How to Create Content for AI, Gymshark’s Sorority Play, and Rethinking Burnout | Ep. 3037

Let's go with creating the content for generative engines, like how to go and create content generative engines, the biggest nuance.

Is that yours? Yeah, that's yours.

Uh-huh. So right now people are like still focused on keywords. I already know, you know, you and I agree that it's all topic-based. I want to give an example of how people should think about content in today's world. You probably have lots of examples as well, but let's go through a journey of like buying a car. You have a nice, beautiful car. At least your color of your car is amazing. I don't know what kind of car it is, but I think it's a Porsche. Electric Porsche?

Porsche Taycan.

Okay. So when you go about buying a car, you're going to look at specs, right? You're going to look at potentially resale value, how much it costs new, used, how much it loses value, safety ratings. You may ask some friends about it. Did you ask friends about your car?

No.

Okay. Did you do or watch YouTube videos about the car?

Yeah.

Yeah. So you're looking at a lot of different elements. You looked at safety ratings, mileage, range, stuff like that.

I look at range. I don't look at safety rating. I probably should have. But yeah, I've researched a handful of sites.

Yeah. Yeah. And then you probably compared other cars in similar price points. Yeah, I did.

I did. I looked at multiple.

Yeah.

Yeah.

Right. So you're not just typing in, you know, electric sports car and just going and clicking on a website and reading an You're looking at a lot of different factors, like, you know, to recap, uh, specifications, what other people are saying, whether it's a friend or YouTube videos.

Yep.

You're looking at, uh, competitors. What's the pricing of competitors? Are their range faster, lower, seats more comfortable? You're going really into it. Color options, right? Is it 4-door, 2-door? All these types of things. So when you look at those kinds of things, that's your whole cycle of trying to figure out if you should buy that car.

Mm-hmm.

The problem though, when people create content in today's world for blogging, they just look at like, is a Porsche Taycan the right car for you? And they just try to shove everything into one article, breaking down specs, reviews, and saying like, here you go. People don't buy cars that way. And in the generative engine world, people want you to cover the topic so thoroughly, and that's really hard to do in one article reviewing a car. You would need Images, videos, and the list goes on and on. And together that allows you to figure out, should you buy that car? And the problem that I think that people are still having when it comes to creating content is they focus on one simple thing. That one simple thing is, well, what are some keywords? And let me go super thorough on those keywords versus covering everything around that topic.

You know, I'll go on a little tangent, then I'll come back to this because I have a graph that's related to this. But what's bothering me right now is when I talk to my team, when we talk about YouTube is a major factor with LLMs, right? Or Reddit is. They look at YouTube with an SEO lens. Oh yeah, with YouTube you have to do keyword research. With YouTube you have to get the tagging right. I'm like, dude, none of that matters as much as the packaging that you have, the idea that you have, the 30-second retention, your thumbnails, your titles, all those things, right? So every game, whether it's Reddit or YouTube, whatever these LLMs are sourcing from, they have a native way to play. And I think the problem is, if you're an SEO, SEOs like to look at things, oh yeah, you're going to do keyword research, you're going to do these things. It's like, dude, you can't look at this the same way as you used to. Even if you're trying to play traditional, even that's evolving quite a bit now. And I think all these SEO companies, even the stuff you have going on with Answer the Public and Ubersuggest, it's shifting so quickly that you You have to adjust your product. And so when you look at this, coming back to what you're talking about, so you've seen this graph already, but where AI gets its facts, I have this on camera right now, but Reddit is 40%, Wikipedia is 26%, YouTube is 23.5%, Google is 23.5%. So yes, this is why you and I are getting a lot more inquiries for Reddit and YouTube because it's top 3 here, but Google is still a major player. So you should, yes, still do traditional SEO, but YouTube and Reddit and even Wikipedia up here, you can't just try to play it the standard SEO way. Oh, we're going to Because when you search for a query, when you search for something on ChatGPT, it usually pulls up 30 to 60 different sources. It's looking at all these things at once. If you go deep research, it's pulling a lot more sources than that.

Dude, I agree. And this is why it's more important than ever to participate in all these community sites. I think businesses have the issue and they're like, oh, we'll control our own narrative. We'll just go and create content on our own website. Oh, sure. We'll pay someone to help us with the Wikipedia page, but they're forgetting about what users are creating on Reddit, on YouTube, and all the other sites that's really easy for people to publish content. Wikipedia is a little bit harder, but those other sites are really easy and they're starting to pull from everywhere and it can screw you over.

Yeah, dude, here's the thing though. So like when people are coming to us for Reddit, when they're coming for YouTube, they're all, they're looking for all different things, which shows to me that people don't know what to look for yet. Right? So, okay, if I'm coming to you for YouTube services, that means I can't just go to an agency and say, oh, you're going to make videos for us. It doesn't work like that.

Yeah.

Right? We actually need you as the client. Let's say you're the client. I'm the person coming to you for YouTube services or like I'm going to do YouTube services for you. Maybe I can help with strategy. Maybe I can come up with ideation. Maybe I can help with the thumbnails. Maybe I can help with the, maybe the editing as well, right? There's not much more I can do. You still need to publish at the end of the day. You want me to do it for you? I can't do it for you. Okay. Number 2, you take a look at Reddit. Oh yeah, we'd like you to just, you know, make thousands of accounts and post for us. So you want us to spam Reddit? Like, no, you can't do that either. You need to have really good community managers. And the best we can do to make sure that we don't get you into hot water is we probably need to make a page for you and just hire multiple community managers to manage that for you. Okay. Okay. Some reputation management, all these different things. People just don't know what they're looking for. And 2, guys, don't try to game these platforms. Try to play the native game. That's what I'm getting at.

Dude, I'm with you. And when I look at what's happening with corporations, when they're looking at these sites and these profiles, they're looking for shortcuts and they don't want to spend the time really investing in the community. And it doesn't work without that. I just haven't seen anyone do it successfully. But even if Reddit's— these AI platforms are pulling data from Reddit and YouTube, it's just like people can create their own narratives. But interesting enough, my team and I were working on a chart that I'll publish in, I think, today or tomorrow or the next day. And this chart didn't have really tons of proprietary data. It was more so on Google's growth over time. Okay.

And the rise of zero clicks, which a lot of people have published data from.

So we're gathering from different sources. So I was using ChatGPT to do research and I was getting it to really think and it was going through, it was taking a while to come up with it. I don't know how many minutes, but a while.

Mm-hmm.

And it was crawling the web, looking at images, deciphering it, because you can see when it's analyzing, it actually tells you what it's doing, like verifying out of the images what percentage is actually no click. And it starts giving you a table of the results. Some of the stuff that it pulls from these external quote unquote sites or social networks is just really off.

Mm-hmm.

Like I know, like it'll say like analyzing SparkToro's graph or image or whatever, and then And I'm like, I can see it myself. And you pulled the wrong data from that graph.

Yeah, it still hallucinates quite a bit.

It hallucinates so much. I found that when I'm creating blog content, which is a similar type of topic, but I love using AI for research, but it hallucinates a lot when it's super technical. If it's not technical and it's basic, it doesn't hallucinate too much. But when I get super technical, it's just like, In many cases, it creates more work than actually helps. Yeah, for me.

I've noticed that too. So, but that goes back to user. I mean, I want to call this user error, but it's user knowledge. You have to, you have to understand how to use these tools and understand when it's wrong and when it's right. And if you're, if you're not playing around with it enough, then you're not going to know. And then what's going to happen is you're going to get in trouble for publishing misinformation.

But it's not just that. Yes, it's knowing how to use these platforms to its best ability. But you and I know this because we've been in the marketing industry for like, call it 20-ish years, right? I'm rounding here. But if you're new to marketing and you want to write like an expert and you do the research and you start pulling it, a lot of it you just won't know.

Yeah.

We know without actually even referencing the charts and the articles because we've just been doing this for so long. We know a lot of the original research pieces out there. We know when something's off, but a lot of people wouldn't. If you and I were doing this in the medical space, And creating blog content, I bet it would hallucinate and we wouldn't catch it all the time.

If you're in B2B, here's the deal. LinkedIn account-based marketing sucks, but it doesn't have to. With Carrot, it allows you to create personalized ads towards your customers and also personalized landing pages that align with the ads. So you're going to have higher conversions and influence pipeline reporting to prove that your ABM works. So if you want to learn more, just go to carrot.ai and you can join the waitlist. Again, it's karrot.ai, and we'll see you on the other side. So this is what Aaron Levy from Box, the Box CEO, says, and he's been tweeting about AI agents a lot. If you're going to create subagents or AI agents, subagents are more focused on one type of role. You have to understand how to context engineer, and you have to put in the right data. You have to— only someone that knows the craft is going to know the right context. And so that's why it was really easy, or not easy, but simple to build like Cursor or Windsurf, because it's the developers that know the problems to solve. But someone like you and I are very deep, let's say, with marketing with SEO. And so when you're rebuilding, retooling Ubersuggest right now, and AnswerThePublic, and I guess even Crazy Egg now.

Yeah.

Yeah. So you guys understand what to do. When I say you guys, Hiten, your partner has gone back to Crazy Egg to focus on that. So I'm like, okay, well, that means that if you have deep domain expertise, you're gonna have a big edge with building these agents and sub-agents in the next couple years, because You're going to know the right questions to ask and you're going to know how to prompt it and you're going to know how to feed it the right information. You're going to know how to call it out too.

So yeah. I was talking with Ethan a few weeks ago because I was in San Francisco and he was talking to me about Crazy Egg and the developers and what they're working on right now with AI and hallucinations. And he was just like, you know, when you're running A/B testing, you want to modify your website. Why can't the AI just figure out your brand guidelines and help you rejigger the page both on copy and with design elements? And then when you don't like something, you notify it, it creates variations, but also learns from why you don't like it. So then it gets better and better over time. And he was just like, you know, because he was showing me it and I was like, oh yeah, it works actually pretty good.

That's where it's going.

Yeah, that's where it's going. And I'm like, you think about the whole marketing ecosystem, AI is pretty good at seeing what's there, learning what you like, and adapting over time as long as you keep feeding it data. May not be perfect on the first run, but it can get better and better over time. I just think the issue that I see in corporate America, especially in marketing, they want it perfect the first time.

Yeah.

And they're not willing to invest to make it better over time. And that's a big mistake.

You have to keep feeding it. It's just like, imagine it's when you feed these agents, it's like onboarding a new employee. You have to give it as much context as possible. The other problem is you have limited context windows right now, up to 250K tokens to a million tokens. It's going to get higher over time, but that is a problem and you can't try to feed it too many things at once, otherwise you're going to confuse it.

But what are you saying with 250,000 or 1,000,000 tokens?

So there's a context window, like with each chat, right?

You can't—

you can only chat with it so much, otherwise it becomes overloaded.

Yeah. So I thought you meant you're only allowed 250,000 tokens in total.

Oh, no, no.

Like in their API.

Just like one shot.

I'm like, no, no, no, they won't make as much money.

Yeah, yeah, yeah, yeah, yeah, yeah.

Give them.

Yeah. But the Yeah. Anyway, so that's that. But by the way, if I switch directions on this one, this is interesting to me. I think you'll like this. So check this out. So, you know, the company Gymshark?

No, I haven't heard of Gymshark.

Gymshark's like a very successful, like, athleisure or athletic type of wear company. They're big in the UK. The founders, I think he's only 30 years or so. I think he's a multibillionaire or maybe billionaire. But check this out. So Gymshark, They're sponsoring a Big Ten sorority. So a Big Ten sorority is one of the divisions in like football, like Ohio State, Oregon, USC, for example, right? So you can sponsor a top-tier Big Ten sorority for under $3,000. If the video gets a million views, this would be a cheaper CPM than a Pornhub ad. Not saying you should advertise on Pornhub necessarily, but it's just, you can see in this video, they're dancing around as Gymshark, right?

Yeah.

Some of these videos can probably get a million, maybe tens of millions of views or so. That's a good deal. So for me, this is underpriced And I'm like, why would you— this will work for consumer companies.

Yes, but I'm going to tell you right now, because Eric's showing me a video of these girls dancing, uh, people are going to get called out for being sexist, womanizing if they go this route.

Woke crowd.

Uh, whether they're woke or not, I think people are going to get called out for it. Um, and I think it works more for Gymshark because of their appeal.

Yeah, yeah.

So it works for them.

But let's say if you or I did this as a B2B company, it doesn't work.

We get eaten alive.

Yeah, it doesn't work. But consumers, this stuff can work, right? Or you— I mean, if it's like food apparel, for sure it works. But if it's like food type of stuff too, maybe there's someone that's exploding. There's a format that's working because this format right now is working. Basically, there's a lot of celebrities posting it and everyone on Twitter is like, America's back, America's back. Like, we like this, we like this, right? So this is also the moment—

The Sydney Sweeney stuff, right? We talked about that a few weeks ago.

We're back, we're back.

And yeah, I think American Apparel stock's still doing Decently well and their sales are up.

American Eagle.

American Eagle.

Yeah, yeah, yeah.

What'd I say? American Apparel?

You said American Apparel, which I think they're gone.

It could be. I have no idea. I meant American Eagle, but thank you for the catch.

Neil, have you ever been burnt out before?

Yes, I have.

Okay. Tell me the story. When have you been burnt out? When's the last time?

Typically the— I've been burnt out many times, but the last time I really remember it is when I tried to create Vision Web Hosting and we changed the name to Surf. It was a cloud computing hosting company back in the day before like AWS. I'm not saying we were going to beat them or anything like that. Picked a bad team for it, poor execution, spent millions of dollars of borrowed money on it and lost it all and had to pay it back, but really felt burnt out from that.

Would you say true or false here? Burnout is caused by a lack of believable motivation, not working too hard. So, Basically you not seeing that there's a way out of this.

Yes.

Okay.

Because of working hard, I've worked very hard for many years and not felt burnt out. It was more so when things don't work out, that's when I feel burnt out.

So how many times in your life have you been burnt out?

I don't know. It's hard for me to count, but that one was definitely one. Kissmetrics probably was number 2. The losses really screwed us up at the end. I'm only counting the failed businesses.

Yeah.

But that's probably because I've always had this notion of working hard. I don't really look at it as working hard. I just love work. It's more so when I can't figure out how to make things work, that's when I feel burnt out. When I feel like there's a solution, like even NPD, my digital ad agency, the overall ecosystem right now for marketing, Most agencies are struggling right now. You agree with this? But I still see light at the end of the tunnel and I still see a way out and a way to grow even in a bad economy.

But you guys, even in a bad economy right now, and we'll come back to this topic in a second, but I feel like you guys are moving, things are picking up because things are picking up for us. Yeah. Yeah. So I see the momentum, but anyway, I want to come back to this real quick.

I'm cautiously optimistic.

Same here. So, you know, the founder of Boom, the airplanes, the new supersonic planes, this guy's name is Blake Scholl.

Okay.

So he wrote this, Everything You've Been Told About Burnout Is Wrong. So he says burnout is caused by a lack of believable motion— motivation. Sorry. Burnout is caused by a lack of believable motivation. Not working too hard. Breaks won't help. Engineering motivation will. And so what he was saying was, look, in 2020, the entire Boom team, again, they do these supersonic planes. Have worked extraordinarily hard to get the XB-1, that's her model, ready for its October 20th public rollout. Many work 6 to 7 days a week, 12 hours a day. People skip vacations, miss kids' birthdays, and went all in to get XB-1 into beautiful shape for her debut. The rollout was a magnificent success. The world was introduced to a soon-to-be historic aircraft and the role it would play on the journey to supersonic renaissance. We were in the thick of COVID so the main event was streamed, but we still had in-person celebration for employees and their families. The families had been making sacrifices on behalf of the mission, and we wanted them to enjoy it. what their loved ones have been working to birth. Now, long story short, what happened afterwards was like he noticed that people were getting burnt out, so they— he gave them a week off. Okay, so, you know, you're working 6, 7 days a week, 12 hours a day. So they— he gave the week off. They came back complaining of burnout and overwork. And his first reaction as a CEO was like, dude, what the hell? Like, I just gave you guys an entire bonus week off, which is actually not a lot of time if you've been working 6, 7 days a week. But point here is He realized that the goal was so unattainable, it was so far out. And him as an entrepreneur, he can take a lot of pain. He can wait, wait, and wait, right? But for the Vision Hosting, like, if it seemed like the goal was so far out where like you're just getting punched in the face, you continue to sink money into the company. Or for example, situations where I had to face with Single Grain back in the day, where it seems like you're staring into the abyss, that is when you start to lose motivation. But other than that, I don't really feel burnout. Other than that, you don't really feel burnout.

No.

The key here is that he's like, oh, he realized that You need to give your people like milestones, break up the things into chunks so you can celebrate these things. Just like Elon is with the different Falcon versions, you can't put it all in one because otherwise it feels like it's too far out and your people will get burnt out a lot faster than you will.

Yes, I agree with that. And I, every time I felt burnt out, it has to be, it's related to feeling like something is failing. And if you can start breaking up things into small wins and you celebrate them, It really does motivate and bring the team together, creates a stronger culture, and people push harder forward.

Also, you and I are very bad at celebrating things, so you need to find someone on your team who's really good at it because I don't care to celebrate. Like, you'd even celebrate your birthday. I don't really celebrate my birthday either. I just keep going.

That's right. I didn't celebrate my birthday. I was speaking at—

Yeah, you're at the YPO event.

Yeah, but I did have a free ice cream.

You had 2 free ice creams.

Yeah.

You didn't even tell me it was your birthday that day. I didn't even know.

I don't want anyone to know. But yeah, we're really good at celebrating failure. When things fail, we make it known and we try to learn from them. And then when things go well, we have the personality of like, okay, let's do it again.

Yeah. That's why when you compliment me or I compliment you, it doesn't do anything. I like, but if you tell me where I failed, I'm like, I get a lot more value from that. Yeah. So, because by the way, so Neil and I, we, we, this is a little side tangent, but when we both took the personality test, you're, we're both 99% receptive to criticism. That's not like that for a lot of people that I go through that. No, a lot of people aren't like that.

You know, I haven't really looked at anyone else's personality test.

Oh no, I've looked at a ton. I have a ton in here.

I only took the personality test because Eric made you. Yeah, Eric's like, take this test. I was like, okay.

Yeah. And that is it for today. We'll see you tomorrow.