Your Content Is Corrosive: Anda Gansca on the Future of Websites in the AI Era

0:00What we constantly see is that marketers will confuse impressions with relevance. Let's say a piece of content is getting a ton of traffic. You may think to yourself, this is great, but what if they're actually having a really shitty experience where they're getting to the end of that content asset and they're thinking, boy, this was completely useless. And that's what we call corrosive. Is that pervasive still? Are we still operating? Yes. Okay. It's really pervasive. About 60 % of the content we measure is corrosive. When we show up to websites, about 80 % of us end up leaving immediately. So that's that immediate bounce rate where someone comes in and goes, whoa, this is not what I wanted to get.

0:40My guest today is Anda Ganska. She's the CEO and co-founder of Notch, where she helps Fortune 500 brands build content that captures and converts. Why does this matter? Because in the world of AI, your website, your digital presence is more important than ever. It's where AI learns about your brand and expands upon it. And Ander has the playbook for success. I met up with her in New York City. We recorded this episode in front of an audience of CMOs live at the Guggenheim Museum. We get into how to crack personalization without being creepy, how to beat the rising tide of AI slop, and why high-quality content remains both king and queen.

1:19Here's my conversation with Ander live at the Guggenheim.

1:27Hey, Ander, how are you doing? Hey, Josh, I'm good. I'm very excited to be here. We're recording here live at the Guggenheim. That's not something you do every day. I'm just going to translate all that positivity into we can do amazing things. It's great. I completely agree with that. I think there's this new era after having a lot of chaotic, negative press cycles that I feel like it's turning around. So hopefully we'll have a good summer. But now I'm going to ask you a depressing question. That's okay. I love depressing questions. I'm Eastern European. So I'm a marketer. Every single day, multiple times a day, I feel like a publisher, an agency, a consultancy, especially an ad tech firm will reach out to me and claim they have the latest and greatest AI silver bullet that will solve every known problem I could ever have.

2:14It's like an AI barrage happening. You're on the other side of the fence from me in terms of you're actually in an ad tech company or a consultancy. Can you advise me on how I should handle all these calls? And how do I know I'm not just getting snake oil? Great question. I think we're going through a hype cycle right now where because of the nebulousness of LLMs and the fact that fundamentally not even the builders of the LLMs know exactly how they work, it lends itself to a lot of speculation and fear mongering. So some combination of your brand is not getting mentioned in LLMs and here's a brand visibility rank for you does very well when it comes to catching someone's attention and getting them to want to purchase something.

3:00How do I deal with it? It's actually quite frustrating for me too, because I'm not one to make promises I can't keep. We've been building Notch for 10 years and we work with most of the Fortune 500 brands. So we're not here because we are in the business of selling snake oil. I like that. That makes sense. We're going to get into Notch in just a little bit. Not everyone listening to the pod may know of the company. Just give me your sense of the state of the marketing industry today. What's working? What's not working? What are you hearing? What are you seeing from your perspective? I'll start with what's not working.

3:32I think as an industry, we've gotten a little bit distracted with agents. I think we're getting too enamored with the idea that we need to market to agents, that now there's more agents that are coming to websites than humans, etc., etc. And we're starting to lose focus on the human. And the human is ultimately the decision maker. And I think this idea of zero clicker marketing is not right because ultimately every time there's been a big shift in technology, whether it's been social media or something else, we've always said, oh my God, this is it. We're not going to need a website anymore. The reality is that human beings are yes and creatures.

4:07We like to have choice. And so I think LLMs are another channel and I think we need to keep our focus on the human and creating a great experience for the human and agents should be a part of how we think about marketing. So it seems to me that agentic commerce is still a little away away. As you work across your customers, as you connect with them, is that true for them as well? Is it anyone who's cracking agentic commerce right now? Or are these agents, like you say, a little overblown in the near term? I actually think the direct commerce efforts are important. They're a big part of the future.

4:42But once again, I think it's a yes and. The journey of the consumer has just gotten more complex. There's now this LLM that is amazing at condensing a lot of information into a short process and giving you information in a way that you're inclined to want to consume it. But it doesn't mean that all the other things will die. What we know about LLMs and GEO for sure is that it makes all the other things important as long as you can be consistent, highly qualitative and distinctive across all channels. So you just said a phrase there, what we know about LLMs thus far. What do you know about LLMs thus far?

5:19I think the most important thing I know is what I don't know. That's a good cop out, but. Well, that's where some of that snake oil salesmanship is happening. It's where people are claiming to know things that can't actually be known. In terms of the things I know, what I know, put very plainly, is that if you do great marketing, you're going to get mentioned. And yes, do you need to have metadata and structures in there to be picked up by agents? Yes, you do. But the important thing is to have great marketing, to have great earned mentions, great reviews, great customer love across the board.

5:53And that's really going to increase your visibility. So let's just stack rank those things. I'm a CMO. I've come to you. I need help. What are the three things as a CMO trying to get ahead today do I need to really be thinking about to actually deliver on what you're saying there? The first thing I would do is I would run what we call a knowledge gap analysis, because what you want to understand is what are people searching for or likely searching for out there versus what is the knowledge you actually have. This is a really important thing, because if you don't have your source of truth as a brand, then you're basically relying on LLMs to make it up.

6:25The second thing I would do is I would create this agentic knowledge base and really think of the website as one of those inputs. But the reality is knowledge is currently siloed in many aspects of the organization and you need to bring it all together and make it both agent readable and conversational for humans. The third thing I would do, which I don't know how many marketers here are doing, is really think about how I could get that peer department revved up. The more earned mentions you get, the more credibility you automatically have, especially in places that are of high trust. PR, comms, content.

6:56Let's just touch on Notch's past 10 years. You've really been focused on understanding engagement and content more than many other metrics that advertisers are focused on. Can you just give us a little potted history of what Notch has been doing and why it's interesting for Mudd brands? We've been capturing data about how audiences journey through brand properties for the past six years or so. We capture data from the biggest websites in the world. And as such, we have this first seat in terms of the intelligence that we see around how the audience journey is actually changing versus not changing.

7:32And so what we've seen is audiences may start their journey in LLMs, but then they will go to social or they'll go to search or they'll directly go to a website. Then they'll come to a website. They're going to go through their validation process. They're going to try to connect with the brand and then ultimately they will transact. What we've done for brands is really enable them to understand how to better create content that will resonate with their audience. What is the content that doesn't actually resonate with their audience that's actually corrosive and is actually hurting their business results?

8:01and how should they organize their digital experience around all that content in order to essentially make it a slam dunk from a customer perspective. Corrosive is a good word. Can you just unpack this a little bit further, though? Give us an example of something where you found something out that you thought was going to be high performing for the brand and was not. And what underpinned that, why it wasn't and then what you changed it to? I would say across almost every one of our customers, whether it's Zillow or Cox Automotive or some of the biggest banks in the US, What we constantly see is that marketers will confuse impressions with relevance.

8:38And I know that everyone talks about how impressions are a vanity metric, yet they are the number one metric that everyone still looks at. And so when you see that a, let's say, piece of content is getting a ton of traffic, you may think to yourself, this is good. Maybe you see that it's getting engagement. You may think to yourself, this is great. But what if people are actually consuming this content asset and not getting what they actually came there for? What if they're actually having a really shitty experience where they're getting to the end of that content asset and they're thinking, boy, this was completely useless.

9:07This was SEO games that basically caught me and got me to read all the way through. So across all of our customers, one of the first things we do is we create this quad chart where we have one axis is impressions and the other axis is either conversions or brand affinity. And so they can easily tell which are the assets and which are the experiences that are driving a ton of traffic, but they're getting people to leave and never come back. And that's what we call corrosive. I remember once building a website for Hargendas. It was about Mayan chocolate. So we had 45 pages about Mayan chocolate.

9:38And then we discovered it cost a fortune to build this thing. We discovered like sort of, you know, six months later, the depth page was two. So it was just, but is that pervasive still? Are we still operating? Okay. It's really pervasive. About 60 % of the content we measure is corrosive. And to keep hitting you with some pessimistic stats, we've now seen that what we call this AI native audience, which is like all of us because we've now been conditioned by LLMs. When we show up to websites, about 80 % of us end up leaving immediately. So that's that immediate bounce rate where someone comes in and goes, whoa, this is not what I wanted to get.

10:13So how do you counter that, given that most of our brands have hundreds of thousands, if not millions of customers? one website is not going to work for everyone, right? So dynamic, fluid websites. Yeah, I would give you two answers. One that's probably more self-serving and one that's a bit more general. I think the most important insight for brand marketers is they should stop trying to make their websites about what they think they should be. And they should really start asking themselves, what does the audience actually want to see? I cannot tell you how many times the product marketing teams within large organizations will say, we need this product.

10:47It has to have this name because we brainstormed it so hard. And then under that, there has to be the sub menu. And it's just not the reality of how people search for things nowadays. The example that I like to use that my team is sick of is I've been looking for an electric car that doesn't look like a mom van. And that's the way I express myself. And I'm not going to express myself in keywords. And I don't know anything about car brands. And so manufacturer's website should be intelligent enough to allow me to say to it, I'm looking for an electric car that doesn't look like a mom van and respond to me in a way that feels personalized and relevant.

11:19Earlier today, we announced that we launched ACE. It's essentially a personalization engine that allows every single person to get an experience of one based on their semantic input and a bunch of other data that we capture. It's always fun talking to you because you have all the answers. But I know that along the way, you've made some mistakes as well. I read about your first sort of AI launch and it kind of crashed and burned or didn't work the way you wanted to do. Just tell us a little bit about your learning journey yourself in your organization as a self-professed data nerd getting it wrong?

11:48I would say that the mistakes that I made and that we made as a company were when we were trying to really be something we're not. We were listening more to the buzz and the tech and the hype and tried to feed into it as opposed to really stay true and core to what we do best, which is build performance-driven data-first engines. There's been a couple of different products that we built because there are VCs who are like, guys, this is it. It's the crypto era. Not that we built a crypto product, but there was that era too a few years ago. And it's tough because you listen to something that you think is a higher order voice that knows better than you.

12:26And then you realize that the only thing that really matters is listening to your customers and to your intuition. You're using AI day-to-day. You're vibe coding yourself. One of the perennial questions that's come up on Frontier CMO is that how much of a CMO needs to learn new skills like vibe coding or just coding versus the fundamentals. Where do you see the balance right now? What does a CMO need to be developing that is new? What needs to continue? The answer to that question is curiosity. I think you have to stay curious and you have to try new things. And that's why I vibe code as well.

12:58And that's why I try to build on my little apps, because I think there's something about pushing AI to its limits to see where it can really go. And it also makes you significantly more aware and less likely to fall for snake oil salesmen as well. So you're talking about walking the walk, not just talking the talk. How have you made that happen inside a notch? How have you got your organization to actually embrace using the AI in their work as well? In our team, we had a group that just naturally gravitated towards AI and using every possible thing under the sun, myself included. And then there was a group that was probably more resistant to it.

13:31And so we decided to declare an AI mandate and really make it something that everyone has to use, that a certain percentage of our code has to be written with it, that the way that our sellers and our account managers prep for meetings with clients has to be done with an AI model. And I was really nervous. I honestly didn't know if we were going to be faced with a little bit of a revolution or not. every single person on our team adopted it. And it almost needed that extra level of push to try to start to work with it. And our speed got faster. And so did our number of bugs. And so then we started building ways to navigate that and so on.

14:07So let's get a little listicle here. I'll share a couple, you can share a couple, but what are the one, two or three things that have either helped you or help the team actually embrace this AI moment and take advantage of it? So one thing that I personally did is I organized these live virtual sessions where we were just essentially vibe coding together and all working on different ideas and kind of sharing what we were working on together. One of the things I think is interesting from a marketing perspective, certainly the more senior level is you need to start building this stuff into the system.

14:38People get promoted earlier or get bigger bonuses or compensation if they are shown to be leading in AI. And that becomes one of the gating criterias actually helps set the signal both at the top, but also then do those smaller things in terms of day-to-day coding and lessons and learning. I have a lot of our Google customers, a lot of CMOs want to come to Google and get lessons of people actually using the tools, not really realizing you can use all these tools yourself directly right there and then, like Nano Banana, VO Flow, and a bunch of all these other ones. It's very apparent to me, and I'm interested in your take on whether you can see which CMOs, which teams are not really embracing the technology, but claiming to at this point?

15:15I mean, for sure. No, no, but I can tell you that I think 25 % of CMOs are probably pushing the boundaries of AI and they're seeing around corners. You know, they're the folks that adopted a lot of this geo stuff early and are the first now to call it a bit of snake oil. And 75 % of them, I think, are still following suit and probably just now starting to learn some of those lessons. So let's put a positive spin on that. They're held back by compliance, by finance, by bureaucracy of some sort or another versus they're just not engaged. Any advice or how to tackle those other cross-functional disciplines to get them to unlock and enable marketers to actually embrace AI?

16:01Have you seen anything that's really worked doing that? I think it's about creating very small pockets of experimentation and not waiting for cross collaborative consensus for everything. That's one. It's really about allowing people to make mistakes. So create low risk environments, like whatever you're doing with AI, create something that is not going to blow up your brand, but allow people to fail because AI makes a lot of mistakes. And so we need to be able to allow it to make mistakes, to learn from it, to teach it, to move it forward. It goes back to curiosity and fear because ultimately it's all about these basic kind of human emotions.

16:37If you're afraid of AI or you're afraid of being left behind, you're going to be held back from being curious. And I think the sooner you can switch that mindset to curiosity and just going to these tools yourself and start playing with them and try to break them, turn it into something fun. Like I think the better you're going to realize how to lead your team. But from a professional perspective, can you give the audience here a couple of examples where companies you've worked with have actually embraced using AI to solve even a small problem or a new challenge, which has actually been successful?

17:07What was the problem they were solving for and how did AI help? There's a lot of really great examples in the realm of generative AI for how generative AI can be used to make average content better or make corrosive content really good. And we've done that across a number of our customers. Realtor is one that we have a public case study with and we drove some really incredible results lift and everything really from engagement conversion and so on with ace which has really been live for about three months with a few of our beta customers we've seen define ace again because i'm not sure agentic content engine ace is the product that we launched today here at the guggenheim we've seen 20 decrease in bounce rates 5x increase in engagement and 3x increase in conversions the punchline is when you're using AI and you're doing it in a governed way, it really works.

17:55It can make, and that can actually deliver on the promise of personalization, which is something I'm really excited about. And do so in privacy compliant and safe ways. Totally. Of course. I think what you need to do to make personalization really click is to not use a lot of generative AI. And we've learned this, my co-founder and I were up last night making our own website live on ACE. And we learned a lot about how much to restrict generative AI versus how much to let it go. But if you can create the right rules and regulations around it, can really drive a good experience. Were you making the website the last night because you forgot to do it or because this is the way you operate in our last just-in-time production?

18:31We were making the website last night because we're so customer first that we were obsessed with getting our customers' live experiences live. And we allowed ourselves about 48 hours to get our own live, but we did it. It's great. It's live. I asked a little earlier about what doesn't change when it comes to marketing fundamentals. And organizing your data seems pretty important. And I just want to take a step back and really ask the question, it still seems like many brands, many marketers don't have their data house in order. Yeah, I think because there was another snake oil era of a different kind that we called the CDP era.

19:07We all believe that if we move it from one place to another, it's going to be automatically structured. But putting all your data in order is about getting it to be structured, not just the kind of data repository you have today, but also to have really structured data collection that's clicking with the data you already have. It's a hard challenge. Let's work through the timeline here. I come to you, any technological partner, what are the two or three things I need to really get in order? And what is the timeline to do that? And I know that's how long is a piece of string, but I want to try and give people in the audience who are concerned that they can never catch up.

19:38With AI, you can move faster, right? Yes, I think what's important is I would go on a hunting mission to find where's my data, where's my information, where's my knowledge, where are the silos, bring it together and work with a partner who actually knows how to bring it together and make it match. That matching part is really difficult to do. I think you have to augment your data with more first party data if you can. And then I think you have a good solid foundation from a data perspective. You also have to do the exact same thing with your content in order to be able to get your stat and content from today to be modular enough to then be powered by data in order to drive personalization with AI.

20:14Where's this all going in your mind? It feels to me like we're sort of in an in-between moment, right? We've gone past the first era of AI transformation where a lot of it was defined by fear and uncertainty. And now I'd like to think we're entering this area where our teams are starting to think about what they can do with AI. So positive, uplifting opportunity. Where do we go from there? How do we encourage people to take that mindset more and where do you actually see the technology going itself? Some of the marketers here before said we are now in pilot purgatory. I think we have to get through this phase and we have to decide who are the two to three partners that we can scale with as a brand.

20:52After that, I think it's going to be a really fun time because we will have already realized what are the things that are not worth focusing on? What are the things that we can really automate and put away so we can focus on the stuff that really matters, which is to do great marketing. I think in today's world of so much AI slop and noise, the thing that will differentiate a brand, the thing that will differentiate a human is taste, thoughtfulness, judgment. And so having the actual time to invest in that and create that distinct brand, I think is going to be a lot of fun. It'll be like the golden era of marketing 2.0.

21:27I like the sound of that. Let me just push back respectfully on one thing there. It seems to me that marketers have for decades and decades made a significant proportion of their output as unmemorable, crappy, bad, dare I say it, slop ads. Decades and decades. I kind of think that the floor is actually higher. And what I'm seeing is using Google's generative tools and other generative tools grounded in your own best practices and learning, you actually can raise the floor as it were and start off from a better place. Is that a fair thing to say. I just feel like AI swap is over plan. I don't think so.

22:03I think what you're doing is you're just creating a lot more average content, which is really just adding to the noise. And we've had enough noise. I mean, as consumers, as humans, that's why we love LLMs because they condense all that noise into an answer. I think it's about time that we take note of that and really try to make great marketing that's going to speak to humans. And then the slop, I think, is highly relevant to the new audience that loves slop, agents. There is genuine feedback in the industry from large platforms that you need to be reducing hundreds, if not thousands of assets and just let the algorithm decide.

22:37That seems counterfactual to AI slop and to what you're saying. So what's your take on that thesis? I think there's a time and place for content that you need to feed the machine. And that's what I mean when I say agents. I think we need content that feeds the machines. And that content is at scale and that content has to have many different variations so that the machines know what to do with it. What I'm talking about is the human experience. I believe the human experience needs to be really personalized, really special, really immersive. It has to be different from all the stuff that we're putting out there for machines in order for our brand to stand out.