fortune 500. But to your point, there's also going to be more kind of flora and fauna in the ecosystem. So you're going to see, you know, data companies. You're going to see analytics focused companies. You're going to see optimization focused companies, measurement focused companies. You're going to see, you know, pay media focused companies.

52:53there will be an ecosystem for better of course there will be another you know luma scape and and i and i think that what happens is is that these industries tend to deploy in three stages there's the you know everybody does everything kind of stage there's the specialization stage and then there's the like big tech consolidates everything stage and it sort of re reconverges usually that whole cycle takes 10 to 15 years to kind of play out maybe a little bit a little bit more. But right now, I think, you know, you're just starting to see real maturity start to emerge in this category. So I think I think we're still in the earliest innings of AEO and agentic marketing.

53:33It has been 24 months of this industry and there's only so much you can, you know, so much water you can run through the pipe in 24 months. But you are starting to see we see this at Bluefish. The largest enterprise organizations in the world are thinking, OK, we've got some points on the board. We figured out how to do this thing. Now we need to scale it. What does that mean in terms of our org? What does it mean in terms of our tech stack? What does it mean in terms of integrations? It's time for the next level of investment. And that, to your question, is going to bring out specialization. And I think we'll see a lot of new companies in 12 months that don't exist today.

54:12And that will ultimately be a healthy thing for the category. I mean, it's been a bit refreshing too. And I feel like, you know, coming from the SEO tool provider and kind of like software side, I think we went through some of those first phases and like, just kind of getting into, I mean, it's been in the consolidation for a little bit. but it kind of feels like now with this industry shift it's like reset the clock and now we're going back to the like specialization which i'm excited about because i think like that's kind of like the fun it's like one of the more fun periods to go through is because you get so much variety and so many interesting perspectives on a similar problem i think that's so true there i i really I think in the early days, there's so much hype cycle and so much noise that comes with the hype cycle.

55:04Like there is, I like can't go on LinkedIn anymore. Like it's just like, there's, there's just too much like really fluffy, uh, you know, speculation. And I just, I don't think that's helping anybody. Um, we, we really try to focus on like large enterprise brands who literally have billions of dollars at stake. The decisions that they make count. There's real jobs on the line. And so they're not looking for more noise. They're looking for signal. And I think ultimately, when you're looking for signal, you start to reach out to people who aren't going to blow smoke up your ass. They're looking for data points.

55:40They're looking for real research. They're looking for something substantial. And that's the thing that's going to move the industry forward. And I think it's healthy. And you saw it in search. And you're definitely seeing it in the gentic marketing now. Which is interesting. I feel like there's also the parallels in like what's going on within LLMs from like the conversation. Yeah, really sure. And it's like we see the same thing almost in like the influencer bucket from it. Final question for you though, so I can let you go. Five years from now, will search marketers spend more time optimizing for humans or AI agents?

56:19You know, I definitely think that the agentic traffic is going to grow. I think the agentic use cases and applications are going to grow. I would be surprised if humans basically stop using the internet at all. And so I sort of struggle to think that brands will no longer solve for human stakeholders. And I think that there are definitely channels in particular that are, you know, that are not going away, like social, I think is a really good example where, you know, I think consumers are very happy with social experiences and their adoption rates and usage rates sort of show that. But I definitely think that the split is going to shift quite a bit.

57:08And I don't think that we have seen anywhere close to, you know, to maturity on the agentic use cases side. I think that is still in kind of its like Cambrian moments where there's just a proliferation of different applications and use cases and benefits, and that is just going to accelerate. So I think, you know, 12 months from now, 24 months from now, we will be using agents in ways that we can't fathom today. And search marketers and agentic marketers and the rest of the marketing teams will be solving for those applications. So I sort of think of building an agentic marketing practice as kind of climbing a mountain where you sort of get to like the first kind of threshold, the first hill, and that's what enables you to see the next one.

57:53And then you climb that and you can see the next one. And so I think we're going to go through that process here. And I think ultimately that rewards the builders. It rewards the folks who are running experiments. It rewards the folks who have a, you know, a growth mindset and kind of, you know, open to new paradigms. And it punishes the opposite, right? It punishes the folks who are in like wait and see mode. It punishes the folks who are pretty fixed. And, you know, like I always love when I see people have the like SEO versus AEO debate, you know, because of course, like everybody's wrong. And so, you know, you sort of hold that line at your own peril.

58:33And I think that's good. I think anytime we've seen a big, you know, marketing channel cycle on the internet, it has opened up sort of a new class of talent, a new class of companies, and it's created opportunity for the players in the space. And I think we'll see the same thing here. absolutely and i mean i feel like we all decided to join and enter an industry that's like volatile and fast-paced and changing so it's like enjoy it enjoy the volatility the uncomfortableness it's just part of the industry well um alex it's been a pleasure um it's going to wrap up this episode of the voice of search podcast thank you again to alex german co-founder and CEO at Bluefish AI for joining us.

59:22If you'd like to get in touch with Alex, you can find a link to his LinkedIn profile in the show notes or go on over and check out his company's website at bluefishai.com. If you haven't subscribed yet and you want a daily stream of SEO, AEO, GEO, whatever acronym you want to give, hit the subscribe button in your podcast feed and we'll be in your feed in the next following day. That's all for this episode and we'll see you in the next one. Thanks for having me. Thank you.