Episode Transcript
[00:00:01] Speaker A: Welcome to ten Speed Sessions. Today, Kevin and I are going to be talking about kind of all things around LLM visibility and prompt tracking and what we've recently done in a research report specifically looking at what AI sites in the evaluation stage. So very much kind of bottom of funnel consideration, you know, late in the buying journey is kind of what we're looking at. And then also want to kind of talk through the more practical side of LLM visibility, which involves, you know, deciding what prompts to track and how you want to go about structuring that to get your best picture into what your visibility and share voice and all that stuff looks like. So first going to talk about the study and then we're going to kind of jump into some of the practical stuff that we've seen has been working well for us. So, yeah, excited to jump into that.
[00:01:00] Speaker B: Let's do it.
[00:01:01] Speaker A: Yeah. All right. So like I said, we, we recently published a study. We have the, the link in the show notes where we looked at, I will see the number in a moment. A little over 7,000 prompts or, sorry, 7,000 citations across a number of prompts that were specifically built across our clients and lookalike companies. So the competitors or similar companies to get a really good broad set of B2B companies specifically and very much looking at prompts in the LH stage. So we can go ahead and get into that. So the evaluation stage prompt set, so it was 220 prompts. So you can see how those break down in terms of how those were structured here on the right. So, you know, 44 prompts that were comparison focused, 44 that were alternatives, etc. So that's kind of just the, the baseline and kind of establishing how we, how we broke that down.
[00:02:07] Speaker B: I think one thing probably worth clarifying why it's, it's important to look at the evaluation stage specifically versus any other ones.
Was there any intention behind that when building the report?
[00:02:20] Speaker A: Yeah, I think we, we've seen and just understand a lot of disruption around top of funnel and you know what that means. I think a lot of the studies that I've kind of criticized that, you know, evaluate millions, you know, of prompts or millions of citations that are looking all across the funnel. A lot of the, I feel like just noise around this conversation has come from the stuff that maybe is a bit more top of funnel oriented. Yep. And is very generic. And so we wanted to look just at B2B and you know, we're always trying to orient towards conversions towards the parts of the business Impact. And so this is a logical place for us to start, which is, you know, what's actually happening at that, like evaluation, buying, consideration.
[00:03:12] Speaker B: Right. And the other thing too which like will come through in some of the prompt tracking stuff in general and how we approach it is just that it's like really not worth the squeeze to track really many non brand prompts, which is. You alluded to it like.
[00:03:26] Speaker A: Yeah.
[00:03:27] Speaker B: In a similar way to how we would track like a ton of non brand keywords just because they're less likely to be citing brands in those responses. Unless. Which there's like some caveats that like it's like somewhat solution based and we'll get. We can pro. We'll probably touch on that, but I think that that's a key thing as to why this is. We're not really looking at those in this.
[00:03:51] Speaker A: Yeah, absolutely.
[00:03:52] Speaker B: Yeah.
[00:03:52] Speaker A: So this, if you want to kind of talk through just some of the examples here, but this kind of gave some clarity into the study of like what, you know, how these were categorized on that prior slide around, you know, use case and alternatives and that kind of stuff that you can kind of just see. So of the examples here.
[00:04:10] Speaker B: Yeah, I mean I think the important things to look at or call out here are, you know, use case, obviously buying specific high buying intent or alternatives, which I mean all of them have like a varying level of, of intent or clear intent. For a B2B brand, you can kind of think of them as like sol, like solution aware, pro problem aware, solution solution awareness and then brand aware, which is also something we'll probably touch on when we talk about like more about how we build prompts. But yeah, I like these are just really important types of prompts to be tracking because of like what they're going to tell you is best around.
Sorry, getting you the best sentiment around like how your brand is represented within LLMs.
That's really, I think one of the bigger things with these because use cases, alternatives and the buying intent will ultimately lead to that, I guess is that we're getting at. Right.
[00:05:10] Speaker A: Yep.
[00:05:10] Speaker B: Yeah.
[00:05:13] Speaker A: And then so what we. This is kind of getting into now the results. So As I mentioned, 7,387 citations came through here across 170 of the prompts.
So this then was breaking down
[00:05:31] Speaker B: the
[00:05:31] Speaker A: type of content that's actually cited the most for those prompts that we tracked that were evaluation stage. And the kind of the big takeaway and kind of that you see in the headline is brand controllable content dominates. Like this, this category.
[00:05:47] Speaker B: Yeah, yeah. I think this is one of the most important things to probably dive into is the narrative, probably for the past year or so is that there's a lot more third party content related to your brand that's being cited.
I think it's going to be on another one of these slides in terms of just like, you know, a community driven stuff, but just that you, your control, controllable content is the stuff that you should really be heavily investing in. And I think the product pages is something we emphasize a lot but I think we've gotten away from in the narrative around like what's important to focus on to get service LLMs is that it's just heavily weighted there. Product pages are number one.
[00:06:37] Speaker A: Yep.
[00:06:38] Speaker B: And they tend to fall by the wayside.
[00:06:41] Speaker A: Yeah. And it's yes to, to your point. I've Posted about on LinkedIn, I have written newsletter on our sub stack about how everyone was kind of just preaching that Reddit was, you know, the kind of the ultimate destination, the ultimate unlock if you wanted to be cited and. Right. That stems from those studies that were extremely broad. It also stems from studies that were done in partnership with Reddit. Like it's not even like that hard to see some of the, the flaws in that.
And again, you know, we did focus this here. So it's not that third party sites don't have a place in B2B.
But you want to talk about where it matters that your brand shows up the most. It's in these evaluation stage prompts and like the, the results that are being generated, that's where it matters. And you can see discussion is kind of where you would have that like Reddit type of citation is here and it's much further down at 3.7%.
So I agree, I think that I'm the product page. You know, you think about where you can invest time. Like product pages have historically been very, I think we moved from like just telling people what we do to like talking about the benefits. You know, that's been kind of a big thing in product marketing.
However, I think there's probably a bigger need from an LLM standpoint to be very literal about what your product does. Totally. Which can be, you know, maybe a big opportunity for companies to reevaluate through that lens.
[00:08:32] Speaker B: Right. And that also product page, there's like more to it than just we should define that like that can be solutions pages, any like feature pages. So like core marketing pages, what we call them. But to your point, the big thing that people probably aren't talking about as much is that approach. It's about the execution versus how it used to be, which is like being quite literal about what, what you do and not bearing it in like marketing speak and all that. And I think that just doesn't sound like super exciting to like have a big focus on, to help in continuous focus on, to help influence LLM visibility, which I think is one of the bigger like takeaways or like insights I think from this piece that I have gotten.
[00:09:18] Speaker A: Yeah, yeah, and I think so. Yes, you're right. It's product page does kind of bucket several different types. The features, solutions, use cases.
And I think that it, you know, it isn't just a matter of burying all that important detail into FAQs at the bottom of that product page.
It's also not, you know, the, the what seems to be more gimmicky at this point. Like put it in your footer. Hey LLMs, read this like right. You know, try to force all of it in there. I think there's opportunities like you said to, to reevaluate even if it doesn't feel, you know, that, that sexy to, to go do that work. But like there definitely is a huge impact there. Articles, you know, are a big part. Obviously Listicles would typically kind of fall into that same part of the site, but different approach to the content. And then I think, you know, it's, it's still worth mentioning that like the, the homepage is still, you know, oh yeah, an important piece. It's lower on this list but obviously does, you know, carry a lot of weight there and then certainly from a conversion standpoint and kind of getting people further, getting to the right places from the homepage, all that at this stage is very important.
[00:10:42] Speaker B: Yeah, I think, yeah. Considering all those pages and page types or content types is really important.
I think there's a way to look at this from like a strategy, overall strategy approach to your website, which is how we think about it. And I know it seems biased because this is our report, but like if you think about what's stack ranked here you have product pages and then you have articles comparisons, listicles, how to guides. I mean you broke that down. But like we always emphasize like it's so important and depending on the size of your website that you should be expanding that initial foundation. Right.
To use a classic house cliche metaphor. But like that's the foundation. You expand that as the product grows. Right. That should be solutions, pages, use cases, all that. And then all those other things that are underneath it are the things that feed into that. So like you can have that those product page foundations that support the additional educational or not just like broad educational things but like core to your ICP that feed into that and then higher intent being comparisons and listicles and how to guides. But like it just, it just like seems so very like obvious that that's like the hierarchy of how you should approach your strategy.
[00:11:54] Speaker A: Yep.
[00:11:55] Speaker B: And we kind of approach that with all of our clients.
[00:11:58] Speaker A: Yeah. And that is a good opportunity for me to clarify that was a big part of why we wanted to include lookalike companies in the data set as well and not just look at our own client data was to remove that bias that we would be the results would just be a reflection of our approach to strategy.
[00:12:23] Speaker B: Right.
[00:12:23] Speaker A: Even though it's not identical for every client. And we, you know there's variation, there's certainly you know, core beliefs and things that we've seen that we know work.
So that was a big part of adding the lookalikes was to.
To make sure that that wasn't like a bias coming through in the data. Yep.
Yeah. So this is not. We don't need to spend a lot of time here but really when you add in all the community driven sources here again on the evaluation stage, it does make up an extremely small piece of that. So only 4%, 88% of the of the piece was. Was brand controllable content format. So yeah, I don't anything else to add here.
[00:13:07] Speaker B: Well, I was going to ask you what you think of then, what the broader narrative that seems to be driven around like community driven sources being so important.
Why do you feel like, why do you think that that ends up becoming the most the big focus? I have thoughts there but like your thoughts on that being what everyone is emphasizing now and this data is telling us otherwise.
[00:13:30] Speaker A: I still think it's just largely from the extremely broad studies being kind of the main driver. But what do you think?
[00:13:38] Speaker B: Well, it's just, I think it's just like we said earlier, the sexy new thing that it seems to be able to be like hey, let's get there first and we can focus here and those broad data sets or studies have the data that leans that way, then it's a really defensible way to go. Like we should be investing here. Why are we doing all these things that we used to do? Like we should go here.
[00:13:58] Speaker A: Sure. Yeah.
[00:13:58] Speaker B: So but like it's easy to sell or buy into those newer things because it's not the. Everyone's seeking that new like innovative like forward thinking thing when at the end of the day, like who knows if Reddit will end up getting filtered out like Google did and at one point.
[00:14:18] Speaker A: So yes, and I've, I've said, and I will say again, I do think when you go really prompt by prompt citation, look at what, you know, Reddit threads are surfaced, look at the quality of the conversation in there.
Again, particularly around B2B, I think that it's pretty weak. And so I do think that another piece, whether it's filtered out or just LLMs add a layer. I mean you think about the decades of refinement that have happened with algorithm updates in Google.
Like the LLMs are so early on that like, right, there's going to, it's going to be the same thing. Like if we want people to keep using us and trusting the results, we need to go put a little more effort into what exactly are we citing and surfacing and is that actually the best or is it just say the brand name and therefore we're going to pull it in?
[00:15:14] Speaker B: Right.
[00:15:14] Speaker A: And so I think there's will probably be on the LLM front a bit more of a curation of or you know, filtering or algorithm aspect to, to understand the quality of, of those pieces and therefore it may diminish. It's like broad representation across so many queries. Yeah, but yeah, I, I agree. I think there's a lot, a lot there and it, to your point, it's, it's easier to go to your CMO or CEO and say, yeah, we're going to invest in Reddit and you know, this is a new opportunity for us versus we're going to update our product page.
[00:15:53] Speaker B: Right. In touch regularly.
[00:15:55] Speaker A: Yeah. Yeah. So, yeah, I think it's a fair point.
Yeah. So the next takeaway was that comparison prompts punch above their weight again at the evaluation stage.
So any thoughts on this part?
[00:16:10] Speaker B: Yeah, I mean we do weight prompt tracking in this area so like we'll have more data but like it's still really important to create content. I think it's just a reflection of one, the value that that content can provide to a buyer in the journey to buying your product.
They want to, you should provide resources that compare content strategically and, or, I'm sorry, thoughtfully and well thought out and researched and all that. So I mean it's kind of a. Yeah, duh. Like this is what, where you should be focusing. I think if anything it's another one of those, the, the bigger narrative that has been out there for the past like year, six months to a year is that like this type of content is not like worth investing in anymore. But I think that that's separate from.
That's going after the format and not the execution. Like the execution is still worthwhile. The content is still worth going after if you execute properly. It's not that this type of content you shouldn't invest in, but I think people out there would say otherwise.
[00:17:20] Speaker A: Yeah.
And I think that the thing that came to mind for me on this was again, understanding what the LLMs want to see in the content.
The nature of how these are written feels very favorable for that. Like, this isn't surprising to me because you are, you're very literally, literally saying we do this, we have a bidirectional sync with Salesforce. They do not. We have this.
[00:17:55] Speaker B: Right?
[00:17:56] Speaker A: They do not. We do this. They do not. Like, it is very factual and it is very direct, you know, versus the, some other like things that you could be doing. So it does make sense to me that it would, you know, have an outsized influence on versus the number of prompts that it, that we had in the set. Yep. Yep.
Yeah. So that was that. Anything else on the study particular?
[00:18:24] Speaker B: No, I, I think, you know, validating in a lot of ways. There's some insightful stuff. Like one of the things I, I recently posted about was I just even being surprised by the number of the homepage citations which you called out earlier, which I think is really interesting just because of the, the way I, I kind of put it was it's kind of seemed almost old school SEO where when you first originally were trying to optimize for visibility within search, you were really trying to get your homepage to like kind of hearkens back to that, that like maybe that's a new goal. Like I know you want to have all these pages but like at the end of the day there's a brand in homepage play that seems more important than it has been in recent memory, which is interesting. There's probably more to unpack there. But like that's one of the more interesting takeaways I found.
[00:19:12] Speaker A: Yeah.
Yeah. So like I said, we'll have the, the link to that study in the show notes. The other piece was we like that's just the first. We are looking at, you know, similarly like the prompts in other parts of the buying journey.
So we will look further up the funnel and kind of understand how that difference, the difference in citations and sources and page types and content and all that.
So we do have more of that coming as well that I think will Be good to look at over time and then we can start to compare, like I said.
Yeah. With that we can shift into the visibility and like. So there are a lot of tools out there at this point.
Many of them will generate prompts for you or recommend that like look at your website and just recommend them, which is fine. But like that's the really, really important piece of it is like you are tracking your visibility, your share voice, your sentiment across whatever prompts are put in to track. And so depending on what you put in, like, or if you go manually, put in all the wrong things, it can look really, really bad or it can be kind of an impartial or incorrect view and make you think that you're doing better than you are.
[00:20:46] Speaker B: Right.
[00:20:47] Speaker A: So we, recognizing this, wanted to build a framework to have a very consistent view of the different parts of the buying journey you mentioned before, like problem aware, solution aware, et cetera.
So we wanted to find a way to do that in like a consistent way so that when we track visibility in LLMs for clients, we can know we have a pretty good baseline and that we're confident in the things that are being tracked versus letting a tool, you know, be pretty good for this brand, but not very good for this one.
So. So yeah, I want to kind of let you go ahead and start to jump in on what we're doing and kind of how we came about it and why it matters.
[00:21:30] Speaker B: Yeah, I think building off of what you said, it's about, it's all about the inputs and you controlling that. Like you have to have a framework, otherwise it's just going to be really, just all over the place and not understanding what your visibility is. And at the end of the day it's all still pretty like much a proxy for what your visibility really is. It's because alums are so dynamic. Prompts are unpredictable. We don't have access to like, like keyword data. Like we don't. We have newer proxies for volume and stuff with some of these tools, like prompt volume, which is still, you know, questionable but. But yeah, you have to have a framework to define what that is and then have strong internal or if your agency, you know, a client buy in alignment on what those things are. So anyway, long way of saying is like you just really need to have like a really structured and strategic approach to it.
The way we looked at it was to your point, like breaking it down by what kind of prompt it is and then where the buyer is at in the journey. So what kind of prompt it Is, is the like, category comparison, all that type of stuff, and then the funnel stages being problem aware, solution aware, product aware.
[00:22:40] Speaker A: Yep.
[00:22:41] Speaker B: And that has really helped to then understand where we should focus, because within that, we wanted to weight it heavier in certain areas because of what we were talking about earlier. Clearly there is more visibility and more important visibility to be tracked within the higher intent, lower funnel prompts.
So that's at least big part of the framework, majority of it at a high level. I'll stop there.
What questions do you have to continue to drive this discussion?
[00:23:17] Speaker A: Yeah, I mean, I think probably don't have any at the moment. If you want to go ahead and dive in a bit deeper into, I'll keep rolling to what they are and then. Yeah, go from there.
[00:23:29] Speaker B: Yeah, so the, that, that problem, that funnel stage definition is like, really, really important. The other aspects of it that we wanted to make sure we locked in on was a set number.
[00:23:43] Speaker A: Yeah.
[00:23:44] Speaker B: So, you know, historically, especially on the SEO side of things, keyword tracking was a thing. We've gotten away from that a long time ago in terms of just like, we need to track all these different keywords. Like, we want to track pages and specific topics and things that are mapped to the business. So we take a similar approach.
But there's only so much you can track at this point, like in. Because otherwise there is a significant investment, one which isn't the, like, number one prior or a reason why you wouldn't want to track more, but it's also just unnecessary. So like, we set on a number of like 40 to 50 is like the MVP of like a proxy of AI visibility.
[00:24:23] Speaker A: Yeah.
[00:24:24] Speaker B: Mainly because if.
If you need more than that, then like, there's probably something along the lines like your ICP is probably either broad or not well defined or maybe you have too many of them. Yeah. Um, but within a 40 to 50 set, you should be able to build a number within those tiers and, and categories or, sorry, funnel stages that like, give you a good sense of like, visibility throughout all of that. And then it's within that framework that you should be probably, like, adjusting and optimizing over time. But, like, that should give you a good setup. Like, they have visibility.
[00:25:00] Speaker A: Yeah, yeah, I could see that. Or if you have like, a lot of.
Not a lot. If you have like, multiple very distinct products, then in that case you probably need to treat each of those as, you know, 40 to 50 or.
[00:25:18] Speaker B: Totally.
[00:25:18] Speaker A: Yeah, something along those lines. But yeah, I think it's like, it's really interesting and it's also you know, looking at the data, it is something you want obviously you need, you need to be flexible and update things over time. But like it is one of those things that at the aggregate level if you go keep changing the underlying prompts you're tracking on a regular basis, then it's hard to see that roll up view of like are you actually improving? Which you know, you should be looking more granularly anyway. And we do but like that is a big part of it.
And I do think the going back to just like the reason for it in general is also understanding that the query fan out data that comes from tracking specific prompts does matter a lot. And so putting in, you know, a bunch of very broad top of funnel prompts is not necessarily going to give you that same view into the, you know, the fan out queries and then how that can help inform improvements to your site or new content to create or any of that kind of stuff. So yeah, I think that's good.
So what, what's the breakdown of kind of how we of like the waiting building?
[00:26:47] Speaker B: Yeah, well we go heavier on, on the feature use case. So like if you're looking at a 50 prompt, you're looking at 10 for category, 10 for comparison, 25 for the like feature use case. And then you don't need to track that many on brand. I think it's important because you want to have sentiment on like what your brand is and that's like really why you're tracking that. If you're not showing up for your brand, there's some crazy problem there that needs to be explored before you know what I mean? But it's more for like making sure that you're understanding how it's talking about your brand and then influencing and changing the sentiment there in perception.
So that's kind of the breakdown there. And then like we kind of break them down evenly throughout the funnel stages at least. But I think it's more in the category or the tiers that are where we weigh them a little bit more.
[00:27:38] Speaker A: Got it. So kind of equal numbers of problem aware, solution awareness, product, aware, product, whatever. And then within those you have more granular weighting on.
[00:27:53] Speaker B: Yes.
[00:27:54] Speaker A: The types of prompts that we showed in the study, the labeling and yeah, all that. So.
[00:27:58] Speaker B: Right.
[00:27:59] Speaker A: Yeah.
Sorry, were you gonna say something?
[00:28:02] Speaker B: Yeah, well I was gonna we. Because we have to talk about also like how you actually build the prompts.
[00:28:06] Speaker A: Yeah, yeah, yeah.
[00:28:08] Speaker B: Which is like maybe what we would have like should have led with. But I think the framework is how you actually have to like you need that to then choose which ones you're going to put into those buckets.
[00:28:19] Speaker A: Yep. So anyway, yeah, go for it.
[00:28:22] Speaker B: So, yeah, I think what's critical because the prompts are as we, what we've seen is you mentioned query, fan out.
Prompts are still like this, like proxy. Right. Like people are changing the way that they speak or prompt or ask an LLM a question. So it's like, it's kind of like a representation of what could be possibly a bunch of variations of a possible question or prompt. Right.
But the goal is to get it to be like a natural, like something that seems natural language because of the way people use those tools or how we understand they use them. Now I do believe, I'm sure there are some people who are still prompting them in a way that they would prompt search where they're just going best X software. Right. Like, but I think, you know, the way it's heading is that people are more like asking, hey, what is the best X software for this specific problem that I'm trying to solve for this specific type of, you know, you know, industry.
So what we've done is especially because we're working with a bunch of different clients, but I think for anybody who's trying to build them is like, it's all about the inputs. You need to provide as much information as possible in a structured way. Right. Like, but you know, sales calls, product information, feature breakdowns, all that type of stuff, the sales calls and like actually like real demo demos, like product demos, even if you have webinars or whatever, anything that has people speaking about things in a very natural language way is like going to get you the like actual good outputs that you need. And this is where it's different than, like you said, those tools that'll recommend prompts, like, they won't do that. I don't know of a tool yet that takes that into consideration. Even when we have, which we do use peak, we can talk about that. It's been a great tool so far.
They have a recommendation engine but it's based off of like what you already have in there. And it still goes really broad with the recommendations. Like, we haven't chosen any of the suggested ones.
So. Yeah, I'll stop there for now. Yeah, I feel like since you wanting to say something.
[00:30:36] Speaker A: Yeah, we've, you know, inputs matter. Yeah. And so, yeah, I think that's extremely valuable and a really strong case for more than what I said at the beginning against just taking whatever the tool recommends based on some, you know, generic description of your ideal customer and you know, it scrapes your website and looks at that and like doesn't, doesn't really get to the heart of it the way some of that does. And I do think it's also important clarifying like you know, the tools track a lot of them daily or weekly. You know, there's a lot of the tools themselves are certainly pushing for.
You should track daily because things are so dynamic in terms of how LLMs recommend and I'm sure it is far more dynamic than serps. But like that's not any different than like you know, what is or yeah, what is project management as a query realistically over the thousands of queries like the pages are not in the same spot. That's why you literally have like average position right. Search console. So like it's not like you. The search was always super static and LLM is, is extremely different but that is a variable.
And then the other that I was thinking of,
[00:32:07] Speaker B: I lost it just means just keep rolling.
[00:32:11] Speaker A: Yeah.
So I, I do think those are important pieces to the, to the equation I guess in, in understanding the.
Oh, the other piece was like you said, the, the natural language and the like collection. It's not necessarily just because the prompts are longer. You know, what is the best, you know, CRM with you know, these features for this size of company.
It's not literally just looking at your visibility and people searching like an exact match query.
[00:32:52] Speaker B: Right.
[00:32:52] Speaker A: You know, it is more like a topic.
[00:32:55] Speaker B: Right.
[00:32:55] Speaker A: And the semantic related, like what's the intent? So like exactly wording may be different but like is the intent the same? And I don't know to what degree but like a 50 person company or a 70 person company or whatever like in your query probably still generally gives the same result, you know, and whatnot. So yeah, I do think that's another important thing that people can kind of get hung up on in terms of like what prompts to track. Like how do I know which one to pick because it's so long and so specific. But like that that is the reality is that it's not just that like exact match right Prompt, it's semantically related.
[00:33:38] Speaker B: Totally. And that's kind of what I was saying about it like being like I overuse the term proxy but like or directional or whatever I'm using as an excuse to play like to, to make it understood that it's not something that you can like lock in the query fan out component is what's really the underlying important part there. Because that topic I think will generally repeat. That's where it's actually more static, I think than anything. If you look at the, the prompts that we track in the query fan out, that happens after that. That's where it gets a little bit more generic like that, that one thing is triggering a series of like, pretty, pretty consistently similar like queries.
So that's why I think you can have confidence in like sticking with that one prompt that might be more like kind of like a representation of a topic.
[00:34:30] Speaker A: Yeah.
[00:34:31] Speaker B: So that's important. And yeah, to your point, like with the tracking over time, like I just emphasize that you need to, you need to stick with it because the more data you have, the more you're going to have a better understanding of like what that visibility looks like. And I think there's going to be a tendency, and I think we've seen that a little bit so far to want to like, let's, let's track more prompts or let's track different prompts or like why we track. Let's change it up.
[00:34:54] Speaker A: Right.
[00:34:55] Speaker B: But you're just where the way this is built right now. I mean you just need to kind of stay consistent.
[00:35:01] Speaker A: Yeah.
And then one other piece I want to talk about real quick that matters I think a good amount and is very different from anyone coming from a background around kind of classic like rank tracking and just optimizing for organic search is, you know, in, in organic search you're ranking your own content which you have created and have full control over.
The LLM is mentioning your brand, but not necessarily favorably. Yeah.
And so that is the sentiment piece that's built into a lot of the tools and doesn't matter.
You know, it can be.
And making sure that's accurate, like what, you know, what it's reflecting and all that. Because it may mention your brand, but it may say, you know, it's best for SMBs and not ideal for enterprise. And that's kind of what you're asking or whatever when you may be going what? No, like we have tons of enterprise customers and. Right.
You know, and then therefore in that enterprise lens it's, it's a kind of a negative sentiment. You know, obviously there could be far more negative things if it's kind of scraping a lot of reviews and stuff to really call out those issues. But that is a very different dynamic but real aspect to ll invisibility training.
[00:36:36] Speaker B: Absolutely. And I know we like not fully represented in the research report, but you mentioned review sites and we've encouraged and always advocate for taking control of that aspect of it that isn't the like, that is owned, like. Right. And an owned motion that you can do is like try to increase as many like external mentions of your brand that are positive. And again, if, if the overwhelming, if you're like been around a while and you have maybe not overwhelming, but if it skews negative, there's obviously a lot of different things that need to happen. You can't just change that. Like, it's very different than it used to be within search where like you could try to not manipulate the results in your favor, but you could like at one point in time and this is more like no. Like if they're talking about your product in a negative way, overwhelmingly so in these areas, then that might be a product problem. So you need to get product involved, you need to get customer success involved. And then you have to go like on an actual like product marketing and CS kind of initiative to try and get those or customer marketing initiative to get more reviews once you've maybe solved those problems or listen to them and use that.
[00:37:44] Speaker A: Yes.
[00:37:44] Speaker B: So like, I think that that gets lost in all of this is that especially like I'd say just when you're talking to anybody who says that they're an expert in this and that they can like influence the models or like the sentiment of things. Like, there's a lot that is out of a marketer, a marketer's control, who is like an expert in AEO or SEO.
[00:38:05] Speaker A: Sure.
[00:38:05] Speaker B: Like, it is a business thing.
[00:38:06] Speaker A: Yep. Yeah. Yeah. That's the one take that I've had for from very early on that I still believe in just as much, if not more, which is that you're no longer able to hide behind good marketing or. Right. A good SEO program or something. Like it's.
You have to be a good company, you have to have a good product. Like, you have to do a good job with your customers. Like all that stuff. And so yes, to your point, if all the reviews are because you're whatever scheduler, part of your product was crappy and people had issues with it. But that was all from like four or five years ago. But that's all. It's out there. Then yes, you do have to like find a way to get some reviews on whatever you've done with the scheduler now to. To make it better so that there's more updated information and the ll can. Can pull that in appropriately. Yeah.
[00:39:01] Speaker B: And I think the good thing to point out or like to make it not so like grim is, is that now and it will continue to get better is that there are tools that actually do give you a better set, like peak. Does it? There's a lot of these LLM visibility tools or AI visibility tools that do give you a better sense of like where to at least like start, hey, this source or this is favoring this brand and this sentence, this is the sentiment there. So like I, I think six months ago maybe or probably a year ago was probably like more of a guessing game. So if anything, at least now you have the inputs to go back internally and go like he, this is where it's happening and this is how what we need to do to like have that influence.
[00:39:44] Speaker A: Yes. Yeah. When you think about the classic like thing that people talk about, like CEO messaged me that, you know, I put this into chat GPT and we didn't show up.
[00:39:57] Speaker B: Right.
[00:39:58] Speaker A: Yeah, you're right. Like I think a year ago no one really knew how to answer that question. And I think now, assuming your CEO didn't put in something completely random, if it's like somewhat related to a prompt you're tracking, then you would have the ability to look into that and go see, you know, what are the inputs, what's being cited and go, you know, kind of dig through and figure out what you need to do to show up there. Yeah. If you're unable to push back against your CEO, why you shouldn't go on a. Yeah. A wild goose chase.
[00:40:32] Speaker B: But yeah, smaller company, separate problems, maybe a separate discussion.
Yeah. One thing I'd mention too because we probably aren't going to, we're not going to get into it here. Maybe another future episode is emphasizing too on the prompt tracking and the LM visibility tracking that like, because I think there's still probably a misunderstanding of like what that really means. Like visibility is visibility and it's not necessarily a direct one to one to traffic and conversions. There are other KPIs that you need to start looking at as you track those that are going to help understand like how all of this is performing. And the big notable one too, which there's been some recent validation by third parties. Again not for now, but we have been advocating for that. These things, if you are tracking the right things and you are doing the right things to influence that visibility and sentiment within elons that you should see a correlation in your brand traffic going up and then ultimately that means that there's higher intent traffic coming through and then if that correlates to conversions and then down funnel impact, I think you have something there. Yeah, but those are other things that like, I think are important, distinct, like to separate out or not separate out, but understand that like this is tracking this and then you should look at this as the other thing that it's influencing.
[00:41:52] Speaker A: Yeah, okay. Yeah, yeah. I, I think there's a lot there.
Hopefully this was helpful on the. The practical prompt tracking side. Happy to, you know, if you ever have any questions and want to reach out, I'm happy to. To chat to you on that. And like I said, we'll have the link to the study in the show notes that has like all the details on our methodology and all that stuff as well. So be sure to visit newsletter tensby IO. That's our substack where you get monthly newsletter plus early access to the episodes that are coming out here.
And visit 10 Speed I.O. for any more information or if you'd like to reach out and chat with us about maybe helping you out with that, we'll wrap it up.
Thanks for all the good discussion on this and yeah, information on prompt tracking and all this stuff.
[00:42:48] Speaker B: It's great. I'm excited that we have more and more information and more confidence in the things that we're able to do here.
[00:42:54] Speaker A: Yeah, yeah, likewise. Thanks.
[00:42:56] Speaker B: Thanks.
