Brands are now defined by AI rather than themselves, says Scrunch AI CEO

In this episode of Simulated Reality, Chris Andrew, CEO of Scrunch AI unpacks the disruptive impact of AI-powered search tools like ChatGPT and Perplexity on traditional SEO and digital brand presence. He explains how user behavior is shifting away from keyword-driven Google searches toward natural language queries in AI interfaces, essentially altering how brands are discovered and represented. As Chris puts it, “Brands are no longer what they say they are, they’re what AI says they are.”

He breaks down how his company helps enterprises adapt to this shift by monitoring how they appear in AI-generated answers, identifying gaps in content representation, and optimizing infrastructure to be better understood by AI crawlers. This highlights how AI search now values human-like, intent-driven content over gamified SEO tactics like backlinking. He also shares amazing anecdotes from Fortune 500 companies finding outdated content influencing AI responses, to AI models picking up on biased internal searches, emphasizing the need for brands to stay proactive and intentional in this new search paradigm.

Scrunch AI helps brands monitor and optimize their presence across AI search platforms like ChatGPT and Perplexity. It tracks how these AI  models interpret brand content, identifies misinformation, and reveals gaps between owned, third-party, and AI-generated data. The platform also audits websites for AI crawler accessibility and maps AI-driven customer interactions, helping businesses improve visibility and align content with AI behavior.

Kashyap Raibagi: Welcome everyone to the next episode of the AIM Media House Podcast. Today we have with us the founder and CEO of Scrunch AI, Chris Andrew. Hi Chris, how are you doing today?

Chris Andrew: I’m excellent.

Kashyap Raibagi: While user behavior did impact me personally as well, I’m using ChatGPT and Claude more than Google directly. It didn’t occur to me how that impacts big companies and how did that realization hit you, and what inspired you to tackle this problem in the first place?

Chris Andrew: My early realization was that brands are no longer what they say they are. They’re what ChatGPT, Claude, and other AI models say they are. That’s a huge shift in brand identity and the definition of a product or service. Historically, we’d ask a question and click on a few links to educate ourselves. I noticed a shift in the customer journey, where we started outsourcing browsing at scale to AI without fully processing it. When we get an answer from GPT, I’m not visiting as many websites. When I ask a question on Google, I’m still viewing fewer sites when I get an AI overview. The shift for a large brand is you might spend hundreds of millions on brand identity, ensuring descriptions are accurate and reviews are represented the way you want, but these models are now the first beret to how your company is represented. That shift is something organizations are starting to wrap their heads around.

Kashyap Raibagi: How does this impact SEO? Can you walk us through an example of a brand that was heavy on its SEO game and depended a lot on this? Let’s first discuss the problem statement in detail. That’s my thought process and then we can kind of go to your solution as well. Can you give an example of a company that relied heavily on its SEO strategy, it worked, but then user behavior changed, leaving them in a pickle?

Chris Andrew:  The shift here is that large language models want language. They seek detailed information in a human structure. If you think about how Google indexed and Googlebot historically worked and  rewarded folks, there’s been a lot of gamification with keywords, backlinks, and tactics to get to the top of search results. These models work in a more human, organic fashion, focusing on the human intent of a prompt and searching for the best way to answer the question. They no longer rely on just one website. This creates an opportunity for emergent players. While they may have lost the first ranking or page, the nature of these questions is now more long-tail and human in intent. SEO still forms the foundation for how models determine which websites to examine. They will look at an index, whether Google’s, Bing’s, or OpenAI’s new search index, which may signal a shift in the Microsoft relationship. 

Historically, GPT relied on Bing’s search index to determine which websites to view. They built their own index. The index is still largely influenced by traditional SEO tactics. Once crawlers access a page, they process and understand the information differently. From there, they decide which brands to represent in an answer and which web pages to cite as sources. This is the new equation organizations are trying to make sense of. The problem we see is that companies come to us saying, “We know this shift is happening, but we don’t know if we’re showing up, how we’re showing up, or why we’re showing up that way. What can we do about it?” Monitoring is often the entry point for organizations because a board member or C-suite executive will say, “I searched for a topic relevant to our brand, and we didn’t show up.” That sends fear through the marketing organization. “Let’s create a spreadsheet with a hundred prompts and monitor it manually.” We can discuss the failure points of this manual approach. But that’s often the problem, I’m not sure if I’m showing up and how.

Kashyap Raibagi: That’s interesting. The shift from Google search to ChatGPT is evident, but the problem statement you’ve defined, where customers are concerned about how they show up on ChatGPT and other SEO-related platforms, is equally important. The shift toward AI-powered search tools like ChatGPT is impacting Google search in itself. While the number of searches will go down, how is it impacting the SEO on Google searches itself?

Chris Andrew: I think you’re seeing Google moving aggressively into an AI search-first experience for consumers. The irony is Google invented the technology behind LLMs but didn’t capitalize on it, while OpenAI did. OpenAI beat others to the punch with a consumer experience that’s been delightful. Just last week, ChatGPT was adding a million new users an hour, having its moment of adoption in the mainstream. For Google, you’ve seen them start testing these AI overviews, which represent information much like how GPT or Perplexity do. If you’re following the Google Labs teams, you see new search experiences emerging that resemble a native GPT or Perplexity experience. Meanwhile, the traditional blue links are getting pushed further and further down the organic rankings. If you think about what that means for an organization like Google, it means they’re choosing to cannibalize the ultimate cash cow, probably the best business model on the internet, which was the mix of paid and organic search results that we relied on to get directed to the product we’re trying to buy. The shift to Google is enormous, but they’re sitting on the best set of data and will participate very actively. We have customers coming to us saying, “We’ve seen a 5% drop in Google search traffic almost every quarter for the last three or four quarters. We’re starting to see that backfilled with AI referral traffic. We need to start paying attention to this shift.” Think about which populations are spending more time in an AI search-first experience, and that will lead you to an understanding of how quickly the shift will happen for your verticals.

Kashyap Raibagi: That’s interesting. I want to enter the conversation about demographics and user behavior across age groups, countries, and cultures. But before that, let’s talk about Scrunch. What is Scrunch doing about it?

Chris Andrew: Scrunch works in three ways: monitoring, insights, and optimization infrastructure. On the monitoring side, we help companies understand how they’re showing up in AI search in a fine-tuned, granular way. We start by asking: What are the personas your brand sells to? Who are you trying to reach? Who’s trying to buy from you? Everyone gets different results in AI search. These models are non-deterministic—ask the same question twice, and you get different answers. If you and I ask the same thing, we’ll get different results based on ChatGPT’s memory and the shape of the prompt. Scrunch builds personas, identifies competitors, and maps key themes about your organization to show. These questions run through GPT, Perplexity, Gemini, AI Overviews, and Claude, showing how you’re represented compared to peers and competitors for the personas you’re targeting. That’s the monitoring platform, which leads to insights on gaps in your content strategy. Where are you not represented for themes and topics you should be? What are the opportunities to publish new content on your web presence and ensure it’s cited and crawled by models, increasing visibility and traffic? Another part is identifying third-party sites that AI search models value. The days of relying on one website for an answer are over, GPT or Perplexity looks at dozens, even hundreds, of sources to decide what content to use. There’s a big opportunity for brands to ensure they are accurately represented in third-party sources that influence the models. That’s the insights layer, focused on content generation and addressing gaps. The final layer is understanding the infrastructure behind what happens when an AI crawler lands on your website. How do you track, through logs, the activities of agents and crawlers visiting your site? How do you understand their experience when they land on a page? AI crawlers can’t interpret JavaScript. Google’s index crawler has become more advanced, but AI crawlers look for unstructured data. We’ve built a site audit platform that shows the experience of an AI crawler arriving at your site. That’s often the reason you’re not represented at scale, it can’t access your content. We work with organizations across this funnel, monitoring, insights, and optimization to ensure they show up at scale, attract referral traffic, and retain control of the customer journey.

Kashyap Raibagi: When SEO specialists analyzed what parts of their website had the biggest impact on SEO, they looked at backlinks and similar aspects. Are these elements the same when analyzing how AI chatbots evaluate websites, or are they different? I would also like to understand that. 

Chris Andrew: My slogan is that SEO remains the foundation of how these models determine which websites to look at. They rely on an index of pages to decide which sites to reference. But once the crawler is on your page, it’s behaving differently. I wouldn’t recommend abandoning your backlink strategy; it’s still crucial for traditional search and guiding AI crawlers to your page. However, once an AI crawler is on your page, it’s trying to match human intent. Very few people go to ChatGPT and type a brand name. They don’t type ‘Nike’ and hit enter. It’s not a directional search. I use Google for that. In ChatGPT, I ask a question like, ‘I’m a graduating university student looking for a great deal on a laptop for my first job. These are my requirements.’ It’s usually a longer form structure. Content that performs well tends to closely match human intent blog content, glossary content, FAQ content. Many organizations are now bringing more content online. One breakthrough when we were building the companies was helping organizations organize their internal data for AI outcomes. All external data owned, competitive, and third-party helps represent your brand’s products and services. How can you influence that? Make sure the right resources about your brand, products, and services are accessible for models to understand. If it’s cited, track referral traffic and conversions. Organizations bring content online at scale to ensure models have accurate sources for answers.

Kashyap Raibagi: That’s really interesting to me. What cool things have you observed? Can you give us an example of something unexpected or surprising you’ve discovered? An anecdote would be fun to hear

Chris Andrew:  Lots of fun things happening right now. I think it’s a better search experience for consumers, which excites me. Browsing, by its definition, is inefficient. I’m searching for an answer. On the consumer side, there’s a delightful experience for those who participate in this for the first time, even if it’s just seeing an AI overview. Surprising things for brands fall into a couple of categories. On the concerning side, Scrunch’s sources capability identifies which websites are used in AI answers. We’ve had many engagements with Fortune 500 companies that find a lot of their owned website content is outdated or misrepresented in AI search results. For example, old white papers with product descriptions from the early 2000s being picked up by crawlers and used to answer questions about a product or service. I’m talking about community or help forums on a company’s website with negative content that has been left unattended for a long time. Title, description, and metadata are critical because crawlers quickly identify relevant pages. If your title and description aren’t representative of your page, the crawler might skip your content and go to a competitive or third-party site. There are risks for the brand and opportunities. These models aim to be unbiased. They don’t just look at your website and accept that you have the best product, that’s what most organizations do. One thing working at scale is lists and comparative sources, like ‘Here’s how progressive insurance compares to others,’ or ‘Here are my features compared to my competitors.’ GPT loves that content. It’s like, ‘Okay, this is a more unbiased representation of the category.’ I’m going to link to this company’s comparative sources as a valuable resource for organizations. We also see a return to FAQ, glossary-style content, and knowledge-based content with lots of text. Why? Because large language models want language. These have been some of the wow moments. One thing to be cautious about is some organizations are trying to do this manually. They’ll run searches. Keep in mind, these models have memory. If you keep running searches from your laptop about your company, it will start learning that you want to see your company in the results, even to the point it will say, ‘Company name in parentheses: your company is great at these things compared to your competitors.’ The model knows you work for the company, so it steers the results toward what you want. That won’t be representative of what your personas see. So, it’s important to be fine-tuned and granular in how these searches are structured. The personas you’re trying to map to, these become moments in how we’ve architected the platform.

Kashyap Raibagi: That’s super cool and interesting. My follow-up question is: both ChatGPT and Perplexity have evolved significantly. Every tool will continue to evolve, and the pace of innovation is phenomenal. I was telling my cousin, an ML engineer, that we’re living in one of the most fascinating times in history.

What might be considered relevant for search today that wasn’t before ChatGPT? Do you think we’ll be in a similar situation in four years, or will some fundamentals of search optimization remain constant? How do you envision Scrunch’s journey in this evolving landscape?

Chris Andrew: The way I envision our journey is by sticking right with it. My co-founder and CTO, Robert McCoy, is one of the most technical individuals I’ve met. He deeply understands how these models work. We spent over a year studying how these searches are structured and why results are the way they are. One thing that’s been true over the last 18 months is that the models are frequently changing. There’s a new model or a new way search results are being represented. Many in the space are thinking rudimentary about how to represent the answer today. Keeping pace with this evolution is massive.

For example, in mid-March, OpenAI made several announcements, including pushing their own search index or search bot. This move away from Microsoft had been in the works for some time but became more prominent. These changes yield different results, so enterprises need a partner to keep pace with evolving results from sources like Perplexity, Claude, and Meta AI. Meta has turned every one of their properties into an AI search platform. For instance, when you search for a friend’s handle on Instagram, you’re triggering an AI search result. There are many new participants in the space, so finding a technical team that understands how these models work and is committed to keeping pace with the changes is crucial. I expect we will see change happen faster than we did with Google’s algorithm updates. My hope is that it’s a steady, positive evolution for consumers and less jarring. In the search space, some algorithm changes can completely reshape a business. AI search has done that. The goal is identifying the answer to a consumer’s question, not gamifying platforms to get desired results. We’ll see how this shift plays out as more monetization enters the equation. It’s been interesting to be at the leading edge over the last year and a half.

Kashyap Raibagi: Chris, this was super fun. This conversation was incredibly insightful. It’s easy to overlook how these developments impact broader systems, supply chains, search management, and various functions across large enterprises. Your journey is truly inspiring. Thank you for taking the time to speak with me.

Chris Andrew: You’re welcome. I really enjoyed it. Thanks for having me.