How Gumshoe Helps Brands Understand What AI Thinks About Them?

Are you a marketer trying to understand how AI search is changing brand discovery? Join us for a conversation with Todd Sawicki, CEO and Co-Founder of Gumshoe AI, as he breaks down how AI platforms recommend products, why “knowledge authority” beats domain authority, and what you can do today to boost your brand’s visibility in ChatGPT, Perplexity, and beyond.
#AISearch #MarketingStrategy #BrandVisibility

Join Emanuel, our host from “How About Some Marketing?”, as he navigates this insightful webinar with Todd Sawicki.
Subscribe to the newsletter 👉 https://www.howaboutsomemarketing.com/

Todd reveals how Gumshoe AI helps brands understand what AI says about them and what to do about it. From AI personas and content formats to why small brands can get surfaced in as little as 2-3 weeks, this session is packed with actionable strategies.

Todd Sawicki is the CEO and Co-Founder of Gumshoe AI, helping marketers navigate the shift from traditional search to AI-powered discovery. Previously CEO of Zemanta and CRO of Cheezburger Network, his career spans Disney, startups, performance marketing, and programmatic advertising.

Key Takeaways

• AI search is replacing the traditional search “popularity contest.” Instead of rewarding the most backlinks, AI looks for canonically authoritative and factually accurate content, no matter how deep it sits in the index.
• Think of AI as a shoe store salesperson. Just like sales reps get trained by brands each quarter, AI needs to be trained and informed about your products so it can recommend you accurately to the right customers.
• AI personalizes its answers using customer personas. The moment a user registers with an email, AI builds a profile and tailors recommendations based on who it thinks they are. Different personas get very different brand recommendations.
• Your brand website matters more than ever. 30-40% of URLs cited by AI are business pages (PDPs, FAQs, blog posts). AI sees you as the most canonical source of information about your own products.
• 50-90% of AI citations come from page 3 and beyond in traditional Google search. AI digs deep for the right information, not just the most popular. This levels the playing field for smaller brands.
• Content recency is critical. The average age of a URL cited by AI is 86 days, and that number is shrinking 10-15% every quarter. Fresh content gets surfaced faster than ever.
• Small brands can see results in 2-3 weeks. Unlike traditional SEO that took 12-18 months, AI search rewards new, authoritative content quickly.
• Knowledge authority beats domain authority. AI cares more about who wrote the content and their expertise than the overall popularity of the domain. Recruiting expert “author panels” to write on your blog can boost your visibility.
• AI prefers specific content formats. FAQs, knowledge base articles, how-to guides, and long-form content perform best. Academic-style writing with cited sources and references is favored.
• Upload everything to YouTube and turn on transcripts. AI loves long-form video content and reads transcripts rather than performing speech recognition. This is a quick win for any brand.
• Celebrity endorsements carry less weight in AI search. AI values authoritative and knowledgeable figures in a category over popular celebrities.
• The future is a real-time conversation between brands and AI. As the content cycle shortens, marketers will need to create content that responds to AI-driven traffic in near real time.

Show Notes

In this episode of How About Some Marketing?, host Emanuel Petrescu sits down with Todd Sawicki, CEO and Co-Founder of Gumshoe AI, for a deep dive into how AI search is transforming brand discovery and what marketers can do about it.

Todd explains how AI platforms like ChatGPT, Perplexity, and Claude are taking over the “consideration” phase of the marketing funnel, personalizing answers based on user profiles, and recommending brands based on factual authority rather than popularity. Using Lululemon as a live example, Todd demonstrates how Gumshoe helps brands uncover what AI says about them, which customer personas are being targeted, and where the gaps and opportunities are.

The conversation covers the three essential questions every marketer should answer: what to say to AI, how to say it (formats and language AI prefers), and where to say it (your own site, YouTube, Reddit, Quora, and more). Todd also shares practical strategies including recruiting long-tail expert writers, optimizing YouTube transcripts, and leveraging the 86-day content cycle to get surfaced fast.

Topics Covered:

• What AI search is and how it differs from traditional SEO
• The shoe store analogy for understanding AI recommendations
• How AI builds user profiles and personalizes answers
• Lululemon case study: persona-based recommendation rates
• Why your brand website is now your most important asset
• Deep links: why page 3+ content wins in AI search
• Knowledge authority vs. domain authority
• Content formats AI prefers (FAQs, how-to guides, long-form)
• YouTube and transcript optimization
• How small brands can compete and win
• The role of Reddit, Quora, and third-party platforms
• Celebrity endorsements in AI search
• The future of paid ads in AI platforms
• Live demo of the Gumshoe platform

Episode Transcript:

[00:00:00] Emanuel: [00:00:10] Okay. And we are live. Awesome. Hi everyone, and welcome to another episode of How About [00:00:20] Some Marketing? Webinars. It’s my pleasure and privilege to have a special guest today. Todd, how do you say your last name? I don’t wanna…

[00:00:27] Todd: it’s okay. So Wiki, so my nickname is [00:00:30] So Wikipedia, so it, because it’s a good rhymes or sounds like.

[00:00:35] So that’s how you know how to stay. It. Sawiki rhymes with Wikipedia. [00:00:40]
[00:00:40] Emanuel: So we keep, it. Yeah. But thank you so much.

[00:00:42] Thank you so much for being here. As said, it’s a pleasure and a privilege to have Todd Sawiki – pedia as [00:00:50] a guest today, and we’ll soon learn what Todd does and what his tool that he created does. But before we jump right in, I wouldn’t be much of a marketer. If I wanna recommend everyone [00:01:00] who’s with us live here or who’s watching the recording too, go to howaboutsomemarketing.com and sign up for the newsletter that we have there [00:01:10] for that’s where you’ll find the information about previous recordings, upcoming webinars, and many other useful information because How About Some Marketing? is a place [00:01:20] for marketers to get better at their… wait for it… marketing.

[00:01:25] We cover paid ads, social media, organic, and of [00:01:30] course AI, which is like in the middle of everything right now.

[00:01:34] So it’s my pleasure and privilege to introduce Todd Sawiki, founder of [00:01:40] GumshoeAI. He’ll be, showcasing the tool, how we can help marketers with their ai. I like to call them [00:01:50] citation. This is a hype word, but it’s not really the right word. Before I’ll pass the virtual stage and the screen, I do have one question and I’ll ask [00:02:00] you first to do a short introduction of yourself.

[00:02:02] Only the best things, of course, and the question that’s on everyone lips essentially, what is [00:02:10] Gumshoe? The name, what does it mean, what’s the history behind it? Because I don’t know
[00:02:16] Todd: It is a, great question. I’ll [00:02:20] start with what is Gumshoe and where does the name come from, and then getting a little bit into my background.
[00:02:26] Gumshoe came from, we were thinking about the name [00:02:30] and we knew for Gumshoe we’d already started developing the product and knew it was about helping companies understand what AI thinks about them. And so it’s a research [00:02:40] tool, right? It, helps you learn things. And so we were brainstorming names around that, like looking into things, [00:02:50] researching and then and, investigating.

[00:02:53] And that was the key word. And one of the members of the team is a Film [00:03:00] Noir buff. And what Gumshoe is, it’s a reference, it’s a nickname for private detectives from the 1940s. [00:03:10] And so you would be called, if you were a private detective, they used to call you like a dick, right? because detective and so another name they used to name call them was [00:03:20] Gumshoe because they used to wear, they were one of the first people or types of people to wear rubber sold shoes. [00:03:30] And back then rubber was referred to as gum.

So gum shoes referred to rubber sold shoes. And so [00:03:40] private detectives were one of the, so they could sneak around quietly because before that you had hard sold shoes, right? Wooden heels that would click, As you walked and rubber [00:03:50] sold shoes were quiet. So you could sneak around. And so in you’ll see references to people saying, I need a private detective. No, I need a gum shoe. [00:04:00] In film noir films from the 1940s.

[00:04:04] So he rep, so that came up as a name and it was like, as you’ve named things as a marketer [00:04:10] you, make sure the domain’s available and was available. And so we really liked that as a fun, playful name that really I think fit what we’re trying to do in terms of helping [00:04:20] brands understand what how they investigate themselves effectively.
[00:04:24] Emanuel: A name that actually makes sense.
[00:04:26] Todd: We, it was, we, when I heard it, I had never heard it before. I was like, oh, because I heard [00:04:30] gum. We, all heard gum shoe. We thought, does that mean like gum? Like literally chewing gum on the bottom of your shoes. That’s a weird kind of name. But when it was explained to us, we said, oh, that’s really cool.
[00:04:39] And [00:04:40] so that’s the story in background of how we named Gumshoe, what we did.
[00:04:44] Emanuel: And as our audience will see, it makes sense and it ties really well with [00:04:50] what the tool actually does.
[00:04:54] A short introduction about yourself before we dive.
[00:04:57] Todd: For sure. I have [00:05:00] been doing marketing related startups for, gosh, 20, 25 years, and building software and tech to help marketers do their jobs.
[00:05:08] It really has [00:05:10] often been around this idea of customer acquisition and tools that help with that. And Gumshoe kind of fundamentally we, only care about AI or AI search as it [00:05:20] will ultimately help us find and attract customers. And so previous to that I built, I worked at a number of startups, probably six or seven now.
[00:05:29] One of them [00:05:30] was, not so directly customer related, which was an early meme publisher called I Cheezburger, one of the generators of photos cats. And so [00:05:40] if you literally get silly cat photos from your mom, I’m one of the people responsible for that. Sorry. And so that was [00:05:50] early on in the internet helping create ekins cheeseburger and low cats.
[00:05:54] Then have built DSPs. I built things in like the retail media space. I built [00:06:00] tools in e-commerce. And so it, this is really part of the, Gumshoe is part of that journey of building tools to help marketers, right? To help marketers acquire. I was [00:06:10] in a toolbar startup 20 years ago. That was one of my first ones.
[00:06:13] Then ended up at I Icon Cheezburger, then built a DSP. And so then it was in the e-commerce space helping [00:06:20] with, merchant ads and advertising around the Shopify ecosystem. And then ultimately that led to Gumshoe. And Gumshoe really was as [00:06:30] the, my e-commerce startup was winding down, one of the things that was happening was the rise of AI and AI search, like ChatGPT had just [00:06:40] launched in 2022.
[00:06:41] This is now summer of 2023, and you start hearing stories around marketers. Starting to see the decline of [00:06:50] traditional search and they’re starting to see this new thing, AI search. And so that was, and marketers desperately wanting to know what’s going [00:07:00] on was really what led to starting Gumshoe.
[00:07:05] Emanuel: What an amazing Journey.
[00:07:07] Then we learned what gum shoe [00:07:10] means, what’s his background, what does gum shoe do and how does it work essentially. And I do wanna, I usually take notes, take a [00:07:20] lot of notes manual notes as well that I like to reference them later on. I encourage everyone to look out for that.
[00:07:29] And [00:07:30] yep, let’s get jump into it. Do you want me to put the slides on? Yeah, let’s go. We, a couple of them.
[00:07:35] Todd: Let’s, we we and, I’ll talk about what AI search is [00:07:40] first and and then how what does gum shoot then fit in and, help with that. And of course for those who are watching live, we’ll [00:07:50] be playing with the platform live as well.
[00:07:52] And, happy to even look at if there’s a companies or that you’re involved in, will take suggestions from the audience in terms of what we [00:08:00] can demo and show with through Gumshoe. But first, I’ll explain how, what AI search is, and then ultimately it’ll help explain how, what, how Gumshoe works and [00:08:10] what it does, and how it helps marketers.
[00:08:12] Emanuel: You couldn’t ask for a better guest thought. Thank you so much.
[00:08:17] Todd: No problem.
[00:08:18] Emanuel: You are on [00:08:20] screen.
[00:08:20] Todd: I see it. Yes. Great. And as, Emanuel said, and, I’ll say again, right? The idea of Gumshoe as a platform [00:08:30] is, we help brands understand what ai says about them, and then what to do about it.
[00:08:36] That’s fundamentally I think the simplest way of, thinking about [00:08:40] that and the way we think about AI search is, because we get this, we get asked this a lot. Is this something that only big companies should be thinking about? Is this something that [00:08:50] everyone, and from our point of view, this isn’t everyone problem.
[00:08:54] And it is it applies to the smallest companies, to the largest companies, just like [00:09:00] traditional search does. If you care and about your company getting discovered by users as users shift from traditional search to AI search, [00:09:10] then you’re gonna care about platforms like Gumshoe and. You’ll see it it to a point of, if [00:09:20] I’m a, marketer, is this a priority for this year, next year, what have you, there was a survey done recently of CMOs and their priorities [00:09:30] for 2026, and the number one priority listed was AI
[00:09:35] And AI search was the number one concern of [00:09:40] CMOs within AI, generative AI to follow agentive workflows after that. But if you’re if you think, if you’re worried, is this a priority? If I work for a head of [00:09:50] marketing or I am a head of marketing work in an agency that services marketers, this is the number one thing that they say that they’re worried about in 2026.
[00:09:59] [00:10:00] So how do I, how should I think about this and, the way I like to think about AI and AI search from a marketer’s understanding, and this is the sort of, the [00:10:10] analogy that I think really frames it well. Hopefully as an analogy that you all can adopt as you explain the importance of, and how you’re thinking about managing [00:10:20] AI.
[00:10:21] And the, one of the things that we think about AI is the best [00:10:30] analogy is you walk into a shoe store and we’ve all walked into that shoe store, and what do they have? They have a wall of shoes behind them and there’s a salesperson in front. When you walk into the shoe [00:10:40] store, they ask you, Hey, how can I help you?
[00:10:42] You say I want a new pair of shoes. Does the salesperson stop there and just turn around and grab a random pair of shoes and give it to you? No. They ask follow up [00:10:50] questions. Are, Emanuel, are you interested? Are you a runner? Are you a hiker? Are you a walker? Are you a golfer? Are you a soccer player?
[00:10:56] And then they’ll say, oh, are you a serious or a casual? Are [00:11:00] you big or small? And do you have flat feet or raised feet?
[00:11:05] Emanuel: All of the above.
[00:11:06] Todd: Exactly right. And so how does the [00:11:10] salesperson know to ask those questions? The way it works in retail is the shoe companies come in to the store once a quarter and [00:11:20] they do a training with the sales reps.
[00:11:22] And so Nike would come in and, present to the sales team and train them and teach them the latest updates to the Nike [00:11:30] shoes, why their shoes are better than their competitors, what shoes are appropriate for different types of customers and offerings, et cetera. And then the next week, Reebok would come in and the week after that, [00:11:40] Hoka would come in the week after that, Asics would come in.
[00:11:43] And so in, in the, way to think about that is now AI is that salesperson [00:11:50] representing and answering questions from consumers and from purchasers. And so as a marketer, you need to now think about how do I train and [00:12:00] inform that sales rep? So that’s the, sort of the problem that gummy shoe helps marketers do is, okay, what is that, what is the baseline understanding of that sales rep?[00:12:10]
[00:12:10] And then what do I need to tell that sales rep so they best understand my product, and what to think about it, what to say about it? And then how do I [00:12:20] inform consumers? Because the way that AI has changed is traditional search. Google was responsible, you think about the marketing funnel, there [00:12:30] was discovery, consideration and then conversion, and Google search managed discovery.
[00:12:37] And then you, they would promote a [00:12:40] website, send you to that website to then learn about a product, get recommendations, right? And then ultimately decide where you were going to buy or not. And the [00:12:50] difference now is that AI and AI search is managing consideration. You have a conversation and it’s giving out answers.
[00:12:57] And so you need to make sure [00:13:00] that it’s representing you in those, questions and answers appropriately. And so that’s the right lens. But the way to think about that and how I inform and how I think about AI and what I do about it is [00:13:10] this shoe store analogy is really the best way to think about it.
[00:13:12] Yes, you’re gonna have to train these sales reps because they’re representing you. They’re the ones helping customers decide, now versus [00:13:20] yesterday. Google would send you off to a website to do that. Now ChatGPT, Perplexity, Claude are all the ones managing that question. And then ultimately, when you decide, [00:13:30] I get sent to a website to purchase, but in the future you look at like the rise of Claude Code and Claude Cowork and Claude Bot, eventually that will be replaced [00:13:40] with the, AI will say, do you want me to buy that for you?
[00:13:43] And so that Agentic future is what we’re also helping think about it. So what’s the right way to, to think about how [00:13:50] AI answers those questions? And how is it solving for that? One of the things that we’ve come to learn is [00:14:00] that AI doesn’t just generically answer questions. And that’s one of the fundamental differences I think between traditional search and AI search, which is [00:14:10] AI is really done a good job of figuring out and personalizing who you are and its answers in response to that.
[00:14:17] And so the way we’ve learned AI [00:14:20] works is that the minute you put in your email to create an account, and we’ve heard stats that now, like on ChatGPT, 80% of the users are [00:14:30] now registered. Not necessarily paid. Paid is a smaller subset, but this, to personalize its answers, it just needs your email; because the minute you put in your email, AI [00:14:40] uses that and looks you up and learns everything about you.
[00:14:43] I dare you to ask what AI knows about you from just, even if you have a free account, you’ll be shocked. [00:14:50] Because it takes that email and it looks you up, it searches you, it’s building a profile because what it’s trying to do is understand who you are so that it can do what all [00:15:00] marketing, tech and marketing software is done, recommendation platforms have done, which is people who like this, like that, people will buy this, buy that, and it’s using it to build profiles of you so that it can customize its answers.[00:15:10]
[00:15:10] And so what it’s doing, this is an example for Lululemon example. And so it breaks down its, customer types and it makes recommendations based [00:15:20] upon who it thinks you are. A sustainable fitness advocate, if you’re selling yoga pants like Lululemon would get different recommendations than an [00:15:30] active empty nester or a corporate wellness seeker.
[00:15:33] In fact, Lululemon gets recommended, if I going on to the next slide, [00:15:40] only 16% of the time is Lulu Lululemon recommended in an answer to that first persona, sustainable fitness advocate; because it turns out Lululemon doesn’t [00:15:50] set, doesn’t use sustainable materials, but its competitors do. So on the other hand, if you’re an active empty nester, sort of someone who appreciates [00:16:00] well-known middle of the road brands, Lululemons recommended 85% of the time.
[00:16:04] Visibility means what percentage of the answers were you shown in? And that’s the way that we talk about [00:16:10] AI is what, how visible is my brand? How many times does it show up in an answer to a prompt in a conversation? And so understanding [00:16:20] how AI is identifying different customer types. Then recommending your brand is part of what Gumshoe does and [00:16:30] helps surface so that you have an understanding of, oh, am I getting recommended to these types of customers or these other types of customers?
[00:16:36] And as a marketer, it turns out they’re not recommending your brand to who the [00:16:40] types of people you think you should recommend it to. You’ve got some work and some training going back to that previous example and analogy to do. And so that’s what we’re trying to help [00:16:50] you understand, which is how is what, who is AI recommending you to and then why are they recommending you to these different types?
[00:16:58] And then once you know that you can then [00:17:00] start building a strategy, then what to, do about it.
[00:17:02] Emanuel: What you said about that, I would like, sorry to interrupt, but I think it’s important sometimes we don’t understand ourselves as marketers. Never [00:17:10] in history before any tool had access to so much intimate information about their users. We’ll, leave the topic of the privacy and the [00:17:20] ethics behind it for another time, but it’s important to know that I don’t think never in history they had access to such detailed information about their users. [00:17:30]
[00:17:30] Todd: Exactly. And I think what’s interesting about AI is that it’s profiling users automatically.[00:17:40]
[00:17:40] And it, and there’s also a reason to be registered with AI because it’s, it reco keeps your chat histories, it learns about you. It, once you experience a, [00:17:50] like a good recommendation from AI because the the follow-up questions it asks you when you’re asking about something, it’s really a magical experience.
[00:17:58] Oh, it helped me identify the [00:18:00] exactly the thing I wanted to find. In a way, the traditional search really struggled because it was only a couple of keywords, a short phrase. It didn’t have this back and forth capability. And so [00:18:10] AI search really does, and I’ll talk about some of the things that AI search solved for versus traditional search here in a in, a second.
[00:18:16] because it really relates to how you, as a marketer, you think about it. So [00:18:20] now that you understand that AI is personalizing its answers, it’s having conversations with users. You’re trying to get in the shoes of those users to understand what’s being [00:18:30] said to those users. Obviously Gumshoe as a platform is, helping solve for that.
[00:18:34] And then we surface that information. Then the next question that coffin comes up from marketers. Okay, Todd, you’re helping understand [00:18:40] what types of users AI is talking about with regard to my brand and products. Now what? So partly what we’re we [00:18:50] think about is, and going back to the analogy of the shoe store, which is, okay, how do you talk to AI?
[00:18:54] And so we’re, we help you try to understand by analyzing hundreds if not [00:19:00] thousands of, chats, what AI is saying. And then we can help understand, okay, what to say, right? Because what we, you can do is you look at the chats that are, that you, [00:19:10] we create on your behalf as a brand. And you can do this on your own as well.
[00:19:13] You can then look at those chats and then analyze them, those chats to understand does it [00:19:20] like my product or brand, does it like my competitors products or brands? What are the features of those products that it’s identifying what are the things that, that AI thinks are an important aspects to a [00:19:30] purchase consideration in a particular category.
[00:19:32] So once you know that, that’s the what to say, so you look at the words of the answers and you look at them intimately and in detail, and that gives you a [00:19:40] roadmap of what does AI. Think that consumers care about in a product category. The next thing is, then how do you say it? And that actually becomes really important.
[00:19:49] Meaning how do I [00:19:50] talk to AI so that it understands me and my products and the features and benefits of those products so it knows to recommend me. And it turns out there are very [00:20:00] specific formats of content that AI prefers over others. And which helps them understand, meaning the AI platforms, what is important about your product.[00:20:10]
[00:20:10] And we’ll, I’ll talk about what some of those, features and those formats are as we go. And then the last thing is, where to say it? Do I say that on my own site? Are their third party [00:20:20] forums that Reddits the YouTubes of the world that I wanna say it at? So that those are really the three questions that is a platform, whether it’s our platform or anys you should be thinking about, which is okay, great.[00:20:30]
[00:20:30] A platform can, run prompts and then help me analyze those prompts. But really fundamentally, what you wanna walk away from any platform in this space that’s trying to help you understand AI search is [00:20:40] great. What do you need to say as a brand, fundamentally as a result of, those learning, reading those prompts.
[00:20:45] Two, how to say it. because that’s again, important. AI has a [00:20:50] preferred way of, it wants to be talked to. And then where to say it, like, where is AI learning about these things? Because there is no way to proactively submit [00:21:00] information to AI. There is no API, you can’t just upload a document. You have to seed it on the internet in certain places.
[00:21:07] And again we’ll, talk through where [00:21:10] those are and, and what those are that’s absolutely true. It does crawl, it does do searches on Google and Bing and looks at those. [00:21:20] However, being the number one, you would think, oh, then there, if I’m the number one answer in Google, I’m gonna be the number one answer in AI search. Not exactly. What we think AI [00:21:30] is doing is, and so we an we’ve analyzed over 10 million, you mentioned use the term citations, which is perfectly appropriate, which are links that AI is surfacing in these answers. [00:21:40] Call ’em, sources, call ’em citations, depending on the, AI platform. They use both terms. So it’s somewhat, you can use them interchangeably. And when we look at the, those citations, again, [00:21:50] over 10 million of them, and we, what we did is we looked up the traditional search rank of those citations.
[00:21:56] It turns out, as you see to the, in, in the box there, [00:22:00] 50 to 90% of the URLs cited by AI as a justification for their recommendation had a traditional Google search rank [00:22:10] below 21, meaning they’re page three and beyond, meaning it’s a deep link. And what this is telling you is that AI search is doing a much [00:22:20] better job of surfacing the right information no matter how deep it is in the index of URLs versus the most popular [00:22:30] information.
[00:22:30] The problem with Google search is it’s really a popularity contest, right? The thing that has the most backlinks basically wins. It’s in America, [00:22:40] in high school, there’s a dance, we call it a prom, in your last year of high school, and the most popular boy and girl are elected the king and queen of the dance [00:22:50] of the prom.
[00:22:51] And it’s. And it’s important to with traditional searches that if you’re a brand trying to get surfaced, you’re like, oh, if I’m not [00:23:00] the most popular on dead. And so what AI search solves for is it can read, it’s think it’s a robot, fundamentally, right? It’s a machine. So therefore it doesn’t stop right?
[00:23:08] At page two or [00:23:10] page three, like users, we, what do you say? It’s 10% users get to page two on Google, 1% get to page three, and nobody gets to page four. And, but a robot will look through a [00:23:20] million links and then pull out the important ones, the ones that are correct, right? That are, they’re authoritative.
[00:23:26] That are canonical. And so that’s an important piece. So [00:23:30] you can get surfaced as a small brand if you’re the best brand to talk about something. And I think that’s really an important point here is that [00:23:40] and, from an end user standpoint. This means that Google, or sorry, AI is solving what Google couldn’t, which is how do I find the right information versus just the most popular?
[00:23:49] So that’s one of [00:23:50] the, takeaways here is that AI search is looking for canonically authoritative information, not what’s the most popular necessarily.[00:24:00]
[00:24:00] Okay, what are the types of sites that are getting referenced when they’re getting referenced? And so this is what this next slide is. And in that same, again, that same 10 million [00:24:10] data set, it turns out that the most popular type of site referenced are business sites.
[00:24:19] 30 [00:24:20] to 40% of the links are in fact business pages and are often PDPs product description pages, FAQs, blog posts, [00:24:30] et cetera. And this is people are like, oh, it’s kinda surprising. Now, again, there, there’s 60 to 70% of the links are outside of your domain, but the number one, the… is your [00:24:40] business site, which, if you think about it, makes fundamental scent.
[00:24:44] If I’m trying to make authoritative statements, if I’m AI, who is likely the [00:24:50] most authoritative source for a product? The manufacturer, the builder, the creator of that product. Whereas the problem with AI with traditional search, the problem [00:25:00] with traditional search was that it was so gained by third party, by publishers.
[00:25:05] It was hard for brands to win that. How many times have we looked? We [00:25:10] just need a simple answer about a product like. I, we’ve all probably imagined this, right? So I bought a TV and you’re like, oh, is the TV gonna fit above my [00:25:20] fireplace? Or I have to put it someplace else? What’s the width of it?
[00:25:22] I just wanna look it up. Like, how wide is my new Sony tv? And it’s you at all. You see when you search for something is, oh, a [00:25:30] bunch of top 10 this, or top 10, that type of websites in any popular product search category, right? In Google. Whereas AI does a great job of just giving you that answer and where’s it like that?
[00:25:39] [00:25:40] Where’s it prefer that answer to come from the creator. So that’s, as a business, it understands that you’re likely the most canonical source of information about your products, and that’s a [00:25:50] fundamental change. Now, at the same time, then you gotta figure out, how do I talk about it? What’s the language and words that AI prefers when you make those statements?
[00:25:57] But the good news is [00:26:00] AI likes business pages. So in a way, if you’re a brand, you’re back. Your website matters. You don’t just have to worry about these. You’ve probably heard stories about YouTube and [00:26:10] Reddit and whatever, and I’ll talk about that too in terms of, are those sites popular within ai? And the answer is, yes, they are.
[00:26:15] But your brand site is incredibly important going forward in the world [00:26:20] of AI search.
[00:26:22] So the next, I did talk about third party sites and it’s funny, some of my competitors are just now talking about, oh, [00:26:30] we see that YouTube is more popular than Reddit in AI search. And my answer was yeah, I, we figured this out a year ago.
[00:26:37] And Reddit is popular. [00:26:40] Social information sites like Quora, YouTube, and Reddit are popular and you should be posting to those platforms as well. [00:26:50] And if AI is referencing a particular thread in Reddit or Quora or LinkedIn or some other community forums, you should absolutely as a brand, [00:27:00] participate and look to answer questions there.
[00:27:02] And the reason AI likes those sites is they’re great for questions and answers. And AI really prefers [00:27:10] looking at question and answers as a way of training itself. How to answer its own questions. Now what’s interesting is that AI loves YouTube and what it loves YouTube are [00:27:20] like product how to videos.
[00:27:21] They love explanatory videos. They love review videos. And so they love, and the other thing is they love long form videos. They also like long form [00:27:30] content. I think that’s different traditional search, SEO. It didn’t necessarily like long form really dense, informative content. But AI loves it. The more [00:27:40] informational it is, more popular it is with AI.
[00:27:43] And the other thing is a quick hack for YouTube. So yes, make sure you’re uploading all your content to YouTube, all your videos, but we don’t [00:27:50] believe it’s doing automated speech recognition, the three acronym is a SR. We don’t think it’s doing a SR, we think it’s reading transcripts. So make sure the automated transcripts are turned on when you upload content.
[00:27:59] [00:28:00] So that’s a little bit of a hack here is YouTube loves long form, or sorry, AI loves long form YouTube videos and they love transcripts. So make sure that’s turned on. You upload your content, but upload all [00:28:10] your corporate video content to YouTube. Absolutely upload it every last piece. That is a must do going forward.
[00:28:17] The other, so then the question is, and you [00:28:20] bring it up, right? Small brands, new brands. Can you win? Can you, are you SEOed out? Is it too late? The answer is no. You are, you have lots of time. You have you, [00:28:30] can still get into it. And one of the interesting things here and what we’re showing with this stat is, and you’ve probably heard about recency of content, which makes sense.[00:28:40]
[00:28:40] And the average age of URL published date in terms of age of a URL cited by the models in our dataset is 86 days old. And that’s decreasing. [00:28:50] 10 to 15% every quarter. So next quarter will be 78 days and it’ll be 70 days and what have you. And this sort of speaks to the models doing realtime search.[00:29:00]
[00:29:01] And so what they’re doing is they’re looking for the most recently, the most recent version of a piece of, content. Of information. Because what’s the [00:29:10] most likely or most canonically correct piece of information? The most recent version of it. And so we’re seeing this, which means as a brand, the good news is if you’re a new [00:29:20] brand, it’s okay because within AB six days, you’re getting crawled, you’re getting surfaced.
[00:29:25] You’re able to really, we’ve seen results in two to three weeks. Once you start creating content to talk to [00:29:30] the models, to work with them, and you’re able to really. Get surfaced in a way that you didn’t, it was harder for traditional SEO where it took 12 to [00:29:40] 18 months, it was a very much longer thread.
[00:29:42] And so in two, in, in the short as two to three weeks, in a more typical time, it takes about two to three months to get surfaced. But you can get surfaced and in a [00:29:50] relatively short amount of time. And we think that this will actually get down to as close to a real time conversation between you and AI kinda going forward.
[00:29:59] And [00:30:00] so this is interesting from a brand standpoint, where we see the future is you’re gonna have, as AI sends traffic to you, basically AI is [00:30:10] gonna look for signals. Did this user convert? Was, it the right kind of connection to a user? And a feedback loop today, if you think about it, is 86 days of, it’s an 86 day conversation between [00:30:20] you and AI.
[00:30:21] In 12 to months, maybe 24 months, that’s gonna be close to a realtime conversation where you’re gonna be creating content to respond to traffic that AI is sending to you. And then creating [00:30:30] signals give back to AI is whether it’s working or not. That’s the way that, that we like to think about this as a marketer, which is the content you’re creating is [00:30:40] really just intended to be a conversation between you and AI and how you can inform and train AI.
[00:30:46] And so that’s the way that we think marketers should think about [00:30:50] this. Again, I’m training ai, I’m engaging with ai, and I need to do it in a much more almost real time basis than I maybe did and thought about this in the past.
[00:30:59] So [00:31:00] hopefully that’s a good sort of starting point for how to think about ai, what can I do about it from a marketer standpoint.
[00:31:07] And from there we can [00:31:10] now explore particular companies and, look at the, like the Gumshoe platform and how it surfaces things. Obviously, Emanuel you, [00:31:20] used Gumshoe and that’s how you found us and me and wanted to, talk about us today.
[00:31:24] Emanuel: It’s an amazing tool. And I’ve introduced to some of my clients, I’ve been able to make some [00:31:30] small tweaks that resulted, at the end of the day, businesses need to make money.
[00:31:33] So what we do as marketers is help them, those business make money. And by using Gumshoe, I made some small tweaks, really [00:31:40] obvious that actually resulted in the revenue growth. So it’s amazing a masterclass on how AI works, how AI search works, and, [00:31:50] I couldn’t ask for something better. I’m gonna come back to it to take some snippets and share them on social as well, because it’s [00:32:00] really, important to, for everyone to understand the the process, behind it and to understand that it’s not exactly the same thing.
[00:32:08] I love what you said, canonical [00:32:10] source. I’m gonna use that in my vocabulary more and more because it makes complete sense right now in some context. I like that you said,
[00:32:18] Todd: but one of the things to [00:32:20] that point that we have seen. And that’s really, I think, one of the things to think about this, which is when I talk to AI, how could I be seen as the canonical source [00:32:30] for, of information for a category for something?
[00:32:33] And we have, one of the things around that is we’ve seen much, much less of this concept of domain [00:32:40] authority and much more of the concept of like knowledge, authority. And one of the examples of that is we’ve had some companies, users of our platform [00:32:50] who we surfaced some of these canonical writers and long tail who were being referenced by AI, the answers.
[00:32:58] And what they did is they [00:33:00] recruited these writers. They created an author panel. The great thing about these long tail writers is unlike traditional SEO, where the top 10 websites that are getting referenced by Google get hammered, [00:33:10] talk about my product, talk about my product, these long tail writers, nobody has ever, very few people read their sites and AI starts surfacing them. And so as a, [00:33:20] what they did is they reached out to these writers and recruited them to write on their own blog. And because they’re usually one of the first companies to knock on their door they’re often like, oh [00:33:30] God, somebody found my little known unknown website.
[00:33:33] And so what they did is they, invited these, they called it an author panel. So it’s a great recommendation is you can recruit author panels. And [00:33:40] then the thing about this is you recruit these folks, you put bylines, you put links to their bios. The bios are what AI loves. They love authoritative people with academic background, [00:33:50] industry background with, like awards and recognition.
[00:33:53] And so they recruited these people to write in their own blog, and they saw within a month or two AI referring to [00:34:00] the blog posts that were now on the corporate domain from these same writers. So there’s definitely the signal around knowledge, authority about canonical [00:34:10] information, canonical sources.
[00:34:12] And unlike domain authority. We call it knowledge authority inside Gumshoe. And so that’s this concept to think about, and I think it’s really [00:34:20] fascinating shift, right? AI is solving ai search solves a lot of the problems of traditional search in, right? It could be gamed, it was it, became [00:34:30] overcrowded, it was less relevant over time.
[00:34:32] And so this is the start of the things that happened. So it a, as a company, as a marketer, you can recruit some of these long tail writers to write with you and [00:34:40] write for you in ways that you, that can really be helpful to getting your own brand site surfaced in AI.
[00:34:47] Emanuel: Before we jump into the presentation, because [00:34:50] there’s the second part of today’s session as well. I do wanna ask you a question before we dive right in. My question was, you said the Reddit Quora and all [00:35:00] these places, which are important for brands to be there.
[00:35:04] From my experience, there’s not enough EEAT – expertise there [00:35:10] on those those platforms, and they also can easily be manipulated, which is a topic in itself. What are your thoughts on that and how would you, [00:35:20] what are your comments on that? How would you comment in a few,
[00:35:23] Todd: couple seconds? Yeah, I, I think that is a, it’s interesting challenge as a brand.
[00:35:28] We definitely have some brands that [00:35:30] have tried that have used Gumshoe to make sure they start paying attention to the right Reddit thread because there’s a right or the right Qora threads and [00:35:40] the, and so as a result, what we do in terms of, if you look at here. In the example that I’m sharing, which is the Lululemon [00:35:50] example. So you’re looking at Gumshoe and this is a report that it runs on. And the way we look at Gumshoe is that you, put in your company and then we help [00:36:00] identify products, the key products, your area, and we focus our reports and analysis at a product level, which is the way people search.
[00:36:07] So that’s what we wanna do, is do insights at, within [00:36:10] a product or brand level. And then one of the things that we do is we create these personas which will then have chats. Then we turn these effectively into a agentic [00:36:20] bots that have conversations with ai. We record those conversations and analyze them, and that becomes the basis of the report.
[00:36:28] And so what, one of the things we [00:36:30] pull out are what were the URLs cited by AI? And so here’s an example of yoga [00:36:40] apparel. And so these are, you can see the persona and the prompts that included this answer. What’s interesting, this answer was given across three different personas to three [00:36:50] different prompts.
[00:36:51] And so this is clearly a source that has some weight in the models. So one of the things that, that this will help you understand, which is [00:37:00] the models don’t just pick out any Reddit thread. They try to find ones that have a writer who has some authority in the space. because they’re looking [00:37:10] at the bios and answers of that writer.
[00:37:12] So that’s one of the interesting things is they wouldn’t just have picked this randomly. They would evaluate who is the writer of this thread? [00:37:20] Is it someone that does have some authoritative knowledge in this space before I recommend it? That’s what’s different about AI is it can look at multiple signals when coming up and [00:37:30] surfacing an answer.
[00:37:31] So I
[00:37:31] Emanuel: think they call it karma points on Reddit. So
[00:37:34] Todd: it’s what they’re doing. And it’s it, which [00:37:40] gets to I see Quirino is asking a question or I have a question. How do you put the tenets of your brand, the soft aspects so that AI can, read them? And one of the interesting [00:37:50] things to your point that we wanna call out, which is we’re in this example, there were almost a thousand chats that [00:38:00] happened to generate this report from Lululemon.
[00:38:03] And the way that it works, as I said, is that we’ll, create these identify these target [00:38:10] customers. Again these types of profiles of users who the AI thinks are relevant to this product category. And then we’ll analyze those [00:38:20] chats. And so I’m gonna pull up here, one of the things that we’ll pull up in those chats and look at what are the product characteristics that were [00:38:30] referenced in the answers.
[00:38:31] In the chats that happened. So Quirino, what you’re seeing here is these are the product characteristics for this particular pant. A [00:38:40] line high rise pant, 25 inches. This is a type of yoga pant that Lululemon creates and then you can see there effectively, what’s the word map of what AI [00:38:50] called out about that product so that it we’re helping, we wanna help you identify what are the things, and then we can do a comparison of, let’s say this is a competitor.[00:39:00]
[00:39:00] What are the things that are referenced by the brand elements that were referenced by for your product versus your competitors? You get a mapping of what [00:39:10] is AI calling out now the question, the follow up to that is, okay, great, Todd, but how do I know what, or feed [00:39:20] that back and how do I understand and how can I talk to AI?
[00:39:23] So one of the things that we do is we analyze the chats and what is, and as [00:39:30] I just showed you what was being said, and that becomes a roadmap of what are the features of a product type or category that AI cares about? And that’s effectively calling out like, [00:39:40] here’s the things that you should be talking about.
[00:39:42] And so you, we could then help you identify, oh, do I want to create content that’ll talk to a particular persona under a particular topic? [00:39:50] And topic, meaning a grouping of prompts, a category of information. And then from there you can generate different formats of content. [00:40:00] In our case, we’ve helped identify that AI likes, FAQs, knowledge base articles how-to guides, et cetera.
[00:40:08] And then you see, also see for [00:40:10] offsite social posts and then surface, that becomes the format, right? The, how do I talk? What you saw before is the word map of what [00:40:20] do I potentially talk about, and then we help generate drafts as well of the content that you can be posting. And so I’ll show you an example of what’s something that’s already generated and that’s [00:40:30] generated on the at as you do it, so it’s not pre-generated.
[00:40:35] And so here’s the things that we would say that you’d wanna talk about. You could talk about [00:40:40] your brand characteristics, soft characteristics, whatever you wish, that becomes something that you can talk to. Now, one of the things that AI does and you’ll see here, this is a very [00:40:50] academic style. So if you’ve ever read an academic paper like by university professors, you’ll notice very debts, very information informative, and there’s [00:41:00] also citations, references.
[00:41:01] You see how here we call out a fact we, provide a, link to where that fact comes from, FACT, not FAQ fact [00:41:10] and. AI loves reference and verified information, and so you’ll eventually see at the bottom of this recommended FAQ list that we would potentially a brand could [00:41:20] put on is you’ll see then the links of information.
[00:41:22] So you can absolutely talk about things in a software or different brand characteristics, but the question is are, is there some [00:41:30] basis for understanding what are the brand characteristics that AI prefers or doesn’t prefer? And so that’s the point of what you see here is [00:41:40] we’re trying to help you highlight, here’s what AI is talking about with regard to your product category to help give you some insight of what should you be talking about this or that?[00:41:50]
[00:41:50] And then if you want to then inform AI in a slightly different way, let’s say you’re, in this case, if you’re Lululemon, they’ve got a real challenge around sustainability. [00:42:00] I’m just pulling going back to the main page of the report so I can highlight that for you. You’ll see here, right? They have the Sustainable Fitness Advocate, 16% recommendation rate, and [00:42:10] if I go down, you’ll see they’ve got a competitor girlfriend Collective, where they’re getting recommended for this particular type of product, 63% of the time versus only 16% of the time for Lululemon.
[00:42:18] And so if I [00:42:20] wanna talk, let’s say I’m Lululemon and I have sustainable features they’re not getting understood or recommended by AI.
[00:42:28] Emanuel: There’s a follow up from [00:42:30] Q.
[00:42:32] Todd: He was asking the factual aspects. You can talk about those. But the question then Quirino is does AI [00:42:40] care about them? And there are categories where absolutely, that can be a factor, right?
[00:42:45] Fashion is certainly, you could ask questions that are more about fashion and fashion preferences [00:42:50] and AI will reference threads and other videos where they might be talking about trends and so forth. But the real thing that I think is, [00:43:00] important is to understand what a, what AI cares there. And that’s an interesting question Quirino, you’re and, I’ll go in a slightly different direction, [00:43:10] which is what does AI and, how do I market to AI or how should I think about that? Is marketing to AI the same as marketing to consumers? [00:43:20] And you’re getting to an interesting thread and a point, and where I’m bringing this up is a good example is in the pharma [00:43:30] space, one of the things that we have seen is that in pharma.
[00:43:37] There is, there are end [00:43:40] consumer brand sites for their products, for their drugs, and then there’s the corporate site. And at the corporate site they tend to put the legal material, the very dense research reports, [00:43:50] the much more factual information, and then they make the consumer site very brand friendly to your like soft, like the soft considerations Quirino, that you’re talking about.
[00:43:58] So [00:44:00] in the pharma space, we’ve seen this again and again where AI will reference the academic research behind a drug on the corporate site and [00:44:10] almost 80 to 90% of the time and not the consumer brand site.
[00:44:15] And it drives pharma nuts because they don’t want people going to the boring corporate site. They want people going to the [00:44:20] branded product site. So there, I’ve literally had the comment come back to me like, Todd, I don’t care about the corporate site. I want this site. I’m like [00:44:30] you have a choice.
[00:44:31] You either have to put the content that AI cares about on that consumer site or just [00:44:40] accept that this is the way the, it wants to think in the world’s gonna work. So it’s an interesting question we don’t wanna put that stuff, that content, that boring legalese or research [00:44:50] academicy content over here because this is the brand soft site. And I’m like I don’t know what to tell you, but AI likes this informative content more than it likes [00:45:00] this brand only content. The sort of the soft content to your point, Quirino. And so it is an it to date, AI has much more, preferred [00:45:10] factual content versus the pure brand content.
[00:45:15] However, if we were to, if I were to go back and ask a [00:45:20] prompt. About trend information, right? What is the, what is the trendy, what is the most popular yoga pan this week? This month? I would actually [00:45:30] get a different set of answers. However, when you look at the questions that users are asking, because what’s informing the questions that we ask here is if I go to, you can see the [00:45:40] topics and prompts and topics are groupings of, prompts.
[00:45:45] And then you can see the individual prompts below that, that were behind this report. [00:45:50] And the way we generate these prompts, and they’re generated custom for each report, is that we have we work with a user panel users who’ve opted [00:46:00] in to have their search behavior tracked. And when you look at the prompts, and so you see up here you’ll see we’ll, include how [00:46:10] popular a topic is.
[00:46:12] From that panel data. And when you look at the types of questions, user asking AI search, they’re asking factual and less soft [00:46:20] questions today. They wouldn’t go to AI to ask what’s the trendy thing? What’s the most popular? I I need a, cool new pair of pants that’s not being asked of [00:46:30] AI yet.
[00:46:31] So we look at the panel data. Panel data says its informative, factual based questions are what AI is getting asked today, where you see [00:46:40] in social forums more of those trends soft questions, again, that might change over time, but today it’s more fact-based. But AI reads everything. [00:46:50] And if you ask it a trend question, it would look for more trend, soft based sources, which would be more social sources.
[00:46:57] Happy to, help. And one of the, [00:47:00] and I should say this, one of the great things about Gumshoe is that you so Quirino, in your own case, you could inject prompts that were more in a [00:47:10] different direction. And so those soft aspects, you can push that. We start from looking at the panel data and then using that.
[00:47:17] But as one of the great things about the Gumshoe [00:47:20] platform is you can steer it however you want. Hey, I wanna ask more soft questions. I wanna look and see what might be said about these areas. But that soft versus [00:47:30] fact-based discussion I think is gonna be as, as more marketers become more familiar with how AI and the questions that are getting asked and as consumers be more, [00:47:40] become more comfortable.
[00:47:40] Kind of going beyond, I think, fact-based questions. I think this is gonna be one of the big topics for marketers is how do you brand in the world of AI. [00:47:50]
[00:47:50] Emanuel: I have a question from Brian as well. You want me to read it out loud or do you wanna read it?
[00:47:55] Todd: Go ahead. You can read it, Emanuel. Go for it.
[00:47:57] Emanuel: To the soft aspect, how does AI score [00:48:00] celebrity endorsements?
[00:48:00] Example Ryan Reynolds, and then what happens when they no longer trend?
[00:48:07] Todd: It this is a great [00:48:10] question around celebrities, which is in AI cares less about celebrity endorsements and more about these, again, we were talking [00:48:20] this authoritative or canonical sources, right? So they love people who have more of an academic, more of an industry reference point.
[00:48:27] But that being said, fashion is an area [00:48:30] where a celebrity would have potentially a particular voice and have some opportunity. But I think the idea of when you look at what’s getting [00:48:40] referenced on Reddit in AI, what’s getting referenced on YouTube, it’s not necessarily celebrity stuff except if it’s particularly, I think, relevant to an area that might have some [00:48:50] expertise around, like Ryan Reynolds today, because he owns a soccer team, whereas you, Europeans would say football team correctly, he would get [00:49:00] references to football or soccer. Because he’s an authoritative figure now because he owns a team. But for other things like cell [00:49:10] phones, maybe not right or other areas that Ryan is a part of, but I think celebrity endorsements is one of the fascinating things that, that changes with ai. Does AI care about [00:49:20] celebrity? Not as much. It cares about authoritative figures or knowledgeable figures about a category much more than it cares about just someone who’s popular.[00:49:30]
[00:49:30] Emanuel: And I’m, just throwing out ideas right now.
[00:49:33] Paid ads will come soon in ChatGPT and the likes. So I’m assuming that will be [00:49:40] covered more too, in that aspect.
[00:49:43] Todd: And Yes and, I think to the point of paid ads in AI, like that’s coming. We’re seeing the [00:49:50] experiments, what we like to help I think the, what we’re doing with personas, and again, the way that when you build, we generate these personas for each particular brand report that [00:50:00] we generate.
[00:50:00] So these will be unique and, created for each one. And in many ways what we’re, this is who AI thinks are the most relevant [00:50:10] personas or ICPs for your product and category. What we do is we go out and interview 12 different LLMs. We also call your site. And through that combination, we come up with these personas.
[00:50:19] This [00:50:20] takes, what’s amazing about this is, Emanuel has seen, it takes two minutes to do and you all can sign up for Gumshoe. You can create a free report yourselves. No credit card required, no trial. Just it, there’s no gimmick [00:50:30] there. You can, you’ll be able to see this for yourselves. And when we think about paid ads, what’s relevant here is this is who AI thinks are relevant to your product category.
[00:50:39] And [00:50:40] so it becomes maybe a starting point to when I wanna think about targeting, which will we think will be a combination of topics. To go back to this, would be what you would effectively be the targeting [00:50:50] vector targets, topics plus personas will be recommendations and what you could potentially be targeting when sponsored answers, which is what we think the format will be [00:51:00] appear down the road.
[00:51:02] Emanuel: I do recommend every business owner or everyone who has a site to try Gumshoe. It’s a perhaps not a busy conversation to [00:51:10] have with some of the clients when you pull up this report and saying, okay, we identified this persona. You have no mention of. What, how you’re targeting them on your [00:51:20] website and stuff like that.
[00:51:21] Plus you can edit and add more. It’s an amazing tool. It’s been great. And I’m sorry that we actually didn’t have time to run through from a start [00:51:30] to end or at least to test out a new, brand altogether.
[00:51:33] So maybe that’s that’s something that we could look out separately, but I think the Lululemon was [00:51:40] a great example as well. And you are doing right now adding some custom prompts.
[00:51:46] Todd: Correct. So you can add a custom prompt to the point about Quirino and brand and [00:51:50] oh, I see all these fact-based questions.
[00:51:51] How do I explore that? You can add it and then what’ll happen is the next time that you run your report, that [00:52:00] prompt will be added. So I can go in and then rerun the report. What’ll happen is, and you can, you have the ability to select the models that you wanna run [00:52:10] against. And so what’s gonna happen is all the prompts now in your report will run.
[00:52:15] And so we’ll get a new version of that and you’ll be able to look for the prompt that we just added the, trend based, [00:52:20] have the soft question. So that’s one of the great things about our platform. Even though we’ve, we draft those prompts for you to start, you can add custom ones yourselves, and then you [00:52:30] have the ability to move.
[00:52:31] And this is very typical. Like you, you see a report green, you’re like, oh, this is all fact-based. It doesn’t have trend-based stuff. I wanna add that in. You can certainly do that. [00:52:40] And so that’s one of the benefits of a platform like ours is you’re, you can manage it however you wish over time.
[00:52:46] Emanuel: And as you can see there, course, that’s the really interesting part.
[00:52:49] You have [00:52:50] all these these models that you can, you can reference because they differ. Two deep seek to Open [00:53:00] AI, two from Google, search, and the Gemini.
[00:53:05] Todd: And one of the interesting things that, and you, that you pull this out is that [00:53:10] we use APIs to, submit these, which allows us to run and submit the prompts to specific models.
[00:53:18] If you’re using a platform that [00:53:20] uses headless browsing and scraping, they don’t use a logged in experience like what we are mimicking and you don’t have any control over, is it going to ChatGPT mini or [00:53:30] ChatGPT Think or Pro or what have you. Same thing with Gemini. And what you we learn is that the models give very different answers depending upon [00:53:40] whether it’s mini or pro or thinking.
[00:53:43] And the other thing is like mini, which is the, if you’re a not logged in user, the models all [00:53:50] use the, kind of the cheapest versions of the models. They have mini light, basic, et cetera. There’s more hallucinations, there’s less sources or references for them. [00:54:00] And so one of the reasons consumers are signing up for these logged in accounts is that’s the only way you can access these more thinking reasoning models [00:54:10] is by having a registered account.
[00:54:11] And so what we’re trying to do you is a show you the differences. You can see a what does a light model think versus a promo and what are the differences in what’s getting [00:54:20] cited. For instance, you’ll see more Reddit references in the mini model, but in the pro model you’ll see more like brand [00:54:30] references and some of those deep links and some of those deeper answers and so forth.
[00:54:33] And so some of the, that’s one of the things that sort of pops up as a result. [00:54:40] What we’re doing is this is finalizing and updating, and then you’d be able to go in, you can go to the conversations view, and then I’d be able to look for that specific prompt that we just added under the, [00:54:50] first one, the first persona, and have the ability to look at that answer and see what it said in response to that information.
[00:54:59] Emanuel: It’s [00:55:00] almost done, and I, would say, is it market research? Yes, of course. But it’s not just market research persona research
[00:55:08] ICP identifying. [00:55:10] I was surprised to just by the default personas that Gumshoe was able to create for, the businesses. I consulted to discover some new ICPs that [00:55:20] were totally relevant and were completely missed by the customer.
[00:55:23] Not to say that many small organizations actually don’t get through the exercise of identifying the their ICP [00:55:30] name them and spend the time on the characteristics of those personas to address them correctly. Gumshoe does that for. Within a couple of clicks, [00:55:40]
[00:55:40] Todd: right? So here’s the, custom prompt that we injected.
[00:55:43] Here’s the, answers. So you get the chance to see the prompt. And then what we do is we have that chat. And the way we our [00:55:50] chats work is we drive to a product recommendation in the back and forth, and then we parse out that and show it to you. So you can see in response to that prompt what were the product mentions.[00:56:00]
[00:56:00] And so as a marketer kind of read through and then learn what was actually said to a level of detail. Or I can just look at the sort of the summary analysis that we had talked about earlier.
[00:56:08] Emanuel: Amazing. Are there okay. [00:56:10] Excellent. Yeah I’m amazing. Gumshoe is a very complex tool. I would say maybe. Could look a little bit intimidating. Those matrix, those…[00:56:20]
[00:56:20] Todd: it’s a lot of information there, but it’s cool as you dive in, right? So here’s the trending. It’s like the, what’s fascinating though is you look at this and Corina, I think you’ll see [00:56:30] that it’s, there’s still a lot of fact-based information here, right?
[00:56:33] Even we ask it to be trends, like it just AI is, it how you get soft considerations to come [00:56:40] in? I think will be one of the questions. Now, what’s also fascinating is the models change. Like we’re seeing updates the models like ChatGPT is up to 5.4. A year ago was, it was at [00:56:50] a year and a half ago.
[00:56:51] It was at model 2.5, then it went to three, then it went to four. Now it’s at 5 5, 5 0.1, 5.2. And they’re evolving. And I think to your point about why do we keep [00:57:00] wanting to redo and ask, because the models are changing their answers. And I think the soft question, the soft branding one is gonna be one of the most interesting trends [00:57:10] about how does AI, again, if it’s gonna make a agentic purchasing.
[00:57:14] Possible. It’s gonna have to understand the soft considerations that people care about [00:57:20] versus just the fact-based considerations that it is focused on today if it’s gonna select the right product. Are you gonna want any pair of pants? No, you’re gonna want a pair of [00:57:30] pants that are gonna make you, if you’re going out on Saturday night, that are gonna make look good and feel good and be popular.
[00:57:35] So how does AI solve for that? It’s gonna be what I think one of the interesting ones is as it moves into [00:57:40] this world of decision making, product, decision making,
[00:57:43] Emanuel: Not, but when, because it’s happening. I’m looking forward for that shift to happen soon. [00:57:50] We’re getting, we’re already past our one hour mark, but it’s been an amazing webinar.
[00:57:56] Are there any last comments, questions for Todd? If you [00:58:00] have about Gumshoe since we were having him with us live here now? If not, Todd. Any last [00:58:10] comments, ideas, suggestions, or something that you would like our audience, the ones that are live or the ones that will be watching the webinar recording to leave with [00:58:20] from this session today?
[00:58:23] Todd: I, think, thank you Brian, and I think that the thing to leave you with is [00:58:30] that there’s a lot you can learn about what AI thinks about your brand and products and what’s really fascinating. What’s really, actually, one of the things I [00:58:40] think I’m really inspired about is we, start with this lens of AI search, but to your point, learning about these personas and what does AI think about that, that [00:58:50] those are, that’s marketing insights and brand insights and this competitive landscape that I’m showing you here for Lululemon, right?
[00:58:57] Where, they’re popular with certain types of users and where they’re [00:59:00] not, there’s a lot of brand insights that can come from, this idea of AI search, and that’s really where it gets exciting to us in thinking about, [00:59:10] which is I started trying to understand how my brand appeared in AI search, but I really figured out that I’ve got a pro, if I’m Lululemon, I’ve got a product development problem I better solve around [00:59:20] sustainable, fabrics. And so that, that’s not a AI search insight per se. It’s actually a brand or product insight. And so I think that’s where I get really [00:59:30] excited, which is I start with trying to understand how visible my brand is in AI search. And then I learn about personas and I learned about competitive landscape and I learned about product [00:59:40] features and there’s a lot of great things you can learn.
[00:59:42] And again, as Emanuel, right? You can generate a report in about 15 minutes of coming to Gumshoe and you’re walking in within [00:59:50] 15 minutes. You’re walking away with a ton of cool brand and market and product insights in ways that you probably didn’t even realize. And I think that’s the biggest thing is there’s so much you can learn in just a quick amount [01:00:00] of time about your product that used to cost like a quarter of million dollars of
[01:00:03] Emanuel: Exactly what I wanted to say.
[01:00:04] My point exactly. Brands like Lululemon would pay millions of dollars to get these [01:00:10] insights.
[01:00:12] Todd: Right.
[01:00:12] Emanuel: And now you, have actually, you can emulate that, right? You can put the other, your competitors
[01:00:17] Todd: brands every, firm of every size and nature can [01:00:20] with, this. And then if you decide to sign up with Gumshoe it’ll cost you like a just is at, it starts at about 200 bucks a month.
[01:00:27] You can get insights like this about your product on an [01:00:30] ongoing basis. Not hundreds of thousands of dollars, not tens of thousands of dollars a month, but just a few hundred dollars a month is, a great place to start. And this is the type of information you walk away with.
[01:00:39] Emanuel: And once you [01:00:40] put it put it to work and you implement some of the suggestions, it’ll translate in inevitably in revenue growth.
[01:00:47] That’s what businesses are [01:00:50] here to do. That being said, thank you so much. Gumshoe.ai is the website where you can go in and check, [01:01:00] thank you so much, Todd, again, for being with us today and hopefully we had a, we’ll have a chance to do it some other time if people wanna connect anytime. I
[01:01:07] Todd: think
[01:01:08] Emanuel: LinkedIn is the best [01:01:10] source as well.
[01:01:11] How about marketing.com is the website where you can go and sign up for the newsletter. You will see today’s recording, previous [01:01:20] webinar recordings, upcoming announcement with other events as well. Gumshoe.ai Todd Sawiki as our guest.
[01:01:29] Todd: Thank [01:01:30] you very much. Glad to be here.