- Duration: 40 min
- Platform: meet
- Source: Read.ai 01JYF3VWHQ242DWZ1QK8C64C0F (https://app.read.ai/analytics/meetings/01JYF3VWHQ242DWZ1QK8C64C0F)
Transcript
[Michael Houck]: Will, how you doing, man? How are you? Good. [Wil Armstrong]: Good to meet. [UNKNOWN_SPEAKER]: Great to meet as well. [Wil Armstrong]: Where are you calling from Where are you based out of I’m not based anywhere. [Michael Houck]: I’m currently in Philly, actually just spending a couple of days with my parents. So seeing them for the first time in a bit, which is nice. But yeah, based all over. What about you? Awesome. [Wil Armstrong]: Well, I’m from the East Coast. I’m from Maryland. So close by. The DMV. [Michael Houck]: There we go. [Wil Armstrong]: Yeah, that’s right. Sports fan, though, so doesn’t come into conflict at all. Hey, you know, I mean, at least at least for your sake, you guys aren’t our first game of the year this year, so you don’t get to witness us raise the banner. But you know, that’s right. [Wil Armstrong]: Cowboys aren’t so lucky. [UNKNOWN_SPEAKER]: No, no, exactly. [Wil Armstrong]: I was just thinking about whether or not to get Sunday ticket. Finally, that is a decision in my mind because we have a decent enough team. You’re like, we’re making enough money where I can afford the ridiculously overpriced NFL subscription service. I live in Utah now and I don’t get the games, so it’s a consideration. Where in Utah? [Michael Houck]: I just spent three weeks in Provo actually a month or two ago. [Wil Armstrong]: No kidding. I’m in Park City area. Nice. Yeah. So do you go to Utah often? Provo is a fantastic place. No. [Michael Houck]: My first time in Provo, I actually went to Sundance a few years ago. But yeah, just looking for a new home base. And some friends were like, we know everyone’s Mormon. We’re Mormon. But go check it out. I’m like, all right. I don’t know if I’m going to convert, but I’ll give it a try. [Wil Armstrong]: I’m not a Mormon, but yeah. [Michael Houck]: Yeah, certainly. It’s a big culture out here. [Wil Armstrong]: Yeah, it was cool, though. Very beautiful, beautiful place. [Wil Armstrong]: Yeah, gorgeous. Awesome. Excited to excited to meet. I know. How do you know Janet and the team over at Daydream? [Michael Houck]: So Janet at the old company she used to work at, at Ermin, we did some work for them a while back and their CEO. And she was our our main point of contact there. And then I don’t want to butcher his name, but daydreams founder. Yeah, we got connected a while ago, just like casually, and didn’t end up working with them because they’re going after like larger enterprise level customers. And we’re, you know, at least this company is a is a bootstrapped hold pro that I have that I’m reaching out to you guys about. So didn’t make sense there. But they were both like, yeah, you guys should talk. So Nick. [Wil Armstrong]: Yeah. Well, really cool. Yeah. Excited to just chat AI search in general. How, uh, what prompted you to reach out in terms of AI search? I know you mentioned that, you know, SEO has been a, uh, lackluster part of the marketing department, but, uh, yeah, we can call it a weak point cause that’s what it is. [Michael Houck]: Um, so yeah, I mean, look, um, I’m a growth guy, I’ve been a growth guy for a long time, go to market, all that stuff. But SEO was the one area that I never really dove into. And, you know, now, when I look at all the channels, like every marketing channel sucks right now, right? Everything sucks. And the one thing that doesn’t suck is getting discovered through LLMs, jump bots, things like that, at least in my opinion. So I see it as underpriced attention, I see it as opportunity before, you know, the people who need to cross kind of get there, right? And so I want to be ahead of that curve like I wasn’t on SEO. And we have a couple of different products and services that I think could benefit from being discovered that way and that people probably would search for in those settings. So I want to learn and then also would love to see if there’s a way to partner up and kickstart that for our entities. [Wil Armstrong]: Awesome. Well, that’s really valuable. I can definitely make the like our time educational, because there is a lot of nuance to how AI does search. ChatGBT even does search differently than Perplexity and Google AI Overviews. Although we’re a startup, we just launched the platform publicly in February. So everyone in this space is a startup. It’s really a fast moving time. And so with that, of course, the enterprises that are watching it grow underneath them before they make a strategic move. And I think there’s a great opportunity for folks who do want to hop on and optimize for AI. So yeah, happy to make it educational. So at the very least, you walk away with some. [Michael Houck]: I want to make it valuable for you guys to say, I want to make sure that we’re moving towards something that’s like, hey, like, here’s how we could work together. Here’s how Scrunge can help either this company of ours or a different one. So I don’t want to make it a total waste of your time. [Wil Armstrong]: I’ll just be up front with our pricing. We just do month to month right now packages. We think that there’s a lot of value you get from Scrunge, even if it was just for one month, understanding data and where AI is going to for answers. So we don’t lock anybody into long-term packages. We have packages that start at $300 a month, and they scale up with number of prompts. So our costs are prompts being monitored. How many times we’re pinging the models and getting responses from them. So yeah, for transparency, that’s where we’re coming. That pricing sounds very reasonable to me. [Michael Houck]: So no concerns there. Yeah, I’m surprised it’s based on, I guess I don’t even really know this works. So I’m just like, I’m such a, I’m such a noob here. [Wil Armstrong]: Yeah, well, let me walk through just a couple slides and I’ll flip through the platform itself and see how it would intersect with what you’re, you’re looking for here. So I’ll just share my screen. This is our, so some background on me. I run customer success and data for scrunch. Actually, I’m not often in these initial calls. So excited to be here. I was employee number one for Scrunch a number of, well, a couple of years ago at this point, we were building for this idea that AI is going to totally change the customer journey. Is going to change how people discover brands and products and get information. And so even before it was searching the internet, ChatGPT was using its trained in knowledge, right? To you’re probably an early adopter of AI as well, and realize that, hey, it’s giving answers based on what it’s trained, and this is outdated, and where is it getting this information? [Michael Houck]: Yeah, it’s like before we were optimizing, at least in terms of SEO, people were optimizing for what do search engines like, right? And now it’s a whole different game. And so you have these much longer, more detailed syntax and queries that people are writing. And therefore, the long tail stuff makes more sense. But it’s like, well, how do you know which long tail to go after where you’re going to fit in? [Wil Armstrong]: And that’s, yeah. Exactly right. And actually, I’ll just flip right to this. Some of the brands we work with, certain brands of all shapes and sizes. Lenovo being a big customer to Penn State University to my mom’s local veterinary hospital in Maryland, which is not a logo on this slide. But I’ll note exactly what you’re hitting on is, you know, back in 1998, internet was created. It was optimized for humans. People are interacting with websites, but then for a good chunk, a couple of decades, people are just optimizing their websites for how, what Google wants and what Google is looking for playing that Google game. Now, AI, LLM, search is for the answer. It’s looking sometimes the second, third pages of Google to find the answer to someone’s question. And even AI overviews the search completely differently from how Google indexes sites. So it’s totally changing the game. And you see Google’s hand really being forced and all the products they’re trying to throw against the wall and see what sticks, AI mode, Gemini. [Michael Houck]: I’ve never seen so many TV commercials for Google before in the last two months that I’ve ever had in my we’re monitoring all that closely. [Wil Armstrong]: And really where AI is going is not just search. They want to dominate. They want to own the entire customer journey. Instead of someone going to Google, finding out which website to click on, it’ll highlight shopping as well. It’ll highlight the maps, of course, but AI, OpenAI and Google are both thinking about how they could keep you within their platform the entire journey. So how you could check out using AI or book a hotel, an Airbnb with AI in that conversational experience. So it makes total sense. [Michael Houck]: I mean, I think OpenAI is already even doing some like test partnership with Shopify. So, you know, it’s correct. Totally. [Wil Armstrong]: And Perplexity has a partnership with PayPal to accept transactions in Venmo and PayPal. So there’s a lot going on there. We’re trying to skate to where the puck is going. Hockey analogy because Stanley Cup just ended. [UNKNOWN_SPEAKER]: I’m a big hockey guy. [Wil Armstrong]: I was so sad Florida took it. Same here, same here. Trying to stay relevant. I guess that should be a basketball analogy. We’re trying to skate to where the puck is going and that’s shopping and that’s owning experience. It’s those agentic workflows, but it starts with monitoring. How is our brand showing up in chat GPT? How is it retrieving content from our website? Yeah. [Michael Houck]: I think one, one interesting thought that I had is like, I think right now we’re in this world where a lot of people are shifting from Google to discovering through these LLM chatbots, right. Whether it’s Claude or Grok or a chat GPT or, or whatever, but the, Interesting thing is that we might even soon, like in the next year or two, shift to a world where the majority of people aren’t even performing those searches themselves. They have agents that are doing it for them. And then it’s like, well, they’re just getting a digest from the agent. So really what we need to be optimizing for is something that an agent will like to confirm, more so than even maybe a chatbot. I don’t know. [Wil Armstrong]: So that’s my hypothesis. Yeah, exactly right. Such a workflow where you tell Siri, and I say this because Apple has long been working on their AI models and trying to make Siri more useful. So we’re monitoring that, but eventually you could see telling Siri, hey, every morning I want to get a digest of all the important news related to X category. And it will have different AI agents going out, retrieving content, what it thinks is useful, valuable to you, and serving that up. Conversation anymore, like you said, but a workflow, a task that you give it. [Michael Houck]: Yeah, I actually wrote a piece in our founder newsletter about this. We framed them as, you know, in the old worldly internet, like Ben Thompson from Stratecory always writes about aggregators, right? You’re aggregating a market, like Airbnb aggregates home rentals and things like that, Uber aggregates drivers. We’re shifting to this world where the new paradigm is arbiters. It’s these AI agents that are the arbiters. Of what gets through to the end user and vice versa. So I think that’s where the leverage is going to be. [Wil Armstrong]: That’s a new word that we’re going to have to add to this pitch deck here. Go for it. I love that. Well, that’s exactly right. And I’ll just note there’s some data around it being catching on. This is more a slide for folks that don’t understand the shift in the big enterprises. Yeah, I’m bought in. [UNKNOWN_SPEAKER]: I’m 100% bought in. [Wil Armstrong]: in for sure. I can tell. That’s the types of partners that we like to have is ones that want to do things about optimizing for AI. I’ll note on the educational side, the things to worry about now are not only the Google crawlers coming to a site to make sure it’s indexed in Google, but different platforms have different AI crawlers, ones that will build up their trained in knowledge like the the GPT bot, but then there’s also the GPT user crawler. So if you go to your robots.txt, you can see if you have any blocked. Some of the big enterprises a couple of years ago knew AI is coming and we don’t want them to train on our content. So they blocked all the crawlers. They blocked everything. And as a result, they don’t get cited. And now these models, it doesn’t go to their website for information. So this is the first thing that we look for is just, hey, do you have the crawlers blocked? Your site. If so, you might want to consider unblocking them. Now, it’s an interesting dilemma for the Forbes of the world, people who have content that they publish that are behind a paywall, right? Because you do sometimes want to have it crawlable and cited, but you don’t necessarily want it to access that content for free, right? There are some ways to allow it. One, you could block it from using your content as trained-in knowledge, but allow the live search crawler so it could come and view your content on demand. So some creative ways there, if you’re thinking about any of that. But just note that these crawlers are a big piece of what gets through to a user in ChatGBT. [Michael Houck]: Yeah, I think let’s shove it all in there, man. [Wil Armstrong]: Let’s let them take what they want, it’s fine. [Wil Armstrong]: Yeah, that’s the simple solution. And so we’re thinking about, yeah, a world where users are not actually going to your website, but they’re consuming your content, booking things, using your services, paying for products without ever having stepping foot on your website itself. [UNKNOWN_SPEAKER]: Makes total sense. [Wil Armstrong]: So I’ll just flip to a demo here where a few of the, products that we have, some of the tools that we have available, but it starts with monitoring, understanding, you know, how AI is retrieving our content, where is it going to for answers to questions that we think we should show up for, gets into bot monitoring, so we can monitor, we can tap into your web logs and see how often are these crawlers coming to our site, what information are they retrieving, what pages are those on, is it the trained in model crawler? Is it the live web search crawler? What platforms are coming to? So I can pull up ours, scrunch AIs, bot traffic, which is always interesting. Because you see in there, Apple has an AI crawler meta, of course, is coming to our site often. ChatGPT is most of it. [Michael Houck]: Is this like a Google and Bing thing? [Michael Houck]: Is ChatGPT 90% of it? [Wil Armstrong]: It’s a good question. It depends, I think, a little bit on geo. Definitely worldwide, GATPT is the largest consumer, largest referrer of traffic to websites. In the US, though, perplexity is also up there. Oh, really? Yeah. Perplexity is definitely not a majority by any means, but in terms of AI traffic or AI referrals, you see 25% to 30% of is coming to Perplexes, so it’s a big one. [Michael Houck]: That’s a lot higher than I would have thought. Interesting. [Wil Armstrong]: I guess they have more traction than I thought they did. Yeah, and I’ll just note, we can go into, there’s a tab here that just taps into Google Analytics data. So we can track, and you could do this on your own even, just set up a Looker Studio dashboard to see what kind of traffic you’re getting from the AI crawlers. This is an old website, just a demo. Or it didn’t get any traffic, but for some reason I send a couple people once in a while. But this is one of the features of the platform is just seeing, OK, where are we getting referrals from Here’s another one that I have with some built out. Built out data here, so conversion rates, you know, views versus sessions, what pages AI is sending folks to. We’re seeing that growth at like 40% month over month in terms of referrals from these AI platforms. So it’s definitely picking up. And even we see Cloud tapping it, putting its foot in the door in terms of web search. It’s not searching as often as a chat GBT, but if you explicitly asked it to search, it could do that. Same with crawl. But yeah, mostly chat GBT, a little bit perplexity in Gemini. [Michael Houck]: And how can I take advantage of this? Because I’m pretty bought into the idea of, OK, let’s try the platform. Let’s monitor. Let’s learn. Let’s see what prompts. Let’s see where we’re being returned, where we’re not. And let’s see what pages they’re going to. I’m pretty bought into that. That sounds smart. I should do that for sure. What should I be doing to make it so that we’re also seeing that growth in hits from these companies? Yeah, great question. [Wil Armstrong]: And we have some case studies that we’re working on. I’m working on, right. We were a team of three in February. Now we’re up to 13. So Oh, boy. Wow. That’s our exploding, right? Double digits. That’s crazy. [Michael Houck]: I mean, it was that 300% growth. That’s amazing. [Wil Armstrong]: We’re working on it. [Wil Armstrong]: We’re working on it. But of my job is case studies. And so I’m working with a handful of our customers in terms of what’s working, right? Like, how are they achieving so much success? One of them being this company called RunPod. They’re based in Provo, actually. They’re based in Utah. We’ve got a good relationship with them. They’re a software business. And they were getting five to 10 customers per month back in January, February, attributed to AI crawlers. We signed him up for Scrunch. They wanted to just understand, you know, the customer journey there. And they realized one of the big things they looked at was this dashboard to see, okay, how are we doing for, this is just a demo org here, but how are we doing for Google AI overviews? You can see he was running a handful of prompts. They knew his customers were asking or ones that he wanted to show up for. We’ll auto generate some prompts that you can track, ping them against the models, collect the responses. Show some data over time. So I’ll give you an example here. Here are a handful of prompts that we’re tracking for Disney Plus, just as an example. And you can see, okay, they’re getting cited 20% of the time in aggregate. Let me go. [Michael Houck]: So sorry, so being cited in this case means that 20% of the time someone writes a prompt about this idea, they are included in their response. [Wil Armstrong]: In a way, so if I’m thinking about a prompt, is there a streaming service with a user-friendly interface for kids? We can monitor how all these models are responding. We’ll run those prompts, the same prompt, against each model every three days. And so we’re collecting an aggregate view, you know, rolling averages of how Disney Plus is showing up. And the models, they’re non-deterministic, right? It might not be every time that they’re showing up for a specific prompt. I’ll find one that has some variability. [Michael Houck]: Because maybe they know different users who are asking that prompt to have different preferences. [Wil Armstrong]: Exactly right. Yeah. And so actually a big part of the platform is monitoring for different personas. So you can see how AI responds for a family-centric viewer versus a tired parent. Because memory is a huge piece of ChatGPT specifically. They’re leading the way in personalized responses. And so with that comes different geographies. You can monitor for different locations and geos here. In this default, we have one for France. So you can see, how does the French version of JetGBT and Perplexity and AI overviews respond to this person and these prompts versus the United States? So personas are a big piece. Exactly right. But we’ll monitor a prompt over time. So you can see, OK, we’re showing up at this point. Percentage, but maybe we’re not being cited. [Michael Houck]: So what is the difference between presence and citations, I guess? Yeah, good question. [Wil Armstrong]: So presence is meaning anytime you are mentioned in the response. Okay, we’ll go to new releases here, where most of these are at a filter for non branded. So we’re looking at prompts that don’t mention Disney plus by name. Any brand by name and each one of these prompts are run against five different models I can see they’re not cited in any of the responses meaning chat or perplexity or. I reviews isn’t going to their website for information but it’s still mentioning them in the response. [Michael Houck]: So presents is times like the percentage of time that I mentioned in the responses citations is like did it actually. Paying my site or not. [Wil Armstrong]: Exactly right. Yeah, got it. Okay, cool. [UNKNOWN_SPEAKER]: That’s helpful. [Wil Armstrong]: Here’s one. [Wil Armstrong]: Here’s a prompt. What are the best shows I can stream right now? It’s not going to Disney plus for any of these answers. But for perplexity and chat GPT does mention them in meta here, but it does mention them in the response. You can see the actual June 21 collection here where it’s to Forbes.com and Rotten Tomatoes, TV Guide. And it’s mentioning Ironheart is on DisneyPlus.com, but it got that information from TVGuide.com. Okay. [Michael Houck]: So being on these reputable sites and having essentially the equivalent of backlinks still matters. It doesn’t necessarily need the actual link itself. It just goes to the site aggregation. Yeah. [Wil Armstrong]: And so I think one of the biggest things for you would be this sources tab. Of understanding where is AI going to most frequently for answers? What are the specific pages? What are the domains? So it will give you an understanding of what’s used most frequently, what domains maybe we should partner with, the third parties, where should we learn from, or if it’s a competitor page that has the information that we wanna be cited for, I’ll go to one, this Disney Plus org, it particularly is going to some of these alternate pages like CNET.com, Business Insider. But if it’s a competitor page like Hulu, and even though they own Hulu, it’s going to their site because it’s more optimized for AI. It’s structured in such a way that AI can easily retrieve content from it. And a big piece is, one, HTML. Doesn’t read JavaScript right now. So if you have a website that has… I didn’t know that. OK, makes sense. Yeah, it doesn’t render JavaScript. So if a website that is very dynamic and beautiful, it might be elegant for the human visitor. But AI only reads the HTML. And so part of the platform is throwing in a page, and we’ll show you what AI sees on that page. That’s in the Site Audit tab. But this information is helpful to see, OK, what is AI going to, where is it going to for information? What prompts is it tied to? Is it favoring those citations more or those pages more or less over time? And you can click on them and see, for whatever reason, this page is optimized for AI. AI is going to it frequently. Some of that is the things that AI is looking for. It’s looking for pros and cons. It loves comparison type content. These are large language models. They love just language in general, blog posts do well. [Michael Houck]: Totally. Do you guys, do you guys have like, um, like, let’s say I sign up for the platform. I’m in there, I’m monitoring and I’m like, Oh shit, I should go create, you know, a thousand long tail pages in these areas. Right? Like, can I go to you guys? Do you have like a service level offering too? [Wil Armstrong]: Or is it just the, just the platform? We don’t, we don’t have, uh, the page. Creation piece. Granted, we do have a partnership with Daydream. And so they’re sort of talking that themselves, the programmatic SEO might be something to dip it to in the future. And actually, one piece that we are beta testing with a couple enterprises right now is the content generation piece, but certainly not something that we have available at scale right now. What we can do is we’ll look at scrunch data and scrunch the platform is looking for patterns in your response data. So we’ll suggest things like, there’s one for spirit that’s pretty good. A bunch of recommendations like, hey, you’re not showing up for this specific piece of content. Here’s a page that is frequently being cited and it could be updated. It says it’s from 2014. Or here’s an influential competitor page that is frequently being cited. To answer prompts and you aren’t. So consider mimicking that content, mimicking that page. How does it know who my competitors are? [Michael Houck]: Do I put that in, in like the settings or something? [Wil Armstrong]: Yeah, so you can do that in context. You can define competitors and we’ll just track anytime competitors are mentioned. But we also auto-generate some to get you started. [UNKNOWN_SPEAKER]: Okay, cool. [Wil Armstrong]: But yeah, you can always put in some competitors. And we’ll flag any time, like, hey, here’s a competitor that maybe you’re not tracking, and it’s frequently being referenced in the responses. Maybe you should start tracking them. Cool. I mean, that sounds great. [Michael Houck]: I would love it. Yeah. I think there’s like three or four different ways that you’ve mentioned that we could get some serious value from this. [Wil Armstrong]: And I’d say people, and I know we only have a couple of minutes left here, but I think customers spend the most time in this sources tab to see where is AI going to, Could we earn a citation? Should we optimize a page to look like this competitor page? [Michael Houck]: Yeah, that makes sense. And percentage seen there? What does that mean? Like a citation, basically? [Wil Armstrong]: It’s basically saying, for these prompts that we’re running, it’s citing that page in 30% of the responses. Wow, OK. So there is a lot of variability there. Oftentimes, you’re not going to see 100% seen. Each model has different favorites that it plays, like Google AI Overviews. I mentioned this RunPod company earlier. They really wanted to optimize for Google AI Overviews. And they noticed that Read.AI was a massive source of information for it. So they started posting on Read.AI, creating threads on Read.AI. And now they’re up to, I mentioned, 5 to 10 customers per month in February. Now they’re up to 1,500-plus customers coming from AI sources, mostly AI overviews today. And so that’s a case study that we’re working on. A ton of success there. But it came from this Sources tab, seeing that, oh, it’s Google AI overviews is going to Read.AI for whatever reason. Yeah, I mean, keep me in the loop on those case studies. [Michael Houck]: We’d love to take a look at them as you guys publish them. Yeah, I think we’ll probably sign up and get a read just on where things are right now. My guess is that we’re not doing very well right now because we under-indexed on SEO. So we’re probably not showing up very well in this world either. We probably will want to build out a bunch of just like content pages that folks can find. Do you guys have a preferred partner who is, you know, maybe, you know, because I think that with Daydream, they were saying that, you know, they’re just like series B or above companies, and we’re not venture back, we’re bootstrap. So any preferred partner for this kind of stuff who understands the AI chatbot side of it more so than the traditional programmatic SEO side of it that would be a good fit? [Wil Armstrong]: Yeah, we definitely have some agencies. That’s a big piece that we lean into is agency partners. So we definitely have some folks that are smaller and deal with smaller brands that I would happily refer you to. [Michael Houck]: Yeah, that’d be great if you could send over a rack or even a couple that I I could choose from, that’d be great. Because I think that would tie us in more to getting continual value from the platform if we actually are making a conscious effort to drive those results and improve them. [Wil Armstrong]: And then one thing, you could loop them in as well, give them access to Scrunch. So they’re looking at data and have a full circle of information there. [Michael Houck]: That way we can orient them around the same things we’re orienting around ourselves. So that makes total sense. [Wil Armstrong]: Exactly. Yeah. Follow up. I’ll send you just some, some checkout links if you want to go forward and some wrecks there on the agency side. Yeah. I really appreciate the time. [Michael Houck]: And I feel honored that you, you did this one yourself. So I appreciate it. Um, but yeah, let’s, let’s do that. Um, and then we’ll get started sort of on both sides, probably simultaneously. And, um, I can go from there. [Wil Armstrong]: Awesome. Sounds great. Great to meet Michael. [UNKNOWN_SPEAKER]: Cool. All right. [UNKNOWN_SPEAKER]: Take care. [UNKNOWN_SPEAKER]: Thank you.