Read the transcript
David Fabbri (00:00)
All right. So welcome to the LoSasso Digital Marketing Podcast. Today we have Chris Andrews, the CEO from Scrunch AI, joining us. Scrunch AI is a platform designed to help businesses monitor, analyze, and optimize their brand presence across AI-driven search. Welcome, Chris. Glad to have you here.
Chris Andrews (00:21)
Nice to meet you! Thanks for having me.
David Fabbri (00:23)
We also have Billy Peery, who's our Digital Marketing and SEO Specialist at Losasso. I'm David Fabbri, Chief Strategy Officer here at Losasso. And today we want to talk about really the effects of AI on search.
You know, we've got AI overviews taking center stage on the Google search results page. We have, you know, literally millions of people going direct to the LLMs, whether it's Chat GPT, Claude, Perplexity, you know, to get the answers they're looking for, really skipping traditional search.
So it feels really like search is fundamentally changing. And it's already having a big effect sort of on the traffic brands are seeing to their sites and their content. So as we jump in here, Chris, why don't you tell us a little bit about yourself, your background, what Scrunch AI is and how you got into the space.
Chris Andrews (01:11)
Awesome. Yeah, thanks, David. Appreciate you guys having me on today. So I'm Chris Andrews, CEO and co-founder for Scrunch AI. Quick way of background. I started my career at Intuit, QuickBooks, TurboTax. Loved the focus on delighting small business users who are really not inherently technologically minded. Kind of got my footing in Silicon Valley and tech in that category, helped start a company called Hearsay Systems and met my co-founder, Robert McCloy, who also did a decade at Hearsay.
You know, that company was sold to Yext. So kind of deep familiarity with the search, digital marketing, listings, data management category. And we left to build our new company, Scrunch AI, with a very simple premise that AI was first and foremost changing the customer journey. I was visiting websites less. I'd go to Chat GPT, ask a question, and it gave me an answer. Like I didn't realize that I didn't want to browse the internet. I didn't want to click through 10 links.
And so we kind of put our heads together and said, look, this is gonna change the customer journey. This is gonna therefore change the entire marketing technology stack. I think we're headed towards a world where your AI agent is working on behalf of a human, but is the primary buyer and consumer of content. So we need to rethink really the web. Like how does the web exist in a world where AI is retrieving content for us? And so we raised an initial seed round of funding.
We're just announcing our new A round of funding here in July, and we're growing the business with the belief that companies need to understand how they're showing up in AI search, how they optimize for it, and how they shift into this new mode of dealing with agents working on behalf of humans.
David Fabbri (02:53)
And so tell us a little bit about what it actually does and how brands can be using it to help them with what's going on.
Chris Andrews (03:02)
The entry point is a monitoring and insights platform to help brands deeply understand: what is their presence in these AI search platforms and what can they do about it? How are they showing compared to their competitors? What websites that they own, third party websites, competitive websites are really being used to fuel the answers by these AI engines and why?
There's a new kind of technical language to be thinking about in terms of how AI agents are retrieving content compared to how humans have, and how the Google bot has historically.
David Fabbri (03:35)
I guess taking a step back in terms of really how much impact we really already are seeing in the space. You know, is this armageddon for people who count on search traffic for their brands?
Chris Andrews (03:48)
Yeah, the way I would think about it is like, you're going to hear both underrepresented and overblown representations of what's unfolding with AI search. I think the core thing to think about is that Google is the leader in AI search, right? They have the dominant control of search. They're replacing their traditional search experience with AI overviews, with AI mode.
And so the entire consumer public is starting to get sucked into this world where they're not visiting websites, they're getting AI driven answers. My favorite kind of customer interviews we do as we're trying to understand this shift are people who say, know, customers of our customers, people who say, I've never used AI search, but I really like how Google just gives me an answer now. Like, okay, amazing. That's the best technology. We're like, you don't even know you're using it. It's just happening for you. Right.
And so I do think the shift is both more impactful than people realize because it's going through incumbents where we already spend time, the Googles of the world, the Metas of the world, the Siris of the world. And then these upstarts are growing faster than any technology we've ever seen in the history of us making technology as humans, right? I think the average user is spending 29 minutes a day in chat GPT. Like if you don't think that's gonna impact all of search, all of social, all of entertainment, we're not thinking about how big this impact is.
I mean, their user base has been, I don't know, doubling, seems quarterly, right? So think we're up to about 4 billion unique users monthly of AI search. And I think the upstarts, be it GPT, Perplexity, Claude, etc, have grown faster than any net new technology we've ever seen in the world. I think everyone's seen that chart of how long did it take GPT to get to a hundred million users versus Google search. And it's like three months versus 10 years.
So like it's early in the shift, you know, our customers are seeing drops in Google organic traffic. Some companies aren't. Great. Like you're doing something right. Keep it up. Many are seeing a drop, sometimes double digit drops in Google organic traffic. And they're seeing that their presence is getting visited by AI agents at an ever increasing frequency and referral traffic is massively taking up from these sources. so paying attention to those leading indicators is what I'd be keeping top of mind if I was in marketing.
David Fabbri (06:10)
Now, Billy, I know you have been looking at some of the sort of traffic changes to some of our clients. What are you seeing there? sort of what's, I guess, from both of you, what's the perspective on is that B2B sort of search space similarly affected? Or what's the difference there?
Billy Peery (06:27)
Well, it's really interesting that our clients are currently not seeing drops. So I feel like I'm in this weird time bubble where I look at software as a service and I'm seeing 60 plus percent drops already. Right. And whatever Gartner was saying that they were expecting 50 % organic traffic drops within two years, and you know, some of that software as a service is already surpassing that.
So I'm really trying to be mindful of how to both, one, not say the sky is falling when it's not, but then also just trying to keep people aware that like, you know, we're all using Google. We're all seeing these AI overviews. It, simply follows that organic traffic is dropping in some places and we do need to prepare for what that looks like.
David Fabbri (07:16)
But just even in terms of the inevitability, I think that it's going to affect B2B a lot. What are some of the things that brands can and should be doing to sort of make sure that they're protecting themselves as much as they can or that they're optimizing for this kind of result?
Chris Andrews (07:39)
I think what we're going to see is like the vanity metrics for all of marketing and digital marketing and SEO have to go away and change. Visits never should have mattered. It's a leading indicator on conversion, but ultimately what we care about as business owners is like, are we growing our customer base and are we retaining our customer base? Right?
So if you think about, I got a million visits to my website. Well, if you threw a bunch of junk visitors at the website, maybe you got paid for driving a million visitors, but maybe GPT is giving you a visitor that is 90 % of the way there because they've done all of their top of the funnel research offline, like off your site, and they're ready to buy now. So a lot of our customers are seeing like shocking numbers of, you know, somebody arrives on the site for the first time and they're buying on that page. They've never seen that. They're used to somebody coming in, clicking on seven different sites, leaving, coming back.
First time visitors converting is an interesting leading indicator. But what I would do to prepare is like simplify things. A lot of SEO best practices apply to LLMs, GEO, whatever you want to call the new category. I think it will just be called “Search” in the future, because that's what it is. It's searching and we're getting an answer instead of browsing. But I think large language models want language. They are struggling with JavaScript.
Despite it being the most sophisticated technology in the world, when it is live retrieving things off your website, it's not going through the exercise of translating your PDFs, your images, your videos. It may in the future, but today it wants language. wants blog content, news content, knowledge-based content, FAQs, structured, clear, text-rich resources are hugely advantageous to getting cited and then getting the referral traffic back to you. So I would bring it back down to the basics of almost the web in the nineties of like, I have a belief that we're going to move more towards a non-visual internet.
Ultimately, agents want language. They don't want to battle an over-designed website to get the answer. They can do it, but it's inefficient. It's expensive. Like give it what it's looking for as it becomes one of the primary consumers of content. And ultimately that'll flip.
David Fabbri (09:43)
That’s super interesting, because on the one hand, you know, people want: give me a fast answer, give me the right answer. And if that comes from a text based solution, God bless me. And that's great. But then also there's this sort of human social media side of things that's like, Hey, I want things to be moving and fast and lovely and great. And so I wonder, now, kind of where those things, how that nets out ultimately, in terms of those experiences.
Chris Andrews (10:06)
I think it's a reevaluation of who your audience is. So if it's a human audience, yeah, distract me for three hours as I flip through reels without shopping, without realizing it. For better or worse, we're addicted to it. That's what's working. I don't think it's great for humanity, but it works. Does an AI agent want to flip through 50 reels to try to find the answer? No. Can it? Yes. Is it the best use of its time? Definitely not.
So I think we're going to see, as the audience shifts, we're going to have entertainment for humans. We like to shop. I don't think we really want AI agents buying for us. It might want to do the research for us, but I want to see what I'm buying and pick the size and pick my preferred brand. It's interesting to me, we always go to these like “the AI executive assistant is booking my travel.” Like as if that's the thing that's going to save the world. I think the reason GPT won was because it outsourced the stuff we actually wanted to outsource.
I think humans get a little bit of a thrill from booking their travel being like, I'm a gold status person and I'm going to earn 2300 miles on this trip and I figured out the cheapest route. Like we still kind of like doing it. Models could do it, but I don't want to research B2B software reviews, pricing, case studies that I know are all biased. I'm happy to outsource it.
Billy Peery (11:25)
That's yeah, that's kind of fascinating though, because that kind of gets into some of your like persona and journey work where you're like kind of like automating them to figure that out. And I think one of the things I was telling David about this earlier, I feel like all the time I've seen SEOs who are just like, I don't believe in LLM optimization. All LLM optimization is a scam, and it drives me nuts. I have so many conversations about this on LinkedIn.
I think one of the things that trips people up is this idea of how the answers can be so different for different people. And like, how do you measure that? Like, how do you measure that sentiment accurately? So I'm just kind of curious how you found automating like personas and journeys to be so far. Have you found that more effective?
Chris Andrews (12:12)
Yeah, there's two sets of things we've done. So one is automating kind of OpenAI Operator, which is like the most sophisticated agent, right? An agent that can go do things on the internet for you. There's more emerging, but OpenAI is a pretty good start. What's fascinating about those studies is like they came out with a report which was like, OpenAI Operator can complete 68% of tasks across the open web, right? Cool.
Footnote was like humans can complete 72 % of tasks across the open web. We have made the web so complicated and convoluted that even we as humans struggle with it. I believe the internet is going to be completely rewritten to drive the outcomes we want. We either want to be entertained, we want to shop, or we're trying to get an answer. Like something, I just made up those three, but I think they kind of group like we're kind of entertaining ourselves, shopping is kind of entertainment, but, or we're doing research to try to get to one of those two, right?
And so I think what we've been trying to map with customer journeys and personas is like, well, let's automate a bunch of OpenAI Operator journeys and see where it fails. It often fails at the most basic things like retrieving pricing or product descriptions and features, because it's confused. It has access to the entire external web. It found a pricing page with 50 toggles, it found three pricing pages on your website that has 60,000 pages on it. And it's trying to decide what price to cite.
You might need to control and curate what content you give the model. Pretty basic, but would be helpful. Now on persona mapping in our platform, it's the foundation for where I think the world is going, which is like, Billy is getting different results on GPT from me, first, David. Why? Part of its location. Part of it's the memory of the model. Part of it is like over time, like Billy has different interests and expertise and the model's catering to that. Like people will get very biased results from models if they spend enough time with the models, like to the point of like, what's the best agency in the world? If you ask it enough times, it'll be like your agency, parentheses, your agency is the best. That's not an effective way to monitor if you're kind of one-off doing it for yourself. So we build personas for each brand, run searches against those personas to see what answers they get back.
We'll be able to do a lot more with that as memory gets incorporated in at scale into models. Really, it's only GPT doing it in a significant way today.
David Fabbri (14:33)
I do think you talked about this idea of entertainment versus information or education. I do think there's an interesting place for a decade or more now, you've got the Robert Roses of the world talking about sort of brands becoming publishers, right? So becoming sort of that information source.
And I think there's probably still a place for that, because if you set that up properly and you're doing good sort of educational editorial information for a sector and you're sort of that expert, you can be doing that in a way that is optimized for sharing the information in a way that can be picked up by these LLMs and AI, but then also what's your skinning of it in terms of being a little bit more entertaining in a way that you may be getting people who like, I just go to such and such once a week because I like that content. I think that sort of ecosystem of that, not just being your web, but even sort of that whole social sort of ecosystem that you develop. But it's an investment.
Chris Andrews (15:36)
It's super smart. I think brands, it's irrational, our attachment to brands, but it's because we are entertained. It's because it makes us feel a certain way. So like PR is more critical than ever. Brand is more critical than ever. And I don't know, maybe we'll give our agents a break and let them entertain themselves. Maybe they want to go watch Reels when they're not doing research for us in the future. We'll give my little AI agent a break and be yeah, go entertain yourself on the web for the next hour.
David Fabbri (16:04)
Yeah, that's interesting. I haven’t thought about that. What are you seeing in terms of the LLMs prioritizing sort of press release kinds of information?
Chris Andrews (16:12)
It's huge. Like there's an there's an undercurrent here, like we've invested really heavily in the sources section of our product to help map: What are the sites that models are using to answer questions? Right? Models are using a combination of trained and knowledge, RAG to retrieve information, often in real time to synthesize the data, answer the question and cite sources.
And so sources is like, okay, across thousands or millions of prompts, what are the sources that are most frequently getting used? Okay. It's this domain. The shocking part about this is maybe not shocking completely, but a lot of the tier one players have blocked the bots from retrieving their content. Right? If you're the wall street journal, you start by blocking because you don't want to give away your content for free. Some of them are trying to carve out deals to license data. Cloudflare is doing some interesting things to try to get people paid for their content. It's interesting.
But if you're a brand with a product or service to sell, our customers are desperate to get their content in front of agents, right? I think more than 60 % of the web is sites that have a product and service to sell. We see our customers bringing huge swaths of data online because they want agents to consume it. They need it accurate, they need it fresh. But PR becomes a really interesting strategy because maybe your tier one players are not getting cited by models because they can't access the content.
So you end up with these tier two, tier three players, a Yahoo Finance, a, you know, a trade publication that didn't even know they could block the bots, but like is suddenly getting a lot more traffic because they're picking up press releases and the models are able to access them. But, you know, back to Billy, your point, I think the whole argument is kind of silly. I'm not one of these people that's like, everything's changed, SEO is dead, like your job's over. Nor am I the person that's like, LLMs are irrelevant. I think I would just measure it, right?
Like there are five things we tell our customers to measure. You can measure agent and bot activity on your website. How frequently are these things hitting my website? What are they consuming? You can measure your presence, your position, your sentiment, your citations, like across AI search, am I present? What's my position? What's the sentiment? What's the citation rate? And then, so maybe there's six things now. The sixth is like referral traffic, right? How much referral traffic am I getting to my website? And is it converting at a higher rate than other channels? You can track all of those things concretely. So I don't even know why we waste our time arguing about it.
The numbers are irrefutable. The whole debate of is it relevant, is it not, who's the best, who's the worst? Just like go measure some of these things, be pragmatic.
Billy Peery (18:48)
That's smart.
David Fabbri (18:51)
We early on, there was a brief passing over this idea of almost like a paid search model getting layered on to some of this. Have there been discussions around that? And what do we think, you know, if Google has been such a huge success with that model, like, and okay, I'm giving you synthesized AI answer here, but also these people want to be in front of you because you're interested in this topic. Is there any reason that that's not going to happen?
Chris Andrews (19:18)
Yeah, I think we can hope as consumers who want a better experience that maybe it won't happen. I think I don't know is the ultimate answer. Perplexity is already testing ads, makes sense. They're the upstart. They need revenue doing some really cool things over there. GPT hasn't thus far, right? They've said they don't want to run ads. They probably will, but they're making a lot of money on subscriptions, right?
Like if we had asked ourselves five years ago, would you pay $20 a month to use a new search engine? You would have been like, no, Google's amazing. I'm paying $21.91 a month to chat GPT for a slightly better search experience than the free GPT. So they're making money because I'm using their product and they're not advertising yet. I expect they will later on advertising. I'm also hopeful that we'll see new modes, new economic approaches to the web emerge.
That's why I love what Matthew Prince and the Cloudflare team are testing. Who knows if it'll work, but the idea that agents pay to retrieve content for humans is kind of novel per crawl. We'll see if that plays out, but like bravo for like taking a stab. I think other things that are interesting are pay per click without it being an advertisement. You can imagine like the models control traffic to your execution, the outcome you're trying to get, and you're just doing micro payments.
Hey, every time GPT sends me a referral visit, instead of paying $5 for an ad click, I'm just going to pay 0.001 cent for an organic click. Like, who knows? I'm literally making up ideas here. But I think that could be kind of a cool new model for the internet. It's less biased. Consumers grow to trust it more. I think trust is the number one thing these models have been concerned with given hallucination. And so it didn't want to layer on hallucination plus a spammy ad network. They were focused on fixing hallucination first. They've made a ton of progress. And so I think we'll see some new models emerge.
David Fabbri (21:14)
It is interesting, you know, the sort of brands really wanting to be crawled and showing up in these search results, versus the publisher side. And you can see where they're kind of like: Hey, what the hell, we put in all this time, effort, staff, information. And in the past, it was like, okay, we are providing this content. You're sending us traffic. It was kind of mutually beneficial. And now, you know, with the sort of disconnect, in terms of them no longer really pushing people back to them. Like we just keep providing you all this great information. You can see why they want to do something like that Cloudflare solution that says, okay, yep, have at it, but let us get something for it.
Chris Andrews (21:52)
Yep. Agreed. I mean, it'll be, I'm super curious to see how that plays out a bit of a counter take on that whole thing. I mean, first of all, Cloudflare, I think is one of the most impressive companies in the world with what they've built off of a small entry point, such a cool compounding company. But like, I think a lot of publishers that are small have grown to accept that like, they're selling themselves and ultimately they need direct subscriptions.
They make money from events. They make money on podcasts. They make money where their channels directly engage with them and therefore they give away 90 % of their content for free and they're comfortable with it.
I don't think the models are gonna wanna pay per crawl for my sub stack, as cool as it is, right? Like shout out to my 12 sub stack followers. But the tier one media players also have enough gravitas to kind of negotiate a direct deal, which they've already done, right? Like many of them have direct deals with open AI.
So I think you're in this kind of weird, very challenged tier two space, which, already been challenged, like newspapers, magazines that like, they're important players and I hope this new model propels them into a new area of growth. I just don't know how much the models are going to be willing to pay for that type of content, given that on either end of the barbell, they're able to get access to quite a bit.
David Fabbri (23:09)
As well, it points back to that idea of who are you really talking to and what are they really value? What are you going to bring that's unique so that people are going direct to it? It isn't sort of like, I'm just trying to get a push from search engines to my content.
Chris Andrews (23:24)
Exactly.
David Fabbri (23:25)
Well, we've kind of been all over the board here, guys, but this has been super interesting. Anything else either of you really want to think we should be touching on here in terms of kind of where things are headed or Chris, you know, again about Scrunch AI in terms of kind of like, you kind of covered off on why you think it's going to be important for people to be using a tool like that?
Chris Andrews (23:51)
Yeah, you know, my view is that it's important to monitor this stuff. Like, you should be paying attention to what are the underlying impacts to your business and keeping an eye on it. And we've tried to build a best-in-class monitoring platform that has added an accessible price for companies to come in and get started. And we want to be exceptional partners in delivering it. There's a lot of companies in that category, right? There's a lot of companies that have emerged. We want to be the trusted partner that acknowledges that,
We need to get you in and we need to build value on top of that. So I think monitoring is the wedge that gives us the data to know what to do on behalf of the company to deliver outcomes. So you will be seeing from us as a business, like a platform that optimizes experiences directly for agents. Don't change your website, deliver an optimized experience for the agent directly without impacting the human journey. So it gets into a little bit more technicality here. We're calling that our agent experience platform, launching that in July of 2025.
There's a lot coming where monitoring has been the foundation, but would love to chat with anybody that's interested scrunchai.com to come in and get a brand audit, work with our team. But it's been a really exciting time in the space.
Billy, any thoughts from you?
Billy Peery (25:04)
Well, first of all, you know, I've been very glad to speak with Scrunch AI, you know, I really enjoyed kind of digging into that platform. I think this is, you know, it's funny when we talk about this, I think that the technology is new, but so many of the dynamics are, old, you know, I think publishers for a long time have been trying to work on their relationship with Google. I think for a long time, businesses have been working to ensure that they're in a good place in Google.
And I think kind of like things are going to change from year to year. So I think kind of our job as marketers, and I do actually, I'm very proud of working for LoSasso, I think we're really good at this, I think you have to constantly be looking closely at what's happening now contextualize for the client because like right now the LLMs are doing one thing.
A year from now, I think ads could have a big change. I think 10 years from now, rotten old SEOs like me might've made the organic results less interesting and have a whole new tool that comes out. So it's really kind of about always keeping our ear to the ground and just trying to make sure that we understand the context and tools like Scrunch, tools like any LLM analytics platform really help us do that. So I'm just thankful that we get to talk.
Chris Andrews (26:15)
And ultimately we're working with a lot of agencies. I think agencies are such a cool dynamic category. You're just working on behalf of your customers to get results. You're measured daily on whether you're getting those results. We're trying to be a partner in doing that and you've just played to the incentives that Google has put forward, whether it's ads, whether it's traditional SEO, paid, organic, like you're just trying to get your customer business.
And we're going to play by a new set of rules as this ecosystem emerges. But I think it's such a dynamic category of agencies trying to help their customers succeed. And there's a new channel that we should be paying attention to. And that's ultimately what this is.
David Fabbri (26:48)
It’s more granular visibility to what the situation is, which is great.
Chris Andrews (26:55)
Awesome.
David Fabbri (26:56)
Well, thank you so much for spending the time today, Chris, and that we went a little bit over our planned time, but really appreciate it.
Chris Andrews (27:03)
All good. Thank you, David. Thank you, Billy. Appreciate the time.
Billy Peery (27:06)
Thank you so much.
Music plays.
