Meta and Google Are Not the Best Place to Start Anymore

3 underpriced marketing channels for Q2 2026.

By the way, all these were generated by OpenClaw, so you could go first.

All right, so channel number 1, we're seeing the social networks that no one likes talking about— Snap, uh, Pinterest, X. I'm not talking about talking about from a political standpoint, but from a marketing standpoint. These channels, they're performing well. The ROI is still massive and for some reason companies don't try to run on them. And we believe that if you're going to run paid advertising, don't start first with the Metas and the Googles. Start with, especially if you're a smaller organization, start with these smaller social networks because the cost is just way cheaper. They can't scale as much, but that's okay. At least you're driving in more profit for whatever spend you have.

Number 2, so I think a very underpriced marketing channel right now is GitHub. Let me tell you this. So I am basically open sourcing all of my AI skills and I'm just tossing it onto GitHub. I just threw it up yesterday. I have 250 stars on GitHub. So do I have the same amount of OpenCLaw stars? No, I don't. But 250 stars is a good amount. And what I'm going to do is I'm going to keep adding to that repository, that repo, and I'm going to keep upgrading it because I was using Gary Tan's from Y Combinator. I used this for the first time yesterday. It's like, hey, are you okay sharing your data with us on how you're using it? I was like, oh my God, that's smart. And then afterwards, like, hey, are you using the most updated? I was like, I took that scaffolding, I added it to mine this morning, and I'm just going to keep adding to it. And then I can see their usage, I can see what they're using, what they want more of and whatever. And in the future, I can see marketers adding like, oh, by the way, what's your email in there too? I'm sure we can ask all those questions. But I think GitHub is going to become more and more of a marketing channel because agents are using it more and more.

Number 3, GEO. So when we look at traffic from GEO, it converts roughly 2.5 times better than email and SMS traffic. The number 2 highest converting channel, I believe, is SMS. The number 3 is email, and then I believe it gets into some paid advertising channels. GEO converts roughly 3x better than these paid ad channels. So this is massive. The cool thing about GEO versus SEO, you can actually get results in a very short period of time, call it like a few months. One of the big factors that they're looking for now when they pull results is how updated is the content. So your website content is super important to be updated. They don't like pulling old outdated sites. The other thing is when you're getting brand mentions around the internet, you need to optimized not just for velocity, but recency. They're looking for how many times is a company being mentioned as of recent. If you're Coca-Cola and you've been mentioned a lot over 10, 15 years, those articles from 10, 15 years, the AI doesn't like pulling that as much as an article that is 2, 3 months old.

All right. So let's start. Sorry, I was thinking about these earlier. You want more? No, no, no. Okay, I'm going to continue. So thank you. Next topic here.

Wait, they are really good.

No, I just kept calling you something. It's an inside joke in my own head. So how AI turns B players into A players. So this is the multiplier effect. So we can kind of go, go back and forth on this because we're seeing it happen in our organizations right now. In my mind, the biggest multiplier effect is when you're able to talk to it and you're able to run multiple things in parallel. So multiple tabs doing different things. So one, you can have a tab doing conversion rate optimization for you. The other tab, you can have it doing SEO for you. The other tab, you can have it helping doing design work. You can have it ship different features for you without waiting. And so the way it turns B players into A players is the B players used to need to wait. They still need to be told what to do. That's the definition. That's a B player. You need to be told what to do, and A player kind of just does it. But the B player, if you tell them what to do and say, hey, you're no longer resource constrained and you can use these copilots to help you, they can do a lot more than what they used to do before. And that, in effect, is more like an A player. It's still not quite— they're doing autonomously yet, but if the agents push them to work autonomously, maybe they become an A player there.

So I'm going to give you my controversial viewpoint on this. We see AI making the B players still stay B players and the A players being A players.

I don't think that's controversial. I think that's fair.

Because the A players do so much more and they're so much better. The B players are also doing more and they're so much better. But the standards have changed on what is an A and what is a B player. So what we're typically seeing is the biggest thing that is impacting is people are able to just execute faster, no matter if you're A player, B player, C player, and even lower than that. Execution speed is the biggest thing that we've seen. The second big thing that we've seen is it gives you the ideas on what to do and it'll surface things for you. We see it surfacing things for A players, B players, C players. It doesn't matter what level because some decisions like the credit card conversion example that I gave you, that's something that you would want an A player to take charge of and not— roll it out with a B player because it's something related to finance and money and it can actually have a big impact on money.

You need A player judgment.

You need A player judgment with that. Spot on. And then I would say the third thing that we're really seeing with AI is it allows them to run more experiments at a faster pace to assuming you have the traffic volume or you can buy the traffic to figure out what is the direction that we should take our campaigns. Because people all have different ideas. Now you can just run through them much faster and figure out what actually works and then just focus and double down on what works. While before it would just take you months and months, sometimes a year to figure out what's the right path.

So to Neil's point on that one, I would recommend everyone go look for Andrej Karpathy's Auto Research, go pull that GitHub repo, and then you can literally do this yourself. You call that— maybe you're using more synthetic data and then you can combine that with my Auto Growth, which is also open source as well. I don't have a good link for you there. Just search for Eric Siu Auto Growth. You should be able to find it. But Neil, let me ask you a question here. You mentioned the B players and A players, they become more of who they are, but the multiplier effect. So an A player using this now, how much better are they? Are they 5x better, 10x better? Just throw a number out there.

We're seeing A players be like 30-40% better.

And how about the B players?

Maybe like 50-ish percent better. We're seeing bigger lifts with B players and C players than we are with A players. This is a biased viewpoint from my perspective. Most of the A-players that I know of and interact with in my organization and in other organizations that have marketing, they're typically in management. We're only seeing so much efficiencies that you can provide to manager-level positions with AI versus the people who are actually doing. I do interact with the people that are doing, but I don't interact with them nearly as frequently as I do with management in organizations, even excluding mine. It's just usually I'm dealing with VPs and CMOs much more frequently than mid-level marketers and managers.

Okay, so 2 more things on this. So you look at your organization, okay? What percent of people, so if you think of AI as 4 levels, so you have unacceptable, you have capable people with AI, you have adaptive, and then you have transformative. Transformative, they're building end-to-end workflows on their own. They don't need to be like, they're completely reinventing how to work, right? What percent of your company would you say is transformative?

Because— Less than 10%.

Less than— so here's the thing. When I talk to my other friends and I see this the same for mine, it's like probably less than like 2%. Yes.

Yeah. I don't know the exact percentage, but it's not high. I know it's well below 10%.

So that— I was just curious. I just want to ask you that. But I— my controversial take is I think as a leader now in an organization, it used to be you wanted to be the chief inspirer and you still definitely do. But I find myself in a weird spot now. I have to be both the cheer— the chief inspirer and the chief fearmonger. Okay, so the inspirational piece, you will get your best people to perform at the highest level. That's what you want. That's a good result, right? Also, if you're the chief fearmonger, you're going to get some people to tap out. That is also an acceptable result. Okay. And so for me, it's like I've never had to do this in my career. But also the last time something big like this happened, we're both children. Okay. We didn't have to experience this change. This is the first time I've had to play both. And it's a very weird thing, but I see a lot of my friends doing it. I know you're maybe not as front-facing with it, but I just, it's like a thing that I've just been looking at. It's like, it's so weird to have to do that. All right. So I wanted to take a moment to tell you about my podcast co-host Neil's agency called NP Digital. And they work with a whole host of global companies or a global organization. Also, Neil has SEO tools such as Ubersuggest and AnswerThePublic. All you have to do is go to npdigital.com to learn more. And we'll see you on the other side.

I don't really see on, at least in our organization, we don't really have to do too much of the fear stuff. When we look at the most transformative people in AI, we've taken our approach instead of saying, hey, we need to make everyone an A-player with AI, we now look at it as that's unrealistic. It's like saying, hey, Eric, you're recruiting 50 new employees. I want you to take every single employee and take them through this training methodology, and they need to all go through this practice, be onboarded in this specific way, and learn how to use cloud code. All right, so that way they can do their job better in marketing. And then how realistic is it then 30 days later, hey everyone, OpenClock came out, you guys all need to go through onboarding and learn how to be the best at OpenClock. It's just not realistic in an organization. So what we found is there's— every organization has their elite tech people who are really good at adapting with times, and you don't have to push them. They tend to adapt before you even tell them about the new products and services because they get a kick out of it and they just love doing it. With those people, we just put them in charge of figuring out how to help different departments with AI and build the tech and provide it to them and integrate the SOPs. So that way people are just more efficient and we don't have to keep training these people on how to relearn a new technology or how to be amazing at everything. Instead, the AI adapts to them versus having, having everyone to be amazing at using AI when it comes to at least marketing.

Yeah. So our philosophy on this is we want to build the bridge for everyone, but not necessarily everyone's going to want to cross it. The majority of people are going to cross the bridge, right? How you bridge that gap. The other thing is, it is within your— like, if you're a leader listening to this right now, you should absolutely— you have to build this because if you know how to do all this stuff alone, it's not enough. It's on you to teach everyone how to do it, right? Because if everyone can spend the 20 to 40 hours over like a 60 to 90 day period to learn how to do this, Everyone's a lot stronger for that, and that's good. The other thing that Neil's mentioning here, so I said earlier, we have 4 levels that we're looking for, and this is adapted from Zapier. You have unacceptable level, capable level, adaptive, and transformative. We are not looking for many people to get to transformative. That's extremely hard to find. Okay? That's like best of the best. But Neil, we want everyone to get to adaptive because today's capable is tomorrow's unacceptable. So if you get them to adaptive, they at least will quickly fall back to capable, and then you gotta get them back up to adaptive, right? That's how we see it. But I will say, Neil, there's good, obviously there's There's going to be some stragglers that don't want to adapt, right?

And that's okay.

And that's why I say those people, it's okay for them to tap out. I'm okay with that. But I think the majority of the people at my organization, they want to do it. And, you know, we've been facilitating that. And I think it's the same thing for you guys too.

Yeah. But I think that is, you know, every organization is going to have to figure out what works for them. And there's no right or wrong, but this stuff is just moving so fast. It's too complex to try to get everyone to adapt.

You can't. And I think it's just like, remember I told you, I didn't tell you this. My dad said that, you know, it took years and years for all the companies, different companies he worked at, to adapt to getting email addresses, right? So like, I just, I don't think we have the luxury of that to wait this time. It's very different. So, and to illustrate this, and then we'll cover maybe another classic topic, but the Anthropic team doesn't write code anymore. Okay, so this guy's friend got hired at Anthropic 3 weeks ago. Nobody on his team had handwritten code in 3 months. Okay? So they haven't written like anything handwritten. They run multiple agents in parallel and act more like managers than engineers. This isn't just an Anthropic thing. This applies to your organization. It applies to my organization. Now you know what I do, Neil? I send people Looms on how I'm just going through Slack on how I'm working with the different— I'm querying different things. I'm like, guys, this is how you do it. You see how easy this is? This is awesome. Look at how much time you save, right? And you know what I'm doing now? We have this thing called Gather Town where we like go around virtually talking to people and I'll just randomly hit the gong and people will hear this gong noise and I'll walk over to And I'll pull someone in and I'll start training them. So the move now, guys, is if you are just watching an agent code, you're already behind, right? You should be spinning up multiple agents and directing them. And then at the end of the day, if you're working really hard, maybe your eyes hurt. Okay, I'm not recommending that, by the way, but my eyes hurt at the end of the day. So the point of this method is not to learn to use AI to code faster. It's you are the product manager, you— the agents are your engineers, and your job is to keep them running at all times. This doesn't just apply to engineers. It applies to people across the board in knowledge work. Yep. And by the way, if you're an electrician, I might tell my kids today, if they're in high school, just go be an electrician.

Dude, electricians get paid an arm and a leg, especially in data centers right now.

It's $260K, I think, starting or something like that. Yeah.

Yeah. Interesting thing that happened to me is I met up with one of my daughter's teachers in our old school. So not the current school.

What incentivized you to do that?

You had to meet up with them because parent-teachers and all that kind of stuff.

But you said old school.

In her old school. This is while she was at the old school.

Okay, I see.

Not current. So this wasn't recent. And the teacher was complaining to me how people who pick up trash, like garbage people, you know, that drive the trucks and, you know, empty out the dustbins, they get paid way more than teachers. And that's not fair. And I wasn't allowed to say this. I was like, But you could go be that. And they're just like, why would anyone want to be that job? I'm like, but that's why it gets paid more money because people don't want to do it. And they were like, that job is beneath me. And I was like, that job's not beneath me. I would just clean trash.

And you have done it before.

Yeah, I used to sweep up vomit. Literally every single day I was working, I would have to like clean up vomit. They paid me extra $0.25 an hour to sweep up the vomit. It was like $6.75, no, $5.75 an hour. To sweep up vomit. That was a lot of money back then.

Nothing is beneath me. I had to put, I had to put expired dates. I had to put dates over expired dates when I worked for a Chinese factory.

Wait, wait, wait. So the food was expired?

The food was expired. So they had us, I was like 17 years old that they had us putting new dates on top of the expired dates and they would pay us all cash under the table. And this is like a college in the middle of between going to college. And yeah, it was fun. And they didn't give us any air conditioner. We're like, we're in the back of the warehouse and it was freaking hot. Yeah. Yeah.

And then the teacher asked me like, is this fair? And I'm like, but you can choose your own profession. No one's holding a gun to your head.

Yeah. By the way, guys, what we're seeing is, again, whether it's with— we're talking about trash or when I look at that building with all the engineers working there, like when you talk about electricians, everything's supply and demand, guys. That's all it is.

They have so many shifts.

Yeah, they're working. I heard the construction 24/7. Yeah. Yeah. So, okay, let's pick one more and this will close us out here. So you can pick one more here.

Let's see, I closed the sheet.

Oh, here, I'll pick, I'll pick. So let's go with why the best— no, no, let's go. You know, we haven't done a tools one in a while. Want to do a tools one?

I think we talked about tools indirectly all the time.

Okay, fine, fine, fine. The 5 skills every marketer needs by 2027 that nobody's teaching. Okay, here, I'll start with that.

The 5 skills.

Okay. Yeah. So number 1, I don't know if we want to call this a skill, but I think You as a marketer, as long as your relationship with change, meaning adaptability, is good, you will be fine no matter what. I think it's the people that are unwilling to change, and I'm seeing in my org, Neil's seeing it in his org also, we're seeing this in other companies. The people that are very scared of the future right now, they're unwilling to change and they have an unhealthy relationship with change. They think they're more comfortable with things being stable when the reality is the only thing that's constant is actually change. So that's number one.

Number 2 is understanding data. Even though AI can end up breaking data down to you, in a lot of cases, it's It's off. And data is the backbone of marketing. Bad data means everything that you're doing is not going to work. So you need to focus on good data. What you need with AI, or what marketers are doing with AI going forward, is they're running a lot more experiments. You're analyzing which ones are working, which ones aren't, so that way you can figure out where to double down. If you suck at analyzing data, suck at figuring out if it's accurate, you're going to be screwed and you're just wasting your time.

Yep. Number 3 is storytelling. So if you don't know how to tell good stories, especially in a world where everyone's making the same thing, everyone's copying each other, you're not going to know how to stand out. The best storytellers, they know how to hook people in. They know how to tell a good story and they know how to make a good offer. That is the premise of how this stuff has always worked from an advertising standpoint. And that's going to become even more valuable. Like the value is standing out.

Number 4, having deep understandings of your ICP, your ideal customer profile. Everything that you're doing shouldn't be about traffic, shouldn't be about reach. It should be about getting more of your ICPs to your website, to your app, whatever it may be, and then eventually converting. But here's the kicker. When people say they understand their ICPs, eventually businesses expand with their marketing. They release more products, they go into more countries. And even if you use countries, for example, If your ideal customer is someone who works at a big corporation in marketing, the way you would market to them in, let's say, Brazil is going to be different than you do in the United States because there's some cultural differences. So when I talk about understanding your ICP, it's not just about knowing here's my ideal customer and who's who I'm targeting. You need to understand them culturally as well so you can figure out how to best target them no matter what country they're in.

Number 5, I was debating between another one here, but you want to, when you are working with AI, don't just use it on one screen. You have to learn how to work in parallel.

Okay.

When I say working in parallel, the creator of Claude Code, for example, he has 6 or 7 different agents running at the same time. That is the future of how you're going to be working. You're going to be running multiple ones. Maybe, for example, by the way, you spin something up, you want 20 articles written. Okay. They're going to create those in parallel for you. And then it'll create a strategy for you over here when it comes to CRO. And then you're going to have a go-to-market strategy over here, and then it's going to do some, some SEO work over here, some paid media work over here, some ad creative stuff over here. Like across the board, you're doing multiple things and your output should be a multiplier or a multiple on top of what you're able to do before.

That's it.

That is it, guys. Don't forget to rate, review, and subscribe, and we'll see you tomorrow.