Latest / Found in AI: AI Search Visibility, SEO, & GEO / Who’s Responsible for What AI Says About Your Brand?

Transcript

Who’s Responsible for What AI Says About Your Brand? · Found in AI: AI Search Visibility, SEO, & GEO

• 0:41
  SPEAKER_02: If AI is doing all that for you and you're not represented accurately or in a positive light, or you know, or against the right c competitors, you've already lost.
• 0:51
  SPEAKER_00: Hey, welcome back to Found in AI.
• 0:53
  SPEAKER_00: I'm Cassie Clark, an AI Search Visibility Consultant, and today we have a returning guest.
• 0:59
  SPEAKER_00: Tommy Landry is back on the show.
• 1:01
  SPEAKER_00: And this time we are talking about something that I think is becoming a much bigger issue as companies start using AI across more of their organizations.
• 1:10
  SPEAKER_00: We're talking about governance.
• 1:13
  SPEAKER_00: Because there are really two sides to this.
• 1:15
  SPEAKER_00: There's the information your company is feeding into AI internally, that's your CRM data, your reports, sales information, your content workflows, all of it.
• 1:25
  SPEAKER_00: And then there are the signals AI engines are using externally to understand and represent your brand.
• 1:31
  SPEAKER_00: If either side is inaccurate, inconsistent, or just plain messy, AI can very confidently give you an answer that is absolutely completely wrong.
• 1:41
  SPEAKER_00: Tommy and I talk about why companies need humans in the loop.
• 1:45
  SPEAKER_00: Who should actually be responsible for AI governance?
• 1:48
  SPEAKER_00: How inconsistent messaging across departments can affect your AI representation, and why measuring AI visibility needs to go beyond simply counting citations.
• 1:59
  SPEAKER_00: We also talk about Tommy's new book, The Signal and the Source, and his framework for governing AI internally and externally.
• 2:07
  SPEAKER_00: Let's get right into it.
• 2:20
  SPEAKER_02: Thank you for having me back for a second appearance on here.
• 2:22
  SPEAKER_02: We uh had a had a chat a few months ago.
• 2:24
  SPEAKER_02: I run a company called Return on Now, uh, and we have two offerings.
• 2:28
  SPEAKER_02: We focus on external AI visibility, so like your AEO, GEO, and SEO, but also internal workflow governance where you're automating things with AI and making sure that AI doesn't run completely off the rails.
• 2:41
  SPEAKER_00: So tell tell me how Go ahead.
• 2:45
  SPEAKER_02: No, I was gonna say, and I forgot to mention I recently published a book about HAFE, which is my framework for blending the two, that talks about how they how you can manage across internal and external using similar processes to make sure that AI has the right signals to ingest so that it outputs accurate information about your company or your data or whatever it is it's analyzing.
• 3:06
  SPEAKER_00: So tell us about the overlap between the external and internal signals.
• 3:10
  SPEAKER_00: What does that look like?
• 3:13
  SPEAKER_02: Well, so it's really kind of a continuum.
• 3:15
  SPEAKER_02: Um AI is in and of itself, it doesn't just it doesn't try to fill in the gaps naturally on its own.
• 3:21
  SPEAKER_02: So if you give it incomplete or inaccurate data or incomplete or inaccurate content, it's the same processing, it's just happening for a different purpose, right?
• 3:30
  SPEAKER_02: So AI is gonna it will just make up things to fill in the gaps sometimes if you don't have a proper governance process in place to manage how it's doing it.
• 3:39
  SPEAKER_02: So um for the internal, you know, do you have clean data feeding it in the first place?
• 3:43
  SPEAKER_02: So a lot of folks I talk to, they end up with a CRM issue going on where the AI is actually pulling from the CRM, but it's maybe pulling inaccurate data, incomplete data, um, dirty data, whatever, uh, a lot of duplicates, and it's thinking you have more opportunities than you do.
• 3:59
  SPEAKER_02: And then it's summarizing all that very confidently into an output report that people are making decisions on without having so all can't speak, without all the time having some person actually analyze the results and confirm that it used the right data from the right system and that the conclusions it's pushing are actually accurate.
• 4:19
  SPEAKER_02: Externally, uh, the systems are actually able to see all the content you have on your website, but there's also oftentimes an ability for those external systems to see documents you don't realize they see.
• 4:29
  SPEAKER_02: They could be looking at PDFs, they might be looking at a sales presentation that somehow made its way to AI.
• 4:34
  SPEAKER_02: Uh, a lot of people integrate it into Slack and email and things like that, and that actually can potentially affect your external visibility too.
• 4:41
  SPEAKER_02: Um, and I know a lot of people don't really talk about this, but it's a it's a real concern when you start looking at the continuum, you know, the internal and the external.
• 4:50
  SPEAKER_02: So um, it really becomes a top-down issue when you're talking about it this way.
• 4:54
  SPEAKER_02: It's not just, hey, let's hand the SEO people a project to go and AEO the site, right?
• 4:59
  SPEAKER_02: And that's you can do that.
• 5:01
  SPEAKER_02: But it's really about as an organization, do you position yourself the same way across functions?
• 5:08
  SPEAKER_02: Right?
• 5:09
  SPEAKER_02: Does your RevOps say the same thing as sales?
• 5:11
  SPEAKER_02: Do they say the same as marketing?
• 5:13
  SPEAKER_02: And then do the third-party uh platforms that you know are so important for GEO, do they say the same thing?
• 5:19
  SPEAKER_02: You've got to have alignment now all the way across.
• 5:21
  SPEAKER_02: And it's really a much bigger problem than sort of the tactical approach we've seen a lot of people take in the last year or so.
• 5:27
  SPEAKER_00: I'm actually really glad that you brought up the sales piece, like the sales presentation that's made its way to AI.
• 5:32
  SPEAKER_00: A couple of weeks ago, I had David Kirkdorfer on, and he was talking about you know, sometimes the sales team speaks the language to whoever they're talking to, so it makes it easier for them to understand.
• 5:44
  SPEAKER_00: But if you do that enough, you really change the positioning and you don't realize it across the department is what you've done.
• 5:51
  SPEAKER_02: Right.
• 5:52
  SPEAKER_02: And now imagine your external AI systems get a hold of that and they've got convoluted messaging.
• 5:57
  SPEAKER_02: So they're making promises that you actually can't fulfill.
• 6:00
  SPEAKER_02: And then your rest of your sales team is sitting around unraveling the yarn before they can actually sell.
• 6:05
  SPEAKER_02: It's an efficiency problem that's creating.
• 6:08
  SPEAKER_00: Right.
• 6:08
  SPEAKER_02: You can't customize everything for everyone.
• 6:11
  SPEAKER_00: Right.
• 6:11
  SPEAKER_00: So you mentioned the governance team governance.
• 6:14
  SPEAKER_00: I'm gonna call it the government's government governance team.
• 6:17
  SPEAKER_00: If I could get it out.
• 6:18
  SPEAKER_00: Oh my goodness.
• 6:19
  SPEAKER_00: Um, what what do you recommend when you're talking to brands about setting up the internal and external governance?
• 6:27
  SPEAKER_02: You know, it really comes down to there should be someone who's I hate to say it this way, but has the X on their back, right?
• 6:33
  SPEAKER_02: Like the final check for does content align with the overarching messaging that the company wants to actually put out in the market, right?
• 6:41
  SPEAKER_02: And internally, there should be someone who is is actually, and it might be in a function, right?
• 6:46
  SPEAKER_02: It might be the RevOps RevOps analysts, the final say on whether a report actually goes anywhere.
• 6:51
  SPEAKER_02: Like, do the execs see this?
• 6:53
  SPEAKER_02: Because they might make a decision on it, right?
• 6:55
  SPEAKER_02: Let's go ahead and have someone verify that it's pulling the right data, right?
• 6:59
  SPEAKER_02: I saw one example where a company uh had Uber automated their forecasting and their historical analysis, and they were making rev, you know, revenue like budget investment decisions, they were making headcount decisions, allocating things, they were making campaign decisions.
• 7:16
  SPEAKER_02: And then someone caught that the automated AI agent was pulling from the wrong fiscal year, but the right quarter.
• 7:23
  SPEAKER_02: So it looked right.
• 7:24
  SPEAKER_02: And when they actually caught it, they had to spend six months making strategic decisions on a year-old data.
• 7:31
  SPEAKER_02: So imagine the mess.
• 7:32
  SPEAKER_02: Yeah, I I never did see the final fallout from that, but it was uh not pretty.
• 7:37
  SPEAKER_00: Yeah.
• 7:38
  SPEAKER_02: And so this could happen.
• 7:39
  SPEAKER_00: That is definitely messy.
• 7:41
  SPEAKER_00: So your inputs that you're putting into your AI engine or your AI assistant, whichever one you're using, absolutely have to be correct.
• 7:48
  SPEAKER_00: And then what you're putting out also has to be correct.
• 7:51
  SPEAKER_00: So I guess that goes back to the argument of you absolutely have to have that human in the loop.
• 7:56
  SPEAKER_00: Like you cannot just leave this alone.
• 7:59
  SPEAKER_02: Yes.
• 8:00
  SPEAKER_02: And that's actually the the book is about how to apply my model hate, which is one way to get a human in the loop.
• 8:06
  SPEAKER_02: But it really starts by verifying the sources, right?
• 8:09
  SPEAKER_02: Are they the right sources?
• 8:11
  SPEAKER_02: What does it have access to?
• 8:12
  SPEAKER_02: Um, it looks at what AI is allowed to make decisions on, right?
• 8:16
  SPEAKER_02: Where where are the guardrails?
• 8:17
  SPEAKER_02: What can and can it not do autonomously before bringing in some human?
• 8:22
  SPEAKER_02: And then it, you know, where are the checkpoints for people?
• 8:25
  SPEAKER_02: Who are, you know, how do you decide who who does it?
• 8:28
  SPEAKER_02: And um, and what what's the final uh the way that they handle like exceptions or problems or like what happens then versus when do they when they let it go through and actually make it to someone else, a stakeholder of some sort?
• 8:40
  SPEAKER_02: I mean, it could be external, right?
• 8:42
  SPEAKER_02: Let's say you're running an agency and you're you got it running your reporting for you, you're sending clients these reports, who's looking at those before they go out.
• 8:50
  SPEAKER_02: That I've seen this examples of that running off the rails too.
• 8:54
  SPEAKER_02: So this is you know, this is every organization.
• 8:57
  SPEAKER_02: They should have someone tab to check it out, and it might vary depending on what the report is.
• 9:02
  SPEAKER_02: You need a subject matter expert of some sort.
• 9:05
  SPEAKER_02: Um I have heard talk of some companies trying to put some put a position in place that's like a cross-functional sort of governance person.
• 9:13
  SPEAKER_02: Um whether or not that's gonna work for like you know, verifying your brand's represented properly versus internal data analysis may or may not be the same skill set.
• 9:22
  SPEAKER_02: Maybe you've got someone who's a unicorn, that's great.
• 9:24
  SPEAKER_02: But um, you know, it it's not really having a new organization that that's governance, it's having a checkpoint, right?
• 9:32
  SPEAKER_02: A person.
• 9:33
  SPEAKER_02: So I mean it could be an HR system using AI, and then you need an A someone who understands HR, and that's not a marketing person for sure.
• 9:40
  SPEAKER_00: Right.
• 9:41
  SPEAKER_00: That was gonna be my next question.
• 9:43
  SPEAKER_00: Like, if you are thinking about this in terms of having an AI agent embedded into your content marketing strategy, how do you get someone if you're creating a governance team, like the person that sits on that team and like here are the inputs we're putting in, what does that look like?
• 10:04
  SPEAKER_02: So really, I think there's a difference between trying sort of setting the strategy and the person who's managing the execution.
• 10:10
  SPEAKER_02: You it may be smart to have a governance person who's like strategically overlooking the whole thing, ensuring the right people are in place.
• 10:16
  SPEAKER_02: I've seen companies do it just by bringing in a consultant.
• 10:20
  SPEAKER_02: That's something I can actually help with.
• 10:22
  SPEAKER_02: Go in, speak to the various stakeholders, figure out the inputs, figure out how AI is being used internally, help them decide who and where needs to sit there.
• 10:30
  SPEAKER_02: And then on the marketing side, you can do the same thing, right?
• 10:33
  SPEAKER_02: Because content is also an automated process.
• 10:36
  SPEAKER_02: But the person who's at the end needs to understand strategically what the company needs to be aiming for, right?
• 10:42
  SPEAKER_02: The branding, the messaging, the allowable and unallowable commentary, um, and then the voice, which the voice is the biggest one because that's you know, we know AI slop is all over the place.
• 10:53
  SPEAKER_02: Um, it has all the bad habits, if not X then Y.
• 10:56
  SPEAKER_02: And so you have to strip them all out if you let AI get involved in your content.
• 11:00
  SPEAKER_02: Um, someone to do all that stuff.
• 11:02
  SPEAKER_02: So that's really that's a marketing person, right?
• 11:04
  SPEAKER_02: Um, but someone to help them know when they need to be involved.
• 11:07
  SPEAKER_02: It could be an external consultant, maybe you bring someone in, maybe it becomes half of someone uh someone in an ops role.
• 11:14
  SPEAKER_02: I you know, maybe it's a RevOps person who does it.
• 11:17
  SPEAKER_02: I don't know.
• 11:18
  SPEAKER_02: It was I think that part's still being defined.
• 11:20
  SPEAKER_02: And um right now it makes sense to bring in someone who specializes in this and have them help you define organizationally how to set it up.
• 11:28
  SPEAKER_00: So you are also like really big into AI search optimization as well.
• 11:33
  SPEAKER_00: How does this all fit into that?
• 11:35
  SPEAKER_00: Like if you are sitting down and you're thinking, okay, we need to get our AI search strategy together so we can start now in Q4 or Q1 or whenever.
• 11:45
  SPEAKER_00: What do we think about in terms of having governance on top of it?
• 11:50
  SPEAKER_02: I mean, I think it's still important.
• 11:52
  SPEAKER_02: Um, when you're actually looking at the external strategy, we all know now, right?
• 11:56
  SPEAKER_02: A an AEO, you're looking at extraction, you're looking at structure, a lot of those kind of things.
• 12:01
  SPEAKER_02: So you know, you can't just like hand clawed your SEO and your AEO and have it do it.
• 12:07
  SPEAKER_02: You still need a person there to govern that.
• 12:09
  SPEAKER_02: So maybe you're using AI to do some of that.
• 12:12
  SPEAKER_02: And I've seen I've seen folks do that with reasonably decent success.
• 12:16
  SPEAKER_02: Um, but I've also seen people use these AI platforms to optimize stuff and it came out wrong.
• 12:21
  SPEAKER_02: That's because the AI is like, yeah, close enough.
• 12:24
  SPEAKER_01: Right.
• 12:24
  SPEAKER_02: And so there's governance needed there too.
• 12:26
  SPEAKER_02: Um, your actual brand representation out in the market affects how your AI representation works, right?
• 12:33
  SPEAKER_02: And while a lot of people are running around tracking prompts, which can be useful if they're the right prompts, uh, I I really don't.
• 12:40
  SPEAKER_02: There's very few platforms that feel like I've figured this out, but it's about representation.
• 12:44
  SPEAKER_02: Like, are you there for the right queries?
• 12:47
  SPEAKER_02: Um, are you in the right short list as a brand, right?
• 12:50
  SPEAKER_02: Now we're gonna know the GEO side strongly.
• 12:53
  SPEAKER_02: Um is it accurate what they're saying about you?
• 12:56
  SPEAKER_02: What is the sentiment?
• 12:58
  SPEAKER_02: Are they showing the right competitors?
• 12:59
  SPEAKER_02: Like all these kinds of things.
• 13:01
  SPEAKER_02: There may be folks using prompt tracting that way already already, and that's great.
• 13:04
  SPEAKER_02: But I feel like there should be more of an overarching metric that we haven't figured out yet.
• 13:10
  SPEAKER_02: Um that I think we will, and I wish I knew the answer, but um, you know, how is the health of your your AI-based representation across the big platforms?
• 13:21
  SPEAKER_02: Right.
• 13:21
  SPEAKER_02: And we need to understand it differently across the answer engines and the generative engines because they process the information differently, even with the overlaps they have and the fact that there are now hybrid systems.
• 13:32
  SPEAKER_02: So there's a lot in one statement, but it really applies if you're internal and external.
• 13:39
  SPEAKER_00: If you had to name that metric, what would you call it?
• 13:43
  SPEAKER_02: I I keep using the word representation.
• 13:45
  SPEAKER_02: Like, how are you represented on AI?
• 13:47
  SPEAKER_02: And is it is it in the best interest of the business?
• 13:51
  SPEAKER_02: Um, and you know, with a lot of folks still looking at old things like, well, but we're not getting any visits from it.
• 13:56
  SPEAKER_02: To me, that that's a full stop wait.
• 14:00
  SPEAKER_02: Put the brakes on.
• 14:01
  SPEAKER_02: It's talking about you, whether it sends you visits or not, and if it's not sending you visits, that's actually something to be alarmed about because 80% of people, once they get the AI analysis, already know pretty much what they're gonna buy.
• 14:13
  SPEAKER_02: There's a stat that came out like in June.
• 14:16
  SPEAKER_02: So if AI is now only your comparisons, funnel, like all that stuff that you used to do on your website and be able to nurture people and all the funnel marketing stuff.
• 14:26
  SPEAKER_02: If AI is doing all that for you and you're not represented accurately or in a positive light, or you know, or against the right competitors, you've already lost.
• 14:38
  SPEAKER_02: So um, so to me, that's a big deal.
• 14:40
  SPEAKER_02: We need representation.
• 14:41
  SPEAKER_02: Like, are you there in the right places presented the right way?
• 14:45
  SPEAKER_02: And I think there's either a master metric or there's a master set of prompts, which I'm actually I've actually been kind of noodling around, maybe create the set of core prompts that everyone should be looking at.
• 14:54
  SPEAKER_02: Um, you know, like obviously who is such and such, and is it even close?
• 14:59
  SPEAKER_02: Um, but there's others.
• 15:00
  SPEAKER_02: So um the unfortunate part about it is that it varies depending on your niche, uh, because you might have different topic clusters you want to check for on the answer engines, which suddenly infinite prompts come into the picture.
• 15:13
  SPEAKER_02: Um, but for GEO, I feel like it's something we can get to faster because it's really about your brand and your products or services.
• 15:19
  SPEAKER_02: Like if it understands what you do, who you do it for, and why it matters to them, then you're ahead of most people who are still just running around tracking if they're cited for some blog posts on the network.
• 15:31
  SPEAKER_00: I I think the core argument here is that you need to make sure your brand is represented accurately inside of these AI engines.
• 15:38
  SPEAKER_00: And to do that, you need to have someone sitting on your team making sure everything that goes out or comes in to your AI agent is as accurate as possible.
• 15:48
  SPEAKER_02: Yes.
• 15:49
  SPEAKER_02: And with the authority to actually affect the third-party sites, which we all know, your LinkedIn page, your you know, review sites, G2 or whatever, trustpilot, um, anyone else who talks about you, you know, and it gets it gets complicated if say you have acquired brands that have like conflicting information that's now part of your brand, that can be a challenge.
• 16:13
  SPEAKER_02: You know, so you start, there's so many moving parts.
• 16:15
  SPEAKER_02: It's really it'd be really important to try to designate someone who's like cleaning it all up because the signals you give AI is what they're going to talk about.
• 16:24
  SPEAKER_02: And if the signals are all over the place, then AI might just be like, oh, this is a forward-looking revolutionary company that does great things for great customers instead of what you actually do.
• 16:35
  SPEAKER_00: Right.
• 16:36
  SPEAKER_00: Right.
• 16:37
  SPEAKER_00: Okay, before I let you go, tell us about your book.
• 16:39
  SPEAKER_00: Where can we find it?
• 16:40
  SPEAKER_02: Yeah, the book, um, it's uh it's published on Amazon only right now.
• 16:44
  SPEAKER_02: Um uh looking at options of maybe moving it to some other source sources as well, uh like Goodreads or whatever.
• 16:49
  SPEAKER_02: Um, but you can buy it in paperback hardcover or Kindle edition.
• 16:53
  SPEAKER_02: It's available in KDP uh for folks uh the for folks who have the Kindle uh subscription, it's available there.
• 16:59
  SPEAKER_02: Um it's gotten really good reviews so far.
• 17:01
  SPEAKER_02: I think it's like a 4.7 rating on the initial like eight or 10 that I've gotten.
• 17:05
  SPEAKER_02: Um, so it's being well received so far.
• 17:08
  SPEAKER_02: Uh a little bit bigger of a book than the AEO to GEO book I wrote last year, so it might take a little more time.
• 17:14
  SPEAKER_02: Um, but I feel like it's actually a natural evolution to okay, now look, let's look org-wide at AI and really how it should be managed internally.
• 17:23
  SPEAKER_02: And I think there's a lot that can be learned from it if folks are open to implementing the actual step-by-step that I include in there.
• 17:30
  SPEAKER_02: There's a chapter or two that tells you how to roll it into your own org.
• 17:33
  SPEAKER_02: So it's not, you know, you don't have to bring someone in.
• 17:36
  SPEAKER_02: Um, it would be faster and easier, right?
• 17:38
  SPEAKER_02: But um, it does give good instructions and it there's a lot of um a lot of direction of bringing in more senior level people because you're gonna need that level of sort of leadership to make sure that it actually happens and you reduce friction across groups.
• 17:54
  SPEAKER_00: So and what's the title of your book?
• 17:55
  SPEAKER_00: Just so we get it.
• 17:56
  SPEAKER_02: It's the signal and the source.
• 17:58
  SPEAKER_02: Okay.
• 17:58
  SPEAKER_02: Using HAF, which is my model, HAF H A I F using HAF to govern uh AI internally and externally, basically.
• 18:05
  SPEAKER_02: I'm paraphrasing, but that's the subhead, so it's not as important as the signal and the source.
• 18:12
  SPEAKER_00: Okay, that was Tommy Landry.
• 18:14
  SPEAKER_00: And I think the big idea that I am taking away from this conversation is that AI governance isn't just an IT problem, and AI search visibility isn't really just a marketing problem either.
• 18:26
  SPEAKER_00: Both ultimately go back to the same question.
• 18:29
  SPEAKER_00: What information are we giving these systems, and who is really responsible for making sure it's accurate?
• 18:36
  SPEAKER_00: So here's something that you can do today.
• 18:38
  SPEAKER_00: Pick one AI-powered workflow your company is already using.
• 18:41
  SPEAKER_00: Maybe AI is helping create content.
• 18:44
  SPEAKER_00: Maybe it's generating those reports, maybe your sales team is using it to summarize CRM data.
• 18:50
  SPEAKER_00: Maybe you have an agent doing something automatically in the background.
• 18:54
  SPEAKER_00: I mention all of those things because if you think about it, they all really do impact visibility at some point down the line.
• 19:01
  SPEAKER_00: Go verify that there is a human checkpoint.
• 19:04
  SPEAKER_00: Who is responsible for verifying the output before someone acts on it?
• 19:08
  SPEAKER_00: And more importantly, does that person actually have the subject matter expertise and authority to catch a problem if one were to occur?
• 19:17
  SPEAKER_00: You can't immediately name that person, you've probably just found a governance gap.
• 19:21
  SPEAKER_00: Because as Tommy pointed out, AI can produce something that looks completely reasonable while still pulling from the wrong source, using outdated information, or even just misunderstanding what your company actually means.
• 19:35
  SPEAKER_00: And externally, the stakes are very similar.
• 19:38
  SPEAKER_00: AI engines are already interpreting your brand based on the signals available to them, whether you're actively managing those signals or not.
• 19:47
  SPEAKER_00: If you want to understand how those engines currently represent your brand, that's exactly what I investigate through my AMSearch visibility audit.
• 19:54
  SPEAKER_00: I look at how your brand appears across AI Search, what's shaping that representation, where the gaps are, and what you should prioritize next.
• 20:02
  SPEAKER_00: You can find more information about working with me in the show notes or over on my website at castycalaricmarketing.com.
• 20:09
  SPEAKER_00: And if you want more nerdiness about AMSearch strategy every week in your email inbox, join the visibility report.
• 20:18
  SPEAKER_00: It goes out every Friday, and you can find the information for that in the show notes as well.
• 20:22
  SPEAKER_00: Okay, that's it for today's episode.
• 20:24
  SPEAKER_00: I will see you in the next one.
• 20:26
  SPEAKER_00: Until then, stay visible.