WEBVTT

00:00:04.000 --> 00:00:21.000
Hi, everybody. Welcome in. All right, we have very exciting episode, we'll call it episode of the signal for June. So as a reminder, first of all, thank you all for joining us. And if you're watching live, that's awesome. If you're watching on the recording

00:00:21.000 --> 00:00:39.000
That's great too. The reason Will and I decided that this would be a valuable use of all of our time is there is such an influx in the amount of content about marketing, about digital, about AI, and it is so difficult for us to keep up with it

00:00:39.000 --> 00:00:55.000
And it's kind of just me and Will, like, this is our full-time job to keep up with it, and we get overwhelmed, and we have to constantly check ourselves to say, did you see this? What do you think of that? So the goal of today is to try to cut through the noise to get to the signal

00:00:55.000 --> 00:01:11.000
And share with you what we were most excited to read about, what really got our gears turning in the last month. So Will is going to kick us off with some of what was most inspiring for him. And then I'm going to get into some topics that I wanted to share

00:01:11.000 --> 00:01:19.000
and tee up some thought starters. Then we'll have time for Q&A, so we can kind of go back and forth there. All right. So let's get right into it.

00:01:19.000 --> 00:01:34.000
Let's rock and roll. So a couple weeks ago, one of my little agents caught that we had a 1900 increase in ChatGPT citations to one of our pages. And if you guys know us, we

00:01:34.000 --> 00:01:48.000
We run a bunch of tests. In a world where nobody knows what actually works and doesn't work, let's stop BSing ourselves, right? It's like, the winners are gonna be the ones who show up to this podcast and are like, yo, so you're open sources, all their tests, they show us what happened, they show us all this stuff. So if you're thinking

00:01:48.000 --> 00:01:56.000
How did you get a 1,900% increase in citations from ChatGPT month over month? Next slide

00:01:56.000 --> 00:01:58.000
This ugly thing

00:01:58.000 --> 00:02:13.000
So while most of you are out there, not you, not because you guys would come and join us on these are the smart ones. While everyone else was fighting over whether or not LLMs.txt should I or shouldn't I, we processed what's the risk if I put an LLMs.txt up

00:02:13.000 --> 00:02:26.000
How long does it take to take the information about our website and create a very simple page? This is not our LLMs.txt, by the way. This is a page that was called AI-information. And we put it up on our website

00:02:26.000 --> 00:02:35.000
Six months ago, because we wanted to test. What happens when you give a page pretty much just markdown, and you're basic nav? Next slide.

00:02:35.000 --> 00:02:40.000
That's the spike. But here's the thing that you should take away

00:02:40.000 --> 00:02:45.000
The test started, I believe, in either October or November. The page launched back then

00:02:45.000 --> 00:02:57.000
Now, this is the ChatGPT bump in citations, what you don't see here, and I don't have in the visual, is Gemini started picking the same page up like crazy back in like

00:02:57.000 --> 00:02:59.000
February-ish, and I missed it

00:02:59.000 --> 00:03:10.000
But this is the problem when people ask about quick wins in GEO. Use this as an example. If you had to pick SEER to do your GEO

00:03:10.000 --> 00:03:23.000
two weeks ago, you'd have been like, oh my god, in two weeks, these guys increased my citations 1,900%. If you had to pick the exact same company with the exact same strategy back in October, you would be sitting here 6 months later, seven months later, like, these guys suck

00:03:23.000 --> 00:03:26.000
So, my point there, my friends, is

00:03:26.000 --> 00:03:39.000
When we're being asked internally about creating quick wins, part of the reason why we have to push back and run a lot of experiments is you never know, these models are changing so fast, we can't even, as humans keep up with what did they launch and when

00:03:39.000 --> 00:03:52.000
imagine trying to keep up with a strategy that keeps on top of them all. So just keep your tests out in the public. You never know when they'll be picked up, and they might ultimately have a positive outcome. Next slide.

00:03:52.000 --> 00:03:55.000
All right, so,

00:03:55.000 --> 00:04:12.000
When this came out, a few people were like, hey, Will, Google has some interesting statements about whether or not you should do what you guys did. And the beauty of, anybody can go to the page. First of all, we did not link to everybody once asked, well, did you link to that page? It was only in our sitemap

00:04:12.000 --> 00:04:22.000
Is where it was linked from, which has been making me think a lot that, like, internal linking probably might matter less than we think in an AI and agentic world

00:04:22.000 --> 00:04:34.000
But the rules are changing, so the other thing I would recommend that you do is that you keep on top of what Google's saying can and can't be done, and running it against your experiments, because you might have experimented months ago before Google said anything

00:04:34.000 --> 00:04:49.000
And now they come out and have something to say about it. And keep in mind, also, I think in the last week, Google said LLMs.txt from their web, I think their web spam team was like, it has no use. At the same time, while the Chrome team did a blog post that was like, you should have LLMs.txt

00:04:49.000 --> 00:05:04.000
So even Google's trying to figure this out across their different divisions. So be careful who you listen to and just kind of use your brain, your marketing brain, to be like, these are just facts about my business. Like, if people get penalized for having white text background, black or black text

00:05:04.000 --> 00:05:08.000
on white backgrounds with facts about their companies

00:05:08.000 --> 00:05:12.000
I don't think that's a bad thing. Alright, let's go ahead and keep moving, Lisa

00:05:12.000 --> 00:05:30.000
So this is the next thing that really kind of got me going. Garrett Sussman, who's somebody I recommend all of you follow because he's running tests like this, he's at IPOL rank. Another thing we do at SEER, we shout out our competitors. I think great competitors make you smarter and better. We're not here to get into the bullshit fights over, like, who said what about what. No, thanks.

00:05:30.000 --> 00:05:45.000
I'll let them do that, right? So, you know, Garrett Sussman, you guys should follow him. I love what he's dropping, and I love how he's experimenting on how does Google personalize intelligence affect AI answers in Gemini and AI over I believe it might have just been Gemini

00:05:45.000 --> 00:05:58.000
And next slide, Alisa. What he's finding is that once you connect personal intelligence to your account, he literally saw that brands were being mentioned more in his AI answers

00:05:58.000 --> 00:06:07.000
When he had receipts from them in his Gmail, or when he had a certain brand of a hoodie in his Google photos. So now

00:06:07.000 --> 00:06:24.000
For the first time, remember Google's had Gmail photos and Calendar and contacts for years. They've never used any of those things to make their search any better. Now that they've got competition with ChatGPT and Claude, they go, oh, we should use all these things

00:06:24.000 --> 00:06:30.000
about your history to make the results better and it's happening. So now many of you, you need to take this and say to yourselves, alright

00:06:30.000 --> 00:06:33.000
How am I partnering with the people

00:06:33.000 --> 00:06:35.000
that affect these things.

00:06:35.000 --> 00:06:38.000
How am I partnered with the team that's sending out my emails?

00:06:38.000 --> 00:06:53.000
How might partner with the team that decides where the logo gets imprinted in our product or service, etc. So there's that one as well. How do my receipts look? All those things start considering those as things that could be influencing the test that you're running.

00:06:53.000 --> 00:06:55.000
Next slide

00:06:55.000 --> 00:07:11.000
And to me, I think something I've been trying to say to so many of our clients and prospects is we've got to start to let go of the old world of I could find the MSV and start to just take it to the extreme of every search is completely unique, which means every search is an MSV of one

00:07:11.000 --> 00:07:16.000
Right, that's the way that I'm trying to think about it, because when you go to the extreme, which we know is extreme.

00:07:16.000 --> 00:07:28.000
It makes your brain work differently and think differently about, like, well, how would I survive and thrive in a world where I don't have this information and every search is a unique search? Next slide.

00:07:28.000 --> 00:07:41.000
All right, I put this out the other day. If y'all saw what we showed on Bing webmaster Tools, I'm really impressed by Krishna and the team and what they're doing over there. So go to the next slide.

00:07:41.000 --> 00:07:51.000
Isn't it funny? This is how fast our world is changing. When I made this video two days ago, Google did not have AI announced in their webmaster tools

00:07:51.000 --> 00:07:54.000
In two days, they went, we got it too

00:07:54.000 --> 00:08:09.000
I think Google's is going to be mostly useless because they're not giving you queries. So when I do say MSV of one, it's like, yeah, well, that's hard. So what I love about Bing's Webmaster Tools is it gives you the grounding queries. So you don't get the prompts. And remember

00:08:09.000 --> 00:08:28.000
Perplexity doesn't give you this. ChatGPT doesn't give you this. Claude doesn't give you this. Gemini doesn't give you this. None of them give you the prompts, and neither does Copilot. None of them give you the prompts, but none of them also give you the grounding queries. Bing is the first one. So what you can do is you can start to look at your grounding queries, like I've done here

00:08:28.000 --> 00:08:29.000
Next slide, Alisa.

00:08:29.000 --> 00:08:31.000
And I can go, oh.

00:08:31.000 --> 00:08:47.000
What drives brand mentions in AI answers is a page that I'm not used to seeing, because I log in every week and take a look at my pages, and I'm used to seeing the TAM, SAM, SOM one, I'm used to seeing some of these, but when you see a new one, and it takes the number one spot, next slide.

00:08:47.000 --> 00:08:54.000
You then can actually look at the content and go, oh my God, Bing, for some reason and their partners are now showing

00:08:54.000 --> 00:09:11.000
grounding query that is triggering this piece of content, and it's old AF. So now, guess what we're going to do as a company? This is real-time strategy. So the minute I found that a couple days ago, I recorded the video, and now my marketing team is like, okay, how do we make sure that the question, the grounding queries

00:09:11.000 --> 00:09:12.000
That triggered this

00:09:12.000 --> 00:09:28.000
This page actually answers, because when I looked at it, I said, oh my God, we're showing up for this, but let's make sure it's a quality answer and not just another citation to put in some visibility report. We've got to turn that visibility into believability. We've got to turn it into something that's greater than just we show up

00:09:28.000 --> 00:09:32.000
I want us to show up and also help people that got to that content

00:09:32.000 --> 00:09:34.000
Next slide.

00:09:34.000 --> 00:09:37.000
So, you know, my take on this is

00:09:37.000 --> 00:09:54.000
The visibility thing is old, guys. Like, I mean, it's what everybody's asking about, but it's already old to me because I already seen this… I've already seen this play before. Everybody gets hyped about a vanity metric. You give it to them. Three months later, they go, what's that doing for the business? You go, but I thought you were just asking about visibility. So that's why I think it's so important to ask

00:09:54.000 --> 00:10:03.000
Is this page with these scrounging queries helpful to the people that get to it? And if that answer is no, make the page better.

00:10:03.000 --> 00:10:05.000
Next slide.

00:10:05.000 --> 00:10:34.000
So just, you know, it's like the Joker in Dark Knight. It's like, just expect chaos, guys. But you know, that's why we're doing these, is we want to make ourselves available. We want to be part of your community, right? To say like, hey, we know this stuff's all over the place. Let's have regular Q&As

00:10:34.000 --> 00:10:35.000
All right, great.

00:10:35.000 --> 00:10:36.000
We learn from our competitors, we learn from each other, right? So I really want to encourage you to continue to show up and give us feedback on these, because we want to make them as valuable as we can for all of you. All right, Alisa, I think that's all I got. Go ahead and take it away.

00:10:36.000 --> 00:10:37.000
So

00:10:37.000 --> 00:10:54.000
That was a great tee up for my section here. And I wanted to talk about five stories. You might have heard some of these. You might have seen some of these. But we're going to share the full text, the research behind each of these. I'm really going to focus on what I think you should all take away

00:10:54.000 --> 00:11:10.000
Whether you're a practitioner, whether you're an executive, why do you need to be aware of these stories? The first one that I'm going to cover is based on a pretty extensive research project that Trustpilot commissioned seer

00:11:10.000 --> 00:11:27.000
To perform. And the basic question was, does having a robust Trustpilot profile where you've got lots of reviews, lots of context about your business, does that correlate with an increase in AI search visibility? And we went into this thinking, yeah, I think it probably does

00:11:27.000 --> 00:11:43.000
I personally was a little bit staggered with how significant the findings were in terms of even a thin profile beating no profile by 52 percentage points in AI search visibility

00:11:43.000 --> 00:12:03.000
We looked at over 800,000 data points here. Nick Hagler performed this research on the SEER side and was heavily aided by John Lovett, our VP of analytics. So really extensive research here. The takeaway for me, you know, Trustpilot is great. This is not necessarily a go buy Trustpilot

00:12:03.000 --> 00:12:18.000
recommendation, but you know I wouldn't stop you if that's if that's the direction you're going. The takeaway for me is that reviews are such an important part of AI discoverability. And what's really interesting to me is

00:12:18.000 --> 00:12:25.000
I think us, if we're responsible for growth, if we're responsible for marketing, often within our organizations

00:12:25.000 --> 00:12:40.000
Reviews are not something that we own, right? So I think there's a few implications. There's a few recommendations we have, one of which is just audit your third party review presence. We've talked about this a lot. This is something we did in the last year

00:12:40.000 --> 00:13:00.000
Soliciting reviews was never something that we really emphasized, and our business is very, you know, one-to-one trust-based, and we get so much more business from the clients we work with, or clients moving on to a new organization, and saying, hey, I know and trust Seer

00:13:00.000 --> 00:13:15.000
We really under-indexed on the digital proof of that. And what we're finding is that's a miss in the AI era. But what many of you might find in your organizations is, okay, but review management, that's owned by customer experience. I can't touch that

00:13:15.000 --> 00:13:34.000
And, okay, how can I actually influence reviews? I think a lot of people treat this as a reactive, nice-to-have thing. You really have to lead a horse to water here, though. You have to meet your customers where they are and have a mechanism for how you're trying to solicit those reviews

00:13:34.000 --> 00:13:41.000
And also how you're acting on some of the criticisms that you might be getting.

00:13:41.000 --> 00:13:54.000
All right, the next story I want to talk about is something that Amazon launched and I have not seen a lot of people talk about this. But it strikes me as something that could be very, very significant for anybody in the e-commerce space.

00:13:54.000 --> 00:14:11.000
So the headline here is Discovery and purchase are moving inside the agent. Amazon just made it mainstream. So what Amazon launched is this feature called Alexa for Shopping, and it's a true agentic commerce experience. And what's most interesting to me

00:14:11.000 --> 00:14:28.000
Is Amazon has launched this to an audience where probably very few people even know what an agent is. All of a sudden, this is a tab in their Amazon app. And they're trying to attract clicks to it, right? It's got a nice little logo

00:14:28.000 --> 00:14:33.000
And it's not just about search. This is not just a search portal

00:14:33.000 --> 00:15:03.000
This is an action portal. This is an assistant portal. So it's not just about, okay, you know, help me find a gift for my mom. Here's what she likes. It's more so about, okay, can you help me find the best deal for this? Can you remind me when I might be getting low on a certain product

00:15:14.000 --> 00:15:22.000
born here that's going to either get your product in front of more people or get your competitor's products in front of more people.

00:15:22.000 --> 00:15:39.000
So there's implications here if you sell on Amazon. I also think even if you don't sell on Amazon, this is a really big step in getting agentic commerce in front of more people, in front of a broader audience. This is not just the early adopters who are using perplexity.

00:15:39.000 --> 00:15:46.000
for agentic tasks. This is, you know, non-technical people who are just kind of falling into this

00:15:46.000 --> 00:16:04.000
If it's a helpful feature, I think we're going to see it grow. I think we're going to see the expectations of consumers rise in terms of what a digital experience looks like. And it's not going to be too long until I'm thinking, okay, Nespresso, why can't you do this for me if Amazon can anticipate when I

00:16:04.000 --> 00:16:12.000
The next product or Amazon knows that it's summer and maybe now I want the little iced coffee pods. Why can't you do that?

00:16:12.000 --> 00:16:29.000
Okay, the third story I want to talk about is what was released in the Google I/O. And this was marked as the biggest change to the search bar in the past 25 years. This story got a lot of play. The New York Times covered it

00:16:29.000 --> 00:16:46.000
My takeaway here is that Google's minimalist roots are long gone. And we've been talking about this a lot, and we've kind of poked fun at Google for, you know, they start as this very simple experience with just the search bar. And then you search what you're looking for and you get the 10 blue links

00:16:46.000 --> 00:17:04.000
Everything is very clean. And when you compare that to what Yahoo looked like in 1999, it's like night and day. Well, now Google is really has kind of gone the Yahoo approach, and they are trying to create this choose your own adventure search experience

00:17:04.000 --> 00:17:07.000
There's a few things you need to know

00:17:07.000 --> 00:17:17.000
One is that now Gemini is powering the search bar. That's pretty interesting. There's more capabilities that Google can offer to its customers

00:17:17.000 --> 00:17:33.000
I think this really depends, though, on how users are going to embrace these new features. You know, you can essentially vibe code right in the Google search box. Does that mean your father is going to start vibe coding inside the search box? Maybe, maybe not

00:17:33.000 --> 00:17:48.000
So Google is kind of famous for this, right? For saying, here's a bunch of features, here's 10 things you can do, and then quietly they'll see who's engaging with what and start to sunset things. I think what my biggest takeaway with this announcement was

00:17:48.000 --> 00:17:58.000
is that they've reported that AI mode now has 1 billion of, I think it's weekly active users

00:17:58.000 --> 00:18:13.000
It's just really huge, right? And in part, it's because they've moved to a broader audience. You know, they released AI mode first to the US and India, and then they've added hundreds of additional countries, so they've gone global

00:18:13.000 --> 00:18:29.000
But this is really a search feature that we have to pay attention to if we're thinking about SEO the way we were even three years ago, we have to evolve that thinking. And yes, Google is still a powerhouse in the search field

00:18:29.000 --> 00:18:41.000
But that fragmentation that's happening to search, it's happening to Google as well, where there's all these different features that you as the user might adopt or might refuse to adopt.

00:18:41.000 --> 00:18:43.000
So what are the implications of this?

00:18:43.000 --> 00:18:58.000
Broadly, I think we've got to wait and see how your audience starts to adopt the capabilities that are released. You know, it's very compelling that now it's very easy to search by taking a photo of something right in the Google search bar

00:18:58.000 --> 00:19:14.000
remains to be seen if people are going to do that, though. And same thing with the vibe coding element of all of this. You know, there's a lot of alarmists who are like, now SAS is dead because you can build your software right in Google. Anybody who's vibe coded knows that's not exactly true

00:19:14.000 --> 00:19:33.000
But we do have to educate our executive stakeholders, because if they're thinking about Google using the same mental model that they used even three years ago, it is getting so much more complex, which means it's really easy to make the wrong investments or make the wrong decisions based on

00:19:33.000 --> 00:19:36.000
Where Google is today and where it's going

00:19:36.000 --> 00:19:40.000
This directly segues into the next story I want to talk about.

00:19:40.000 --> 00:19:55.000
This is something that Kevin Indig of growth memo released. If you don't follow Kevin, you've got to follow him. His his content is so, so strong. The research he provides is such a huge value to the industry as a whole

00:19:55.000 --> 00:20:11.000
I believe in the last episode of the signal, we talked about AI mode and his findings on how users were experiencing and engaging with AI mode. He recently released new research that told a very different story for AI overviews

00:20:11.000 --> 00:20:14.000
And so I want to just pause there because

00:20:14.000 --> 00:20:25.000
Your customers could be changing nothing about their behaviors. They're still going to Google and maybe they're even searching the way Google trained them to search

00:20:25.000 --> 00:20:40.000
But now the experience is so robust they can be dropped into AI mode. They're likely triggering AI overviews. And the way that they are interpreting and engaging that information is very different

00:20:40.000 --> 00:20:56.000
What we learned last month is AI mode, even though people say they don't trust AI search results, people actually trust AI search results, probably a little bit too much. And what he found is with a really robust audience panel

00:20:56.000 --> 00:21:08.000
When you ask a question like, what is the best X? AI moat says it's A, B, or C, something 74% of people just say, okay, whatever A was is what I'm going to go with

00:21:08.000 --> 00:21:25.000
That can be really problematic for a lot of different ways, namely because we know based on Rand's research and the SparkToro team's research, there's a lot of variability in who's ranked first, second, or third within these generated results. There's a lot of trust being given to the model here

00:21:25.000 --> 00:21:40.000
That maybe is not entirely earned that could impact trustworthiness and usability long term. But for now, it means if you're included in those AI mode, that first consideration set, that is hugely valuable

00:21:40.000 --> 00:21:43.000
Right? If you believe this data, it's hugely valuable.

00:21:43.000 --> 00:21:47.000
What we learned about AI overviews

00:21:47.000 --> 00:22:02.000
It's like a totally different mentality, and it's a lot more akin to like browsing on Netflix, because what he found is people are looking at every single citation, every single piece of information. They're scrolling down the page

00:22:02.000 --> 00:22:21.000
And for the first time in a long time, people are scrolling back up the page. They're kind of taking everything in and then thinking about, okay, out of everything I just saw, what is the right answer for me? So there's interesting implications there too. One of the things Kevin calls out that I totally agree with is how does this change

00:22:21.000 --> 00:22:39.000
How we optimize metadata. So if the job to be done is not necessarily just insert the right keywords to get the right visibility, you almost need to apply a meta description mentality to your page titles. So whatever your UVP is, whatever thing it is that differentiates you

00:22:39.000 --> 00:22:54.000
From all of the other impressions on the page, that's what you have to lean into. There's also implications to measurement because we look at Google rankings, it's easy to say, okay, here's what your visibility looks like on Google

00:22:54.000 --> 00:23:11.000
Increasingly, we've got to drop that language and mentality because it's like, what part of Google are we talking about? Because if I'm ranking in position one in the… or even in the consideration set in AI mode, that's hugely valuable. If I'm present in an AI overview citation.

00:23:11.000 --> 00:23:26.000
That's valuable too. It's worth an impression, but it's a lot more akin to almost a CPM model, where you're just trying to get your brand, your UVP, in front of your audience, and it probably is not going to lead to a visit to your website.

00:23:26.000 --> 00:23:44.000
All right, the last thing I want to cover is, again, research from Kevin Indyk. I'm not kidding when I say he constantly changes the way I think about this stuff, but also research from one of our own team members, Matt Boxbom. And there's two different stories here that I'm kind of combining

00:23:44.000 --> 00:23:49.000
Citation visibility, first of all, is

00:23:49.000 --> 00:24:04.000
extremely chaotic. And what I mean by that is, I think a lot of us are chasing citations because it's the familiar thing, it's the leading indicator, it's akin to Google visibility, the old blue link world

00:24:04.000 --> 00:24:20.000
But what increasingly we are finding is what's visible on ChatGPT is very different than what's visible on Perplexity versus what's visible on Claude versus what's visible and being referenced on Google's AI overviews

00:24:20.000 --> 00:24:31.000
Now, the part that Matt contributed here is he found a ridiculous surge in Reddit's influence on Google's AI overviews. So that's interesting

00:24:31.000 --> 00:24:36.000
But by and large, I think my takeaway for this information

00:24:36.000 --> 00:24:52.000
We've got a really, really think critically about the role of citations as a success metric. And we can't chase after them because it's constantly going to fluctuate as new models are released, even if you're only going after ChatGPT

00:24:52.000 --> 00:25:02.000
As new models are released, they're changing everything from the grounding queries that are being used, and that informs which citations are being pulled

00:25:02.000 --> 00:25:18.000
So I think my my message here, and what I get really excited about is let's transcend the citation data, and let's make sure it's correlated. We're not going off on a, you know an island that has no impact to AI search visibility

00:25:18.000 --> 00:25:35.000
But what is our customer engaging with? And if you have a customer base who is engaging with Reddit, at least an average amount, certainly an above average amount, that to me is a really, really good platform to invest more time and energy into.

00:25:35.000 --> 00:25:52.000
If you currently work with us at SEER, and we haven't offered this already, we've got an alpha build going where we're trying to get more prescriptive with exactly what do you do within Reddit because it's not a box to check, right? It's a whole ecosystem. It requires resourcing. It requires a strategy.

00:25:52.000 --> 00:26:05.000
But if done well, I do believe this can be a cheat code that really helps you fuel the increased velocity you need in order to be successful in digital marketing today.

00:26:05.000 --> 00:26:24.000
So what are the implications of this? Again, it's on measurement, but really it's on Reddit as well. If we know that Reddit is such an outsized, an outsized citation source for Google's AI overviews, the first thing I would say is we're not guaranteeing that even three months from now, that's going to be true

00:26:24.000 --> 00:26:40.000
But you can use data like SparkTorio is great for this to say, okay, where's my audience spending time? Then agnostic of where the citations are coming from. How do I make sure my story is consistent, it's being told in the way it needs to be told on these platforms

00:26:40.000 --> 00:26:58.000
To the extent that's possible, you know, you're not trying to get into arguments on the internet with strangers about is your brand the best? But there may be objectively false statements about your brand that would behoove you to have some kind of resourcing set up to address and to engage with folks

00:26:58.000 --> 00:27:01.000
In a helpful, supportive way

00:27:01.000 --> 00:27:09.000
All right. So again, you'll get all of these resources after the fact, but here are the things we really think you should do this month.

00:27:09.000 --> 00:27:27.000
Audit your third party review presence and start to put together a plan for your zero profile platforms, those profiles that might exist, but you've got one or two reviews on them. We're guilty of this as well. We're going to share as much as we can in terms of how we're going about fixing that

00:27:27.000 --> 00:27:44.000
Pressure test your category on Amazon's Alexa for shopping feature. Even if you're not selling on Amazon, if you have, if there are products related to your products and services, I think it behooves you to start to see who's winning in this space, who are the major players

00:27:44.000 --> 00:27:57.000
Getting added as a preferred source could be an interesting, very minor cheat code to try to encourage Google to show you to the audience you've already built, you've already gained trust with

00:27:57.000 --> 00:28:12.000
Splitting AI overviews in AI mode into two separate optimization tracks with their own metrics, something we're actively working on at SEER, and it starts with, do we have access to all the right data? Are we tracking? Are we collecting as much data as we can

00:28:12.000 --> 00:28:30.000
As Will mentioned, yes, now we are able to get some degree of data from Search Console. We will get increasingly more, I hope and believe. But right now we're not getting ranking data, and that's really what we're craving. We want to be able to take a benchmark of what our visibility looks like now

00:28:30.000 --> 00:28:40.000
So we'll have to use impression data as a proxy to that in addition to third-party metrics. You're also going to want to measure your AI visibility per engine

00:28:40.000 --> 00:28:55.000
In terms of citations, and start to think about standing up a real Reddit presence. And again, this can't be, you know something you task your part-time intern with. It can't be something that is the 10th thing on somebody's, you know, 9-point to-do list

00:28:55.000 --> 00:29:11.000
This requires real resourcing. Subreddit communities are their own individual ecosystems. There are huge opportunities. I firmly believe, for brands to engage in a respectful way that abides by the rules that moderators have set

00:29:11.000 --> 00:29:16.000
But still, we'll set the story straight about your brand and the truths about your brand

00:29:16.000 --> 00:29:28.000
And there's a way to do it not so great. And I think if you're not investing the right time and the right resourcing in Reddit as a channel, there are going to be major downstream effects of that

00:29:28.000 --> 00:29:44.000
The last thing we didn't share this explicitly, but Will did update the AI maturity curve resource. I think trying to push your team to be as AI forward as possible, we are going to continue to hammer this

00:29:44.000 --> 00:30:02.000
I have no idea how you could be successful in digital in 2026 without moving at such a ridiculously fast velocity that is just not possible to do if you're not adopting AI methodologies, AI tools, AI platforms

00:30:02.000 --> 00:30:16.000
All right. So that concludes what we've prepared for you all. Now what we want to do is we want to check out some check out some Q&A questions. So let me pop this open.

00:30:16.000 --> 00:30:31.000
First thing… okay, so the first thing we're going to — oh, you know what? We've got some answered already. Cool. So the first question we have, and Will, I'm going to tee this up for you. So there's been lots of chatter on non-commodity content lately

00:30:31.000 --> 00:30:44.000
If you have nothing personal to share, is it better to share nothing versus publishing AI-generated content?

00:30:44.000 --> 00:30:49.000
It's a good one. My thinking is

00:30:49.000 --> 00:30:55.000
If you don't feel like you've got anything to share

00:30:55.000 --> 00:30:58.000
That is not commoditized

00:30:58.000 --> 00:31:12.000
I think you've got to take that all the way back up to the top of the organization and say, what company have we built where everything about us could be copy and pasted onto a competitor's site and we have no differentiation

00:31:12.000 --> 00:31:19.000
That's where I would probably start. I would go all the way back up to the top. What I also think

00:31:19.000 --> 00:31:34.000
is there are points of view. Shout out to one of our alumni. I met her for coffee recently, and Greta, she said to me, she goes, I just scanned my company's internal Slack and realized we've got all kinds of points of view on stuff

00:31:34.000 --> 00:31:50.000
That never made it to the website, and I was like, that's smart, because she's like, I'm getting most of my content pre-written, 30 to 40% done just by mining my internal Slack using Claude's connector to Slack, and being like, great, put all these points of view together around this topic, and boom

00:31:50.000 --> 00:32:06.000
So I think when we go to the content marketing route, we end up being like, I don't have the content. But yet, if you go the scrappy AI route, you're like, there are places where people are sharing points of view. We just never had a really good pipeline of push it from internal to external

00:32:06.000 --> 00:32:16.000
Because in an SEO world, you could just do your SEMrush research, find what keywords people are searching for, stick it on the bottom of a page and win.

00:32:16.000 --> 00:32:19.000
Good stuff. That's such a good tip from Greta. That's great.

00:32:19.000 --> 00:32:35.000
Alright, next question. This is a good one. So, and I love that you all are digging into the the the research and what we're putting out there. So someone commented, Sierra put up a testimonials page, and you all can see this at sea interactive.com slash testimonials

00:32:35.000 --> 00:32:49.000
If we have reviews on other sites already, do you think there would be value in collecting the best ones on a page, on our site, like you did here? So I think this is such a good segue to what Will teed up with

00:32:49.000 --> 00:33:07.000
Because in my mind, there is no reason not to. This is such a worthwhile test, and even if it does absolutely nothing for your traffic, for your business, for 3 months or 6 months, first of all, I would find that hard to believe, right? That no prospect has stumbled upon

00:33:07.000 --> 00:33:23.000
This page, and it's not at least a little bit of a motivation that's like, okay, I know these brands, they trust Sears, so now I trust Sears. That in itself is worth a test. But also, if we're not testing things, we're not trying things, and we don't know what's successful and not successful

00:33:23.000 --> 00:33:38.000
And I can tell you firsthand, it is a huge pain in the ass to try to get reviews on third party websites. I bet a lot of you understand that, right? So if we can take some of the burden off of that and control the sphere of influence a bit more with something we own

00:33:38.000 --> 00:33:40.000
all the more reason to do it.

00:33:40.000 --> 00:33:46.000
Good, Will

00:33:46.000 --> 00:33:47.000
Nice.

00:33:47.000 --> 00:33:48.000
No, no, I'm like, go ahead, girl. I'm like, I was just going to say preach, like go for it. I mean you yeah yeah that's exactly what I was thinking.

00:33:48.000 --> 00:34:03.000
Nice, nice. All right. Somebody, I'll do a quick one here. Two questions, same theme. Tips to track in AI mode. You know, we're tracking in AI overviews, but what else, what can be done here

00:34:03.000 --> 00:34:18.000
It's all about buying data right now, which in some some cases it's difficult. You got to go back and ask for more budget. If you've got the infrastructure built to be able to take all of your data and synthesize it in a way that makes sense

00:34:18.000 --> 00:34:35.000
There's so much value that can be had with a system like that and the return is going to continue to compound. But even for us at SEER, we've got a really robust infrastructure. We're grappling with these questions now. Do we buy the data from data for SEO? It's a really good source. It's affordable

00:34:35.000 --> 00:34:52.000
If you're already tracking for Geo or AI search, you probably have the capabilities to buy this data from that vendor. So I would say, you know, find the cheapest data possible and try to beat people up on price as much as possible. I hate to say that, but also

00:34:52.000 --> 00:35:08.000
In this space, you know, there's a lot of overage charging happening, and so don't take the first offer, especially with your geo vendor. But getting this data is the important thing. Even if you can only afford a benchmark of the data

00:35:08.000 --> 00:35:14.000
I think it's going to be really important to say, okay, when we look back 6, 12 months

00:35:14.000 --> 00:35:31.000
What does our traffic look like today, and what did our visibility in these different Google surfaces look like? And if we want to be able to intelligently answer questions from an executive or from a board, we have to be able to at least have the contours of this story to be able to tell it

00:35:31.000 --> 00:35:35.000
Hey, Alisa. One of the things that I was thinking of or two things is

00:35:35.000 --> 00:35:45.000
One of the frameworks that somebody, and again, another alumni I was chatting with, it was Adam Nelson who mentioned to me that, like, when he thinks of, like, analyzing this stuff

00:35:45.000 --> 00:35:52.000
He likes to show his clients and the C-levels the black box of analyzing today

00:35:52.000 --> 00:35:55.000
And being like, my job is to try to shrink the black box

00:35:55.000 --> 00:36:07.000
And I was like, that's really smart, right? Because to me, it's like, all these things are not trackable. This is def… you know, Rand comes out with a study and goes, you have to run the prompt a thousand times for it to be the same one. You're like, that's good, but

00:36:07.000 --> 00:36:26.000
Even if I run it 10 times, that shinks my black box, so I can give an answer to my manager or whatever, right? So that is one of the things on one of the little devices I'm starting to use is like, look, everybody's dealing with the same black box. The idea is, is how do I help to shrink that black box using the data that I can get my hands on, whether I'm buying it or not

00:36:26.000 --> 00:36:39.000
That was one of the things. And another thing that somebody said to me, and it was so basic this week, but I was like, man, oh, thank God I got smart friends. So Jason Thompson reminded me of something

00:36:39.000 --> 00:36:43.000
ChatGPT traffic, when it comes in as ChatGPT

00:36:43.000 --> 00:36:51.000
Take a look at whether or not it has a high propensity of desktop search. And I'm like, why? And he's like, because if people are using ChatGPT in an app.

00:36:51.000 --> 00:37:09.000
Apps have different referrals. Make sure you're not thinking that, oh, this is my traffic from ChatGPT. If you have a highly mobile audience, and they're in the mobile app for ChatGPT, guess what's not showing up in your analytics, which could make you think the numbers are smaller? So yeah, that's just another thing, that when he said that to me, it was so basic

00:37:09.000 --> 00:37:14.000
But it was also like so important that now I've got that seed planted in my head when I'm looking at data and metrics as well

00:37:14.000 --> 00:37:30.000
That's a really good call-out. The other one I'll call out is, you know, and again, the search experience, these surfaces are so absurdly fragmented relative to a year ago, but it's like night and day compared to even 5 years ago

00:37:30.000 --> 00:37:45.000
Something I really like to do, and I would encourage anybody who works in search to do is when you're doing personal searches or searches on behalf of your own brand, watch the thinking, you know, click that dropdown on Claude so you can see what is being

00:37:45.000 --> 00:38:02.000
You know, the model will kind of, I hate to personify these things, but the model will go through the steps it is taking to reason with itself to say what is the best answer. Increasingly, I see when I'm on my phone, I see it say, okay, and Elisa is using a mobile device

00:38:02.000 --> 00:38:18.000
There is something different it is considering with the the results and the generated response when I'm on mobile versus when I'm on desktop. I don't yet know what that is. But even something like that really, I think is it illustrates

00:38:18.000 --> 00:38:25.000
How many different variables there are, and just how complex this space is becoming.

00:38:25.000 --> 00:38:29.000
And we have this desire to say, okay, what's the one metric

00:38:29.000 --> 00:38:49.000
And yes, that's what we need to work towards, but also this understanding of like at the end of the day, I don't know if that metric comes from an analytics platform. You know the what we might be looking at a couple years from now is just it's all about word of mouth. It's all about getting in front of our audience and increasing our presence in that consideration set

00:38:49.000 --> 00:38:58.000
And that is the extent by which you can really master online discovery.

00:38:58.000 --> 00:39:17.000
All right, we have, okay, staying on this theme, Will, I'll tee this one up for you. Google's AI mode appears to collect and surface entity data faster than any other platform for a specific named entity, whether that's a person, a brand, or a business, what are the most effective methods right now to increase visibility and

00:39:17.000 --> 00:39:25.000
authority within Google's AI mode, specifically versus traditional SEO. And do we even know the answer?

00:39:25.000 --> 00:39:31.000
I tend to look at human behaviors

00:39:31.000 --> 00:39:36.000
And assume that what Google is trying to do is come up with algorithmic ways

00:39:36.000 --> 00:39:52.000
To replicate how humans would make decisions in the real world. So I don't go deep on like entities. Like, so if you watch me, like, I don't get deep on entities, I don't get deep on EEAT, because I'm like, well, now that we've just named it, like, isn't this just about like trusting the source

00:39:52.000 --> 00:40:08.000
Like, that's… that's just marketing, right? So, I don't have a specific belief around entities and all of that. What I will say is my own tests, our tests are on our website, were that, it seems to me

00:40:08.000 --> 00:40:13.000
broadly across AI, so this is less about Google AI over… it's like

00:40:13.000 --> 00:40:23.000
I'm seeing, when you hit the little dropdown sometimes and I ask the question, tell me about Sierra Interactive, which I do every week across a bunch of models manually, so I can see things

00:40:23.000 --> 00:40:25.000
It looks for differentiators

00:40:25.000 --> 00:40:29.000
So, like, I don't know if those differentiators could be entities. For instance.

00:40:29.000 --> 00:40:35.000
B Corp might be an entity. It probably is, right?

00:40:35.000 --> 00:40:40.000
So maybe when you ask, tell me about SEER Interactive

00:40:40.000 --> 00:40:53.000
And you're like, okay, awards and this and that, you can kind of map you could map the grounding queries from Bing because they're giving you those or the reasoning steps

00:40:53.000 --> 00:41:00.000
I wonder if you could map those to entities and then run a test to say, like, as I started to say these things more.

00:41:00.000 --> 00:41:11.000
Do I see changes in my answers in all AI? I would say not just specifically Google AI overviews or Google AI mode. That's the way I think about it. What about you, Alisa? What's your thoughts on that?

00:41:11.000 --> 00:41:29.000
Yeah, I mean, I'm very aligned. I think it's so hard to get to… There's a desire to figure out what exactly is driving all of these engines, all of these models. And I think increasingly that focusing there is just going to lead to a lot of time wasted and a lot of frustration

00:41:29.000 --> 00:41:45.000
As opposed to focusing on where your audience is engaging today and trying to take the data as an input, but don't chase the data to the nth degree, because by the time you crack the case, a new model is going to be out or your audience is going to have shifted to another one

00:41:45.000 --> 00:42:03.000
So that's the kind of high velocity work. I don't think is worth not to say the question is not worth asking because it absolutely is. But I think we've got to bring it up a few levels of, okay, how does this? Does it influence the strategy? Does it influence the tactics that that we're actually deploying and testing

00:42:03.000 --> 00:42:04.000
Agreed.

00:42:04.000 --> 00:42:21.000
Nice. All right, somebody asked a very good question. They said, does the primary review source matter by industry? Trustpilot isn't very important traditionally in travel, especially for one like ours with over 125 locations and just a single Trustpilot account

00:42:21.000 --> 00:42:38.000
Absolutely. You should look at where your citation data is coming from. We've been doing some research at SEER. I'm really excited about, and the idea is that yes, these models are trained on a specific set of training data. And yes, these models are accessing web search

00:42:38.000 --> 00:42:50.000
Through grounding queries or fan-out queries and pulling more information. But there's this middle step, this invisible criteria that's kind of baked into the reasoning of any of these reasoning models

00:42:50.000 --> 00:43:12.000
That is going to tee up for the model, what is the best thing to do to answer this user's query? And I would bet I haven't done the research explicitly, but I would bet that that invisible criteria is going to say things like, what is… let's check TripAdvisor, let's check Google reviews, or whatever the most relevant ones are for your industry

00:43:12.000 --> 00:43:14.000
Now, is it worth a test?

00:43:14.000 --> 00:43:31.000
could be worth a test if it's a very low effort test. If it was something like, hey, Trustpilot, I don't think it really matters for my audience, then, then why do it, right? Because increasingly, there are so many tactics to test, so much to

00:43:31.000 --> 00:43:51.000
to try to execute upon. I think you always have to be thinking about that Venn diagram of what does the data say? What is the… both the third party research, like what we're providing at SEER, as well as your own citation data, and what do I know to be true about my audience and the things in that overlap? That's what you really have to over index on, in my opinion.

00:43:51.000 --> 00:43:57.000
One thing I'll add to that, David. So

00:43:57.000 --> 00:44:01.000
The thing that I would look for, so about

00:44:01.000 --> 00:44:02.000
3 or 4 months ago

00:44:02.000 --> 00:44:22.000
And I can send you this link. Me and Dave go way back. We're, like, college buddies. So, he has my number and can message me. I'll send you the research. But Chris Long dropped something that we also saw at the same time, which is, at some point, ChatGPT, to try to get spam or, like, lower quality answers out of its answers

00:44:22.000 --> 00:44:26.000
Suddenly it started saying, hey, if you're looking for reviews

00:44:26.000 --> 00:44:41.000
for a certain type of thing, or if we believe that reviews are a good part of an answer, it started running what I call a site colon search. So what was interesting is you saw the AI take a reasoning step to say, hey, if you're traveling

00:44:41.000 --> 00:44:42.000
What sites

00:44:42.000 --> 00:44:58.000
are highly likely to have reviews, and then it would run all of the grounding queries against the site colon search for those individual domains to get rid of the low-quality stuff that was happening before. His example is B2B

00:44:58.000 --> 00:45:13.000
But if I were to say to everyone on this call, hey, what's a B2B review site? Many of you might say like, oh, G2, Gartner's Magic Quadrant, right? So what's happening is the AI, because they saw this flood of spam coming in

00:45:13.000 --> 00:45:29.000
They said, if we run a site colon search to only search G2, to only search Trustpilot, and to only search Gartner for these kinds of prompts, then we can avoid a lot of that low-quality stuff people are building. So I would do the same thing in travel

00:45:29.000 --> 00:45:38.000
I would look for what's called the query fan-outs, and see if you… for ChatGPT specifically, this is the one that you see it the most on, I think it's 5.

00:45:38.000 --> 00:45:53.000
Whatever. But if you see it through a site colon search for a domain, I would be like, okay, then those are the domains that are most important, and shoot me an email, Dave, I can record you a quick video of how what that looks like if it gets a little arduous.

00:45:53.000 --> 00:45:54.000
Awesome.

00:45:54.000 --> 00:46:10.000
Alright, I want to do our best to get folks out of here with a few extra minutes so you can get outside, you can decompress before your next meeting, you can reach out to a loved one, whatever you'd like to do. So we're gonna do two more quick ones. I'm going to take the first, and then Will, I'll tee one up for you to close us

00:46:10.000 --> 00:46:25.000
So a question that came in was, do you recommend seeding conversations on Reddit related to your brand or setting up a subreddit for your company, or just trying to join existing conversations? This is a fantastic question, and

00:46:25.000 --> 00:46:40.000
I don't do this to be self-promotional, but we've been building out a system at SEER that we're really excited about that I think can help you think about Reddit in the lens of a maturity model where you're not trying to go from 0 to 100

00:46:40.000 --> 00:46:45.000
You're treading lightly because you may have a CMO or somebody in charge of your brand who's like, don't touch Reddit. I don't

00:46:45.000 --> 00:47:00.000
want to get it's a quagmire. Don't do anything with it. I'm too afraid of it. But there are baby steps that I think you can take picking out specific subreddit communities that maybe aren't the biggest in your space, but can help your profile build some credibility

00:47:00.000 --> 00:47:19.000
Clarifying any statements that are objectively false, things that are genuinely helpful. The goal is not to go on there, and whenever somebody says, hey, what's the best geo agency? Will and I are like, it's seer. Don't think about anybody else. That's not helpful. The goal is like anything in marketing. We want to add value. We want to help our audience

00:47:19.000 --> 00:47:37.000
get to the next stage in their journey. And I think there's some really compelling ways we're working on. So keep an eye out if you're not signed up for our newsletter yet, make sure you are. Maybe Hannah can drop a link in the in the chat so that as we release more information, we love to share as much as possible with you all.

00:47:37.000 --> 00:47:53.000
So what we're going to end on is, Will, have you seen any correlation in your writing style and tone with visibility in AI search? Think is AP style or more buttoned up and professional more impactful than more casual

00:47:53.000 --> 00:47:55.000
conversation.

00:47:55.000 --> 00:48:05.000
It's an interesting question, and I think I have a pretty good test on this because my style of writing is usually full of

00:48:05.000 --> 00:48:10.000
Wrap references, swear words, and typos, all of which are deemed unprofessional

00:48:10.000 --> 00:48:13.000
I don't see that

00:48:13.000 --> 00:48:24.000
decreasing my ability to show up when people use AI for the things that I write. I think the real important thing I'd leave you with on this question

00:48:24.000 --> 00:48:33.000
is spicy language, let's just say language that comes from, like, passion is not always professional

00:48:33.000 --> 00:48:44.000
And you got to balance which of these is more important to me. I would rather connect with people in a way that makes them feel like this webinar is worth showing up for

00:48:44.000 --> 00:49:06.000
Which sometimes I need to be spicy, I need to have hot takes. I need not need to. It's who I actually am. And they go, oh, I'm gonna show up for an hour and listen to what these guys have to say over getting a single click, buttoning it all up so it sounded more professional so I could get one or two more pieces of visibility. But then when you get to my website, it sounds like everybody else's stuff, which means

00:49:06.000 --> 00:49:13.000
You know, if 30 people all were doing the same webinar right now, and we all sounded the same, you don't get 30x the amount of time to watch them all.

00:49:13.000 --> 00:49:29.000
So I want to be grateful, one, that you chose to spend this time with us and drop the question, but also I want to make sure that I'm balancing AI visibility with and what's right for AI with like what makes people say like, I want to show up to this next thing

00:49:29.000 --> 00:49:33.000
And rarely does anybody say, you know what makes me want to show up to a presenter at a conference?

00:49:33.000 --> 00:49:42.000
They use the AP style. So you've almost got to balance those two things against each other, and I think to have the ideal outcome.

00:49:42.000 --> 00:49:44.000
Awesome.

00:49:44.000 --> 00:50:01.000
All right, that's it. We're going to wrap up here. We'll be in touch with all of the resources we talked about, all the links. Highly recommend you give Garrett Sussman a follow. You give Kevin Indig a follow, and make sure you sign up for the newsletter if you want more of this type of information.

00:50:01.000 --> 00:50:02.000
Thank you all for coming.

00:50:02.000 --> 00:50:05.000
Have