Winning in the Age of Clickless Search with Brian Stempeck

Welcome to another episode of AI Loves PR. We know the AI revolution is transforming business and our industry's capabilities are more important than ever. In every episode, we're bringing you practical insights from leading PR minds who are navigating the AI landscape. This isn't just a passing trend, it's our chance to lead, listen in for innovative ways to integrate AI into your work and ensure that our industry leads the charge. Hey everybody, welcome back to AI Loves PR. I'm here today with Brian Stimpeck, the CEO of a platform called Evertune. I was super excited to learn about its capabilities, which could not be more relevant for the moment we're in and moving forward. So thank you, Brian, for joining. Thanks for having me. Can you give us just a quick overview of Evertune and then we'll dig in. Absolutely. So we started Evertune about a year and a half ago because we saw that search was changing. So you've probably seen, if you're in Google, used to be you get 10 blue links. Now you get an AI summary, summarizing those results at the top. And then a lot of people have started to adopt AI tools like JetGBT, Claude, Perplexity and others, which are a new form of search. Instead of having a link that you click and you go read a story or an article about that subject, now it just gives you the answer. So people are calling it click list search. And so we work with brands and agencies to help them understand, how is your brand showing up in AI search? Are you being mentioned or not? If not, why not? And then what do you do about it? So what's the new equivalent of SEO? How do you use PR and communications and content to actually educate and influence the AI models to recommend your brand or know more about your brand to talk about it more often? So cool. I think that clients and their agencies on behalf of those clients started doing those LLM searches on the companies and brands. And so they'd find out what they were saying, but they didn't really know what they could do to shape it. How do you diagnose what's not working and then how to fix it? Yeah. So the biggest thing to understand is that if you and I went to Google and we said, "What's the best tennis racket to buy?" It'll give us pretty much the same responses. It'll give us each the same 10 blue links. AI does not work that way. So AI models are probabilistic. Think of them as autocomplete. They're trying to guess the next word in every sentence over and over again. They've gotten pretty good at guessing what the next word should be. And so as a result, if you have 100 people search for the best tennis racket, all 100 people are going to get a different response to that. In some cases, it'll recommend prints, in other cases, it'll recommend Yonex, it'll recommend different brands, have a different point of view. And that's just a natural way that the models are built. So in order to understand what's happening, what we have to do is think about it like a large-scale survey or poll. We're actually asking tens of thousands of questions to each AI model to understand. If you ask it once or twice, you may not know, but with a large enough sample size, you can understand which brands really rise to the top. Okay, Wilson tennis rackets are being recommended 92% of the time. They're doing really well. Babylon, another tennis brand, they're only being recommended 20% of the time. Why is that? So we start to dig into what are the websites, what are the publishers, what are the press stories, what are the sources that the AI model used to get to that information? And the AI models are black box. They don't make that easy to figure out. In some cases, they give sources or citations. And so when we run tens of thousands of searches, we can aggregate that information. We can say, okay, it's this tennis website, it's this magazine, it's this influencer, it's this Reddit thread. Those are the things we're seeing come up over and over again. And if you're a PR comms professional, I think the good thing here is that to influence AI requires PR and communications. It takes earned media to actually have an impact unlike SEO, which is more so about the brand's own website and putting content there that helps with AI, but that they are leverageable is more authoritative sites, external points of view about the brand, about the category. And so what we've tried to do is identify what those are so that the PR comms content teams can then take action on that. And it's really cool too because it may not be the typical earned targets that our clients would have in mind. And I guess part of that is because of the licensing and usage agreements, but you've really got to have a whole additional set of earned targets, right? That's right. Yeah. What we'll do in our platform is we give out a list to clients when they run a report, so they will stay with tennis rackets. What are the sites that are coming up? And we'll share with them about 100 different domains and URLs that they'll see there. And some of those, as you'd expect, are publishers and sites that they're already working with, they're already talking to, the large endemic publishers in that category. But then there are blogs that show up. There are YouTube videos that show up. There are Reddit threads that show up. We'll see lots of different articles and links in places that maybe weren't on the radar before. And that creates a really interesting opportunity because typically those are smaller publishers that can be a bit more long tail. What AI models tend to gravitate towards is deep expertise on a niche subject. So not just understanding tennis rackets, saying, "What's the best tennis racket that I want to buy for a slicer, for playing at a clay court, or a more specific search?" Because when consumers use AI, it's different from Google. They don't just go in and say, "Best tennis racket." They do that sometimes. But a lot of times they give context. They say, "I'm a player in my 40s, I'm not as good as I used to be, and my top spin's not as good. I'm mostly playing on hard courts in the summertime." And they give context about how they're playing and where they're playing so the model can give you a very precise recommendation. And when you ask questions like that to the model, it's looking for specific information and that's not always on the major endemic publishers, that can be more on niche content. I think the broader trend here is that by doing this, you have a way to get to data driven communications. And I think that's happened in the advertising world. A lot of advertising is done based on cookies, and IDs, and a lot of data matching on consumers. In this sense, what you're trying to do is figure out where should you run content to best influence the AI model, because that AI model is now going to influence hundreds of millions of people to a super influencer. And then what we also help clients do is figure out what to say. What is the AI model already think about your brand, and where can you reinforce that? Where does it have blank spaces where maybe it doesn't know a lot about the best tennis records for clay courts, and you're going to have to make a recommendation there or make some content to fill that in. But we surface those ideas for content as a starting point for the brands and the agencies that we work with. Yeah. I love that example you shared with me earlier about Canada Goose. Yeah. Really cool case. Yeah. That was one that was an early deck. We looked at the winner coat category. And so part of what we try to figure out is what do the AI models think is important in any given product? And so in winner coats, if you look at the broad category, the models come back and say, what's important is how warm the coat is, how waterproof, things like that. And then we can have them evaluate every brand and how it performs. Some brands did really well on waterproof. Other brands like Canada Goose have not done as well. And that's because those coats are more water resistant than truly waterproof, like a shell that you might have in Patagonia. That's a very nuanced area. You could argue that's pretty much a waterproof coat, but that's an area where the models don't have deep understanding. It's a content opportunity for using social media, using influencers, using PR communications to try to build out more information on that subject area. That's really cool. What do you think is most important for agencies and our discipline to take advantage of this AI transformation? I think the most important thing right now is to understand the scale with which this is happening. I think sometimes there's a tendency when new technology comes out to say, let's wait and see. Are these just the early adopters using AI? I think we're actually well past that phase, even if you said, okay, this is my grandmother. She's never going to download Claude. She's never going to download JetGBT, but she might use a computer. She uses Google. She uses Amazon. She might have social media. All of those technology platforms have now integrated AI search into what they do. So if you search for something on WhatsApp or on Instagram or Facebook, it now plugs into the meta AI model, which is called Lama. If you search for something on Amazon, they have an internal model called Rufus, which does AI summaries and AI questions about every single product on Amazon. If you ask Alexa a question with a voice assistant, it plugs into Claude, another AI model. Apple Siri plugs into JetGBT, and so whether someone knows they're using AI search or not doesn't really matter. This is already mainstream because the tech companies have made it mainstream. And so I don't know if enough people fully appreciate how fast this change has happened, that in the last year and a half to two years, we went from search 1.0 to 2.0, and it has a big impact. We're seeing a lot of clients now where they're seeing this change the number of visitors to their website. The amount of organic or paid search traffic is going down, publishers are seeing decreases in traffic, but their information is being picked up more, and AI companies are paying to license that data from some of the larger publishers. So I think understanding the scale of that change is important, and then starting to experiment. This is very early space. There's a lot to understand. The models are changing frequently. They retrain the models every couple of months. They release new versions. There's increased fragmentation. This isn't just a Google world. There's meta. There's Amazon. There's Claude. There's JetGBT. And so that idea of experimentation with how are these models talking about a brand or a client? This is a net new area, and I think it makes it pretty exciting, too. It makes the place where you can experiment and test out what works, and that's what we often are helping clients do. Where do you think PR professionals should start? Is it just in the experimentation, check out as many LLMs as possible, see what they're saying about your clients? I think it's that. And then starting to use what we see today as we'll have clients where we'll do a contract with a brand. They'll be using our platform to measure how they're showing up in the AI models. And then what we help them do is get that list of here's the hundred websites to focus on with content, and then here's what to talk about. Increasingly we're seeing brands loop in their PR agency into that conversation, because the brand likes the analytics side of it. They want to measure the brand health metric. They want to see the impact on lower funnel with sales and visits to the site, everything like that. But when they go to take action on it, they'll produce content for their own site, but then they need help to go figure out where to go next. And so some of the more pioneering PR agencies we're seeing are using our tools to deeply understand in each of these categories, what should you talk about? Where are those blank spaces that the AI models don't know a lot? Whether it's winter coats, whether it's tennis rackets, whether it's software, there's lots of areas. This is kind of like Google was in 2000 or 2001. It's the early days of AI search. The models are extremely advanced. They ingested tons of information. That doesn't mean they know everything. There's a lot of information that's behind company paywalls or publisher paywalls, or it sits outside the models, and that creates an opportunity to go fill that blank space. And so that's what we're seeing the more advanced PR agencies do now, is identify what do we talk about and where are the publishers we should publish that to try to get that into the models as fast as we can. It's what we do best, right? That's right. What's next for Evertune? Where are you focusing for the future? So right now, mostly just scaling. We're about a 20-person company today. We've grown significantly. We work mostly with larger enterprise, global brands, and agencies. So we're just growing super fast, adding more people to the team. And then from a product standpoint, what we're doing is focusing a lot more on insights to the clients, but then also more action steps. How do we get more precise about what type of ideas they have for content? Can we help PR agencies actually generate some of that content using generative AI tools? And I think what a lot of brands want to do is because the space moves so fast, they think a lot about distribution. How do I push that content out faster? And so as a company, we think about where can we do integrations to help push this, whether it's with different press release type publishing tools, whether it's with affiliate publisher partnerships, with social media influencer platforms. That distribution question, that's the hardest piece to solve. And I think that's where the biggest opportunity is for PR agencies to help scale that. Because in today's world of AI search, volume matters. The amount of word count you have, the models, what they're doing is trying to associate, how should we think about the Wilson tennis brand? What words and descriptors do we most closely associate with that? And part of the answer to that challenge is how much volume is there on the internet and other places talking about what Wilson tennis records do best. And so that's a component where clients of ours are thinking a lot about scale and trying to help them meet that challenge. That's so interesting. I've been thinking a lot about this proliferation of content that's just going to speed up and how overwhelming it is, how it's impossible that this content can be consumed by the right people at the right time, all the time. But that makes sense, that the AI is going to consume it and digest it and present it back in a way that people want it. Yeah. And it's going to start to summarize it as well. The other place we're seeing a lot of information or a lot of interest from clients is, this is not as well known, but the AI models themselves are starting to offer shopping and commerce straight in the platform. So if you're in chat, and you look for, hey, what's the best coffee maker under $100? It'll now not only present the different brands like a Mr. Coffee or a Keurig or whatever else. It'll also present the opportunity to click and then go buy that from a retailer. And so over time, what we're seeing is the sales funnel is condensing. The idea of, I'm going to go into a model, look for the best coffee maker, compare a few different brands, and then make a choice and pick it and never leave that experience is really transformative. That's a big deal for a lot of retailers. It's a big deal for a lot of brands. And so there are places where we're expanding our projects to help them understand, help them understand today. What's happening top of funnel? Are you being recommended? What other brands are being recommended? How are you faring lower funnel? What's happening? How much does the price of your product matter? Where in terms of retailers are the AI models pushing to have the sale made? Those questions are becoming much more important as now commerce starts to shift to the AI models. So just a lot happening really fast in this space. That's really cool, but choose opportunity. That's right. But also for the platform that you've created, very cool. Yep. Thank you so much for joining us. Best of luck. We'll be watching and I bet we'll be hearing a lot more about your platform. Sounds great. Thanks so much for having me. I appreciate it.