Tim Soulo and Eric Siu discuss how AI is reshaping marketing agencies. Eric's team generates 21 blog posts in 30 minutes using AI for drafting and humans for editing, cutting time and cost by 90-95%. They agree that taste and experience still matter: AI produces slop without seasoned guidance, and agencies survive by asking the right questions.
Chapters
We'll probably debate on this podcast, but I think there's a lot of respect on both sides.
You created 21 blog posts in 30 minutes. Tell me, what's the state of creating content with AI and how do you approach it at your agency for your clients?
So if someone tells me, oh, it's 30 minutes, I get 21 pieces, I'm going to assume it's crap. It's good. It's not the same BS from like a year ago. It's actually there and it follows my writer guidelines. But think about it, like 1,000 days ago, we didn't have something like this, right?
Have clients who want you to create content. Why would they choose your agency if everyone would just use the same AI processes to create content that is essentially the same?
Why do you have a job right now?
What do you mean?
Why do you deserve a job right now? Why do why do any of us deserve a job?
You want me answer my own question? Yeah. Okay, my answer is.
Hrefs podcast.
Okay, so today I have Eric Sue on the pod. And we're going to talk a lot about AI in marketing because we have kind of different stance on it. You are very enthusiastic about AI in marketing. I'm a little bit reserved. We'll talk a lot about why. And because I feel from our previous interactions that our debate can get a little bit heated, I wanted to say from the very beginning that I have a lot of respect for you. You've built a very successful marketing agency. You're rubbing shoulders with the smartest people from the industry. You've been in the game for many, many years, and you just invited me to a conference, like amazing conference with amazing people. So a lot of people respect you. So do I. But that doesn't mean that I cannot debate with you and we cannot have a healthy debate and disagree on things.
Oh, totally. We'll probably debate on this podcast, but I think there's a lot of respect on both sides.
Yeah, I think it would be very entertaining for the audience. And I think I'm pretty sure you will be able to persuade me in some things and maybe I will be able to also kind of curb your enthusiasm a little bit, but we'll see. So what I wanted to start from is to learn a little bit about Single Grain, about your agency, things like how big it is today, like what's the headcount, what's the revenue?
Mm-hmm.
So yeah, can you tell us a little bit?
Yeah. So Single Grain, I took over this agency when I was about 26 years old. I had no idea how to run a company and made it go from bad to worse. That's a whole nother story. But it was actually Neil Patel's cousin's company. And a year into joining the company, I took the whole thing over and happy to talk about that story. And we have about 45 people today and happy to talk about that story too. We were at 90 a couple of years ago, so we've been, we've been cutting, and we might potentially, if anything, look to reduce the velocity of hiring just with all this stuff that we're going to talk about. But that's where we're at right now. We don't talk about revenue, but yeah, I have a whole— I have a very optimistic way of looking at business in the next 10 years or so. Yeah.
Okay. We don't talk about revenue. That's not good. Can we talk about some biggest clients or biggest checks at least?
Yeah, sure.
People love numbers. You know it. People love big numbers.
Yeah, people love numbers. So, you know, okay, one client, we'll just say they're in the streaming space. Let's just say they were paying us millions of dollars a year. And clients that you would know would be like Amazon, Uber, Lyft, Salesforce, Airbnb. And a lot of the clients that we get come from SEO. Like Uber came from SEO, Amazon came from SEO. And then, you know, you did that podcast with Neil. Obviously there's a lot of land and expand opportunities, right? So Amazon came in, they wanted, services around helping them for Amazon Alexa. Remember they had that SEO tool? So we helped them with that first and they're like, oh, can you just help us with Amazon ads too? So that started growing more and more. Happy to talk about that too.
So we're not talking about kind of services for small businesses from like $1,000 to $5,000 per month checks?
No, we're talking about— we're talking about monthly. Yeah. So us, we start monthly. I mean, there There, I would still say there's— it's more mid-market level, call it $10,000, $15,000 a month to start. And then it starts to go higher and higher depending on the type of service.
Okay. In terms of growth, how has growth been in the past few years?
Oh, it's been terrible. I mean, it's been terrible because we did these acquisitions and I stepped out of the business because the cool thing in 2020 was everyone's talking about this holding company, right? You start a holding company and then you hire a CEO and then, oh, you get out of their way and you let them do their thing. And I, along with many other CEOs on Twitter or X, whatever you want to call it now, everyone has now realized that no, you have to be involved. And so there's a whole story around that. But we're growing again. Last year we grew the fastest that we ever grew, and that was because I stepped back into the business. And interestingly enough, our, one of our finance people who's been with us for 11, 12 years or so, basically since I took over the company, she says, whenever I look, I look at the numbers, I don't even need to talk to you. I know when you're back in the business and I know when you're out of the business.
Mm-hmm.
So, you know, Neil and I did an episode just basically saying that it's a joke that people think they can just not be involved. The founders are going to drive it the most because they care the most and they're incentivized the most.
Absolutely. Okay. Now that we've established what kind of business you're running in the marketing space, marketing agency, let's start talking about AI. So I wanted to start from referencing the small research that I saw on LinkedIn by Pep Laya from Wynter, where he polled, I think, a few hundred, whether it was executives or business owners. And the main use case for AI among them was content creation. Second was research, and very little number of them were actually using AI to automate something in their business. So let's talk about AI use cases. There are a few. As I was watching your YouTube videos and you were discussing how you are implementing AI or how you're seeing the future of AI, there are many, many things that I've noted from that. But I want you to pick any use case where you want to start, or even if you want to comment on that research by Paplaya.
Yeah, I think that research is fine. I think where I'm coming from, even the dinner conversation we had, so We had— Tim was at a dinner. We had a couple of speakers at a— it was like a pre-speaker dinner before I did my event. And the— what I got from that one, the conversation was really around, is AI overhyped right now? Correct? That's the premise, right? And maybe I'll start more on the personal side and then we'll go into business. We'll go into marketing a little more. You see this right here? You see this little bite, right? Okay, let me show you something. So this will be fun for you too, Humberto. So you see this one?
Uh-huh.
This looks pretty bad, right?
Yeah.
So basically, if those of you that can't see this on camera right now, this is a bite. This is what happens when I get bitten by a mosquito. This happened after the COVID vaccines. And I'm not anti-vaccine, by the way. I'm just saying after the COVID vaccines, it just so happens that whenever I'm bitten by a mosquito, that happens. Now, this over here is a mosquito bite. You can't— maybe on camera you can kind of see it. But it's been controlled. And this solution, doctors couldn't figure out what this was. And then I took a picture of this, they figured out, it's like, hey, you have Skeeter syndrome. I'm like, what is Skeeter syndrome? And it's like, I'm like, how do I treat this? And so I'm like, oh, I'm coming back to Miami for sure. I'm gonna get bitten because they love my blood, right? The mosquitoes love my blood. And it was like, okay, you have to take Zyrtec and you have to put cortisone on it. And then you have to put an ice cube on it. No doctor told me that. And Before, I was putting a heat pen on it. You're not supposed to actually put heat on it. I was putting ice on it. This would be huge right now if I didn't get that solution from a simple ChatGPT search. I'll give you one more. So I have LASIK. Neil has LASIK as well, right? And Neil is actually a pessimist when it comes to AI stuff, but he's generally a pessimist.
Oh, I didn't know that.
He's a pessimist, yeah. And he will say, I'm much more of an optimist. And I think it's fine. It's just different personalities. Basically, if you get LASIK, your eyes after a decade— so we both had it for over a decade.
Not medical advice.
Yeah, none of this is medical advice, guys. Sorry. Basically, your eyes start to dry up because your lipids aren't working correctly anymore. And so I found a solution using ChatGPT. It's like, oh, you got to use the brooder. I'm like, what's a brooder? It's an eye mask that has water in it. You microwave it and it heats it, right? So this event, this was a lot of work, right? Obviously. Every time before I go to dinner, I have to go up to my room, heat it for 25 seconds, put it on, and my eyes are good to go again. Otherwise, they'd be hurting like hell. Funny enough, when Neil and I were recording a podcast, right before he was about to speak, we went up to my room to record podcast. I saw his phone. There's a picture of the Bruder. It's a picture of me holding the Bruder. I'm like, why'd you take a picture of me? He's like, oh dude, because I have that issue too. And so my point of saying all this is that told me to use that temporary solution, but it also told me to look into this thing called the Lippiflow. And my friend's like, oh, I used the Lippiflow. It solved all of my issues. I got that for free, dude. And so those are 2 places.
Let me interrupt you. You finished anyway, so no point of interrupting. I don't disagree with this. AI, ChatGPT, I know Perplexity, Claude, whatever, are great research tools. They speed up research significantly. Let's talk about marketing use cases, creating content. So You actually published a post on LinkedIn about Manus AI where you showed that, oh, and you mentioned in one of your videos you created 21 blog posts in 30 minutes. You had a post, a similar post on LinkedIn, and I commented on it and I said, the real question is how many of those blog posts you actually published on your blog. I got kind of quite a few likes on this because this is an ultimate test of how good this is.
Yeah.
Tell me, what's the state of creating content with AI and how do you approach it again at your agency for your clients?
And I think it's absolutely right to be skeptical because I am— this is where I'm actually skeptical, the content creation piece where it just spits it out very quickly. Right. So if someone tells me, oh, it's 30 pieces, 30 minutes, I get 21 pieces, I'm going to assume it's crap. The problem now is It's not a problem, it's a good problem actually. It's good. And we've been publishing a lot more on my blog.
Yeah.
You can actually see it on my blog now. There's a lot of long-tail, high-intent keywords around ABM, 'cause that's one of our products right now with Carrot. And we're publishing a lot of it. And a lot of it is 90, 95% there. And then when my, my SEOs are pretty good now that we've kind of cleaned house on that team, it's a really good team. And they're just like, dude, this is 70, 80% of the work that we do. And it's not the same BS from like a year ago where it's like, oh yeah, we can do 100 pieces content for you. It's actually there and it follows my writer guidelines.
Let me get this straight. So yeah, you have this product Karrot, which we'll talk later about. It has its own website, right? What's the address of the website?
karrot.ai.
karrot.ai. Is it 2 Rs and 1 T?
2 Rs and 1 T, yes. But with a K. K-A-R-R-O-T. Yeah.
And you're saying that there's content on that website in the form of a blog?
No, we're using the domain authority of Single Grain.
So you publish the content on Single Grain?
Yeah. And we have call to actions that go to Carrot.
So what I'm trying— I'm trying to understand how do people who are interested in seeing like the quality of this content find this content that was kind of 21 articles in 30 minutes.
So if you go to Single Grain blog, singlegrainblog/abm should have all of these posts.
And they are created with AI?
Correct.
Fast.
Fast, but we do have a human editor in the loop. So I want to preface this by saying, I don't think we're in a place right now where you can just let it go. We still have editors in there and they're spot checking everything. And oftentimes here's where AI gets it wrong. It gets a lot of the stats, it just makes up stats, right? So we'll fact check the stats and then we'll change some of the internal links and then we'll change some of the external links as well and we'll add images to it. That the human will handle the last, you know, call it 5 or 10%.
Okay, so 90% AI, 10% human.
Yeah.
Okay. Interesting. That, that doesn't fit the paradigm of 21 blog posts in 30 minutes because if a human needs to work 10% on each of the 21 articles, that's not 30 minutes.
So the way I would break it down is you have the human editor and I agree with you, you have a human editor that's doing the work, right? But you no longer have the human bloggers doing the work. So in my mind, the blog posts are still getting done in that 30 minutes. In my mind, the other part of the process is the human editor. But still, at the end of the day, if I look at the entire process, our process says the time to get this done has decreased by 90-95%, and the cost has decreased by 90-95%. That's what I ultimately care about as a business owner.
Okay, okay. So again, I'm not anti-AI, and in Ahrefs marketing department, we are, investing a lot of effort to figure it out and figure out how to implement. And the reason I'm asking about content is because Ryan Law, our head of content, has actually devised a pretty cool AI workflow where he kind of automated, automated, productized. I'm not, I'm not even sure what's the right word to use. Maybe our—
maybe both.
Maybe both.
Yeah.
Our whole publishing process from kind of picking a topic, discussing an angle for the topic, generating outline, expanding on this outline, and then polishing a post. I think he created custom GPTs or whatever. I'm actually asking him to publish this process on LinkedIn because I know that like people would go crazy about this process because it works so well. So he, at Ahrefs blog, we actually published a few articles that are also 80 to 90% AI and like the remaining 10 to 20% is Ryan because he, Ryan, is literally acting as an editor and AI is acting as a writer. So yeah, 2 things here. First of all, he can only do this for kind of generic informational queries. So if there's a lot of content about it online, it is very easy to work with ChatGPT to kind of collate this content. Form it into an article, and Ryan is overlooking every step of the process. So we cannot use this process on 100% of our articles. We can only use it on, let's say, on stuff that's kind of already there, but we want to spin it at a different angle. So that's the first thing. The second thing is, as far as I remember, Ryan was saying that he is still spending up to 2 hours per article. That's still a lot faster because if he, if he needed to write it from scratch, that could easily be like a whole day. But if you spend up to 2 hours per article, you can do like 3, 4 in a day. So 100% productivity improvement. But it is kind of limited in those things that I just explained to you.
Yeah. So it's interesting because earlier you mentioned, you know, AI is good for research. You talked about content creation, all that. This is actually a comment I think that you left on the Pep post, right? It's like, it's being used primarily for those cases, and which is what Pep said as well. Yeah, but isn't a lot of work research, and isn't a lot of work content creation and information sharing at the end of the day anyway?
True. I guess the, the root of our debate is, uh, it's, it's one thing to look into the future and, uh, kind of extrapolate what is going to be possible, which I agree with you. Like if we look into the future, it gets crazy by the year versus what is happening today. And I kind of— I'm not a fan of overhyping this technology based on the future because it creates a lot of anxiety in people. Like, oh, they're talking about all these awesome use cases, but I'm so behind because I cannot one-shot an awesome article. How is it? And people, like, what's left unsaid is what we just discussed in terms of content creation. Because like I said, Ryan is able to create amazing content, but he still spends 2 hours on it. And then another thing that I wanted to discuss, where is it in my notes, uh, is that Ryan is an incredible, experienced professional. He created tons of content himself. He knows what exceptional quality looks like. And this is one of the things I mentioned in my talk at the YPO conference where you invited me, the importance of being able to recognize exceptional work. So if you give AI, if Ryan would give his process to some entry-level copywriter who's just in the industry for one year, they won't be able to create the kind of quality article that Ryan is able to create with AI, because AI is just a tool and the quality of output is still controlled by you. So yeah, it's one thing what people are saying about AI looking into the future, like this is a crazy technology, revolutionary. I agree to that. But as of today, you cannot just click a button and create a result.
I think to your point, what you're trying to say is you need to have taste at the end of the day. You need to have it. And taste comes from experience. And that I agree with completely. Now, who knows what's gonna happen in the future? Are we just gonna implant a chip and then we just know all the knowledge? Like, I don't know, right? But I do think that it is still imperative that people continue to talk to people, they continue to learn, and they continue to experiment. That part has not changed. And I hope that doesn't change for a while. It probably will. But that's why, like, for example, if you're a really good engineer and you know what you're doing, been doing it for 15, 20 years or so, versus like an intern, an intern's just gonna ask basic questions. But if you're an amazing engineer, you are going to know the right questions to ask to leverage these products to move a lot faster. And you saw me, I mentioned that our engineers are 10 times faster with Cursor. And I think your engineers are using Cursor too?
Yeah, yeah, yeah. Dmitry is pushing everyone to use AI because it is like, it is a huge productivity improvement.
Dmitry is your CEO?
Yeah, yeah, yeah. CEO and founder.
There you go.
Yeah. Yeah, he definitely pushes because he's using it himself a lot. And he says like, I'm getting to prototypes super fast. What, what would take me weeks now takes me hours. So with that, again, I absolutely agree. But then once you have the prototype, you have to spend a lot of time refining it and you have to know what you're doing.
This is interesting. So you actually posted something that you lost a bet recently. And I don't, I don't feel like you lost the bet because I actually sit in your camp there when people say you can build a product, I'm like, great. Yeah, you can get it 70, 80% there, but then you're stuck because you didn't— you don't know how to deploy. Or then Replit might be able to help you deploy. But it's still not quite there yet, right? I think we'll get there eventually. So that's where I think it's a little, you know, it's pretty good for where it is right now. But think about it, like 1,000 days ago, we didn't have something like this, right? I do think you just mentioned prototyping. So now our expectations— now before, our expectations at my company was, oh, you have to write a memo if you want something done. Now it's you have to create a prototype. So you might have to log into v0 and make a prototype of a screen, an app or whatever you want, or landing pages. And you can drag in whatever you want and you move a lot faster. And the crazy thing is our SEO managers are now becoming product managers for our internal product called ClickFlow. And we just use it internally for our clients and they are speccing out exactly what they want. And then that allows our engineers to move faster. So when I think it's really where it should be, right? Like we're just, even you guys, like you're, Demetri's saying, we're just a lot faster. And I don't think I've been so excited about work in a very long time.
True. It opens a lot of possibilities. I wanted also to give the listeners a little bit of context of how I lost $500. Basically, I had this idea of a simple LinkedIn Chrome extension, and I can talk about it right now because I'm going to publish it very soon. Basically, I had a spreadsheet where I have a few names of people on LinkedIn who have kind of similar-sized audience with me. Just like 5,000 more followers, 10,000 more followers. And I wanted to track like every couple weeks how our audience is growing. Like, am I catching up with them or are we growing equally or are they like flying away from me and I will never catch up? Kind of to understand how is my content strategy on LinkedIn compares to theirs. Like, is theirs more effective? If mine is more effective. And then I thought, why am I doing this manually when I can just automate it? Yeah, LinkedIn is not happy when you scrape it and like they don't give you follower counts via API. So I messaged like friends from Ukrainian dev agency to give me a quote of how much they would take to develop very simple version, kind of no fancy design, nothing, just like make it work for me in my browser so that I wouldn't have to do it manually. They quoted me for $1,000. And then I thought, what about vibecoding? It sounds like it's not a massively complicated application. So I posted on LinkedIn, who's willing to do it for less than $4,000? And yeah, one guy said, well, I'm actually interested to take this challenge. I don't even want money. If you want, like, we can have it as a bet. You can pay me if you like the app, but that's not critical. I said, sounds like a deal. Like, if you create an app and it would work, it's not the app that I wanted to publish and other people to use, but I just wanted to see, can you prompt your way to an application that would actually work and do what I wanted it to do. And he did it. But again, he has engineering background, so I still doubt that I would be able to do it.
I don't think we can do it because we're not technical.
Yeah, exactly.
Yeah.
Again, would I take his code and publish it for other people to use? Would I take his code and continue iterating on it and improving that Chrome extension? No, I wouldn't. I would still go with the agency. But yeah, the expectation of the speed of development is definitely going to be higher. You're going to expect from, I don't know, your agency, from your development department to be a lot more efficient and move a lot faster.
Yeah, I think, look, expectations to your point are just going to rise, not even for developers, but across the board now. Because again, when I talk to my friends, when they say they're on a rampage, they're either not hiring more people or they're looking at cutting more people that aren't adopting this stuff. And so When I think about our AI standards of excellence that we created at Single Grain, it's a 12-page document just outlining what exactly we're looking for. And originally someone on my team left a comment saying, oh, what about training? And then as I started to think about creating training, I started to get more and more upset as I was writing that document. I'm like, this stuff is changing so quickly and why do you guys need training for this stuff, right? If you need training for it, like you don't belong here. And so what the document now is just like, hey, here's what we expect, here's some starting points to go with, and that's it. The Shopify CEO wrote something recently and it's only 1 or 2 pages and it's just basically we expect reflexive AI and if you don't know it, then unfortunately we're going to have to move on. And I just, I think this will be a good— and this is more philosophical long term. We can go back to marketing, but I just think it's going to be a net benefit as we transition into this new world long term, because the people that want to be doing this stuff, they're going to be the best of the best. And the ones that don't want to be working at these, these, these jobs that they hate doing, they can work on something else that they like doing. And I think it's going to be a positive for all of humanity.
Okay. We need to curb down the philosophical aspect and return to marketing because, yeah, this is the audience I'm trying to create this content for and to help out. I'm also trying to help out myself because I'm learning from these discussions a lot.
Yeah.
You actually, in your videos, you admitted AI slop, the term. You, you, you admit that a lot of people are going to abuse AI to create a lot of low-quality content, and a lot of people would copy each other's framework to create kind of minimum viable content. Let's call it like that. It's kind of— it kind of reminds me When Brian Dean published his Skyscraper method, everyone started using this Skyscraper, everyone started using the same outreach templates, and it just stopped working. So you admitted this, and you also said that humans would want to interact and get content from other humans.
Mm-hmm.
So again, it's still a little bit philosophical, but tell me about How you think of this as an agency owner. You have clients who want you to create content and like, why would they choose your agency over any other agency if everyone would just use the same AI processes to create content that is essentially the same?
Yep. So I'm going to give some couple— I'm going to give a couple of examples here. So I have a friend, her name's Chloe Shi. So Her Instagram grew from 100,000 to 1 million in basically 2 months. And she didn't buy her followers or anything like that. All she did was she really, to your point, she started skyscraping a lot of her content. The types of content she would edit would be like, imagine if it's just you creating a Short, and all of a sudden there's 5 Tim Solos surrounding wearing different clothes. To create that shot, you have to change clothes, you have to sit in different areas. She's doing all that. It's ultra high quality short-form content. And that's how she grew. So there is, I believe there's a flight to quality. You know, I think that, you know, if you create an AI avatar right now using HeyGen or whatever, it's just not there yet, right? So that's, we're talking about slop. And, but to answer your question directly, why would they choose my agency? One, why do people choose any agency? Why did they choose NP Digital? Or why did they choose us? Well, one, most of our customers are SaaS and e-commerce. So we've been there, done that. So oftentimes when clients come to us, They just really want to know, am I going to lose my job if I pick you guys? And have you guys done this before? It's just like if you tear your ACL, I hope you never tear your ACL, but if you tear your ACL, you want to go to an ACL doctor. You don't want to go to like an elbow doctor, right? You want someone who has been there, done that, and you want referrals. And so for us, the vast majority of our customers come from either referrals or from SEO. That's how they come, right? And then after that, it's about just getting on calls with our team and figuring out Can we trust these people? It's just a confidence-building exercise. And I think there's a lot of people out there that sell gelato that tastes the same, right? It's just the story that you tell. And do you have confidence in these people? That's what it is for us.
Again, in one of your videos, you said that you were working with some company. It was somehow related to real estate or housing or something like that. And you asked Claude for marketing ideas and it suggested you some marketing idea. The question is, why would those businesses who typically hire agencies to help them with, I know, marketing strategy, marketing ideas, use you as an intermediary between themselves and AI? Why not just go and ask AI directly, hey, I have this kind of business, I have these kinds of challenges, what do I need to do?
Why do you have a job right now?
What do you mean?
Why do you have it? Why are you the CMO of Ahrefs right now?
But I am— I'm on the skeptical side about AI.
No, I'm just saying, why do you deserve a job right now? Why do any of us deserve a job?
You want me to answer my own question?
Yeah.
Okay. My answer is taste. My answer is taste and experience. I know what would work. I know what would resonate. ChatGPT can give you a lot of information, but it doesn't understand other people, it doesn't understand your audience on the same level that you do. I've been doing it for over a decade. I am genuinely interested in this. I have a track record of doing all sorts of projects. Some of them flopped, some of them didn't. That developed my intuition of what is going to work, what is going to not work. So instead of asking ChatGPT, oh, give me like 10 ideas Based on my experience and based on my taste and based on my understanding, my deep understanding of our brand, our company, and my team resources, what's possible, I am able to make a better decision.
Great. So at this event that you just came to, about 30% of the attendees were agency owners. And I was talking to one of the agency owners. So let's say they— let's just say they do really well. Let's say 9 figures a year or something like that. And he's like, you know why we get hired, Eric? I was like, why do we get hired? He's like, because people like you and me are constantly at the forefront of things, going back to the taste piece, right? And so if it seems like you can trust this person, you're learning from this person, this person's constantly at the forefront, and it seems like the team knows what they're doing too, you have confidence in the team, and they're also getting enhanced by AI. Because the way I look at it is, you know, me, someone like me using Claude, I'm going to ask better questions, and then I'm going to be able to move faster. I'm just brainstorming it with back and forth. And to me, I just look at these AI, these LLMs as AI assistants right now, and I get better results. I get more horsepower for my thinking.
Okay, good answer. I can't disagree with that because we essentially said the same thing. You still need the person who knows what they're doing. AI is just— I really like how Gail Breton, one of my other guests at the Ahrefs podcast and my long-term friend.
Authority Hoppers?
Yeah, yeah, yeah. He said— he put it amazingly. He said that what calculator did for math, AI is now doing for content. It's just a tool to help you content marketing and whatever. So it's just the tool that helps you move faster and like churn out through ideas, churn out through variations and stuff like this. I really like that.
You know, the, the way I look at this is if we have power tools to build a home, why would we not use the power tools?
Yeah.
That's how I look at it.
Let's talk about the use case of note-taking, because this is again one of the most mentioned use cases whenever I'm watching videos about AI and how people are implementing AI. Yeah, you should have an AI note-taker that joins your every call, every one-on-one, every meeting. It automatically creates notes, automatically creates to-do list. My experience with that, because we signed up for the tool called Fathom, I think it started joining all my calls. I never look at those notes afterwards. And let me, and let me tell you a different side of the coin. What I actually realized was very helpful and very useful for me. So after one-on-ones with like heads of departments within marketing, within our marketing at Ahrefs, I realized that sometimes we would discuss things and things get forgotten. So after I was completing a call, I was making notes and sending to a person, here's what we are, here's what we agreed on. I'm expecting that you would do that until our next call. So I was personally making those notes. Then I realized, wait, why am I spending time on it? when it would be kind of more efficient and better for the company if I would just move on. And the person who we had one-on-one with would spend time like making notes from our call. And when I got the notes from the person that I was doing one-on-one with, I realized that we didn't understand things similarly. When I was reading their notes, I was like, no, this is not actually what I wanted you to do, and this is not how I see it. And I was commenting on their notes. And we made this kind of a process. So after each one-on-one, the person has to create notes. I would look at those notes and I would comment to make sure we are aligned. But it seems like the actual exercise of note-taking is universally kind of agreed that it is very good to have your thoughts in order and to have you memorize things better. Because again, when you go to a conference, The MC was encouraging people to take notes. On every table, there were like sheets of paper, there were pens, and people were encouraged.
Nice.
People were encouraged to take notes. Why didn't you just create an AI note-taker which would like listen to all talks and create all notes? Because people won't look at them. So this is my kind of gripe with AI note-takers, but I want to hear your, your experience, how you're using them, and are you actually coming back to those notes knowing that someone else creates them for you so you don't, you don't need to process the conversation that you had?
Yeah. So I'll start with the conference first. There are a handful of people that wanted to have note-takers on for this event, but it's because it's a private YPO event. There's a lot of some of the talks we can't share, so we couldn't blanket record everything. We were planning to do that. So there's that. Now we use Gong, we pay for Gong for sales intelligence, but we also use for client intelligence too. So Gong actually sits in, if you and I are having a one-on-one, for example, let's say I report to you, Gong will be sitting in there. So we have it across the organization. So we can see if we can quickly check if people are doing things correctly or incorrectly. Now from a sales standpoint, the cool thing about Gong is it allows us to ask AI anything. So it has a full transcript. That saves us time. Now, to your point, I actually don't go back to my notes a lot, but my team, like it's, so my team uses Asana and they try to assign tasks to me. I'm like, guys, don't assign me tasks, right? Do as I say, not as I do. I'm trying to create these guardrails for you at the end of the day as the business owner. But our team does quite often go back into the Gong calls and they'll go back and listen and then they'll go back and they'll ask the AI questions. And that does save them time ultimately. I use Granola for my note-taking. I think Granola is one of the best note-taking products out there. And it just automatically takes these notes. It summarizes it very quickly and I'll push it to HubSpot and I can save it to Slack.
And you don't come back to them ever?
I do come back to them, but very rarely. But also keep in mind, Tim, this is what I— if you look at my personality test, my detail-oriented is 6% out of 100%. So if you start talking to me for too long, I start checking out of the conversation. So yeah.
Okay. I absolutely agree with the Gong use case, especially for salespeople. If you have sales calls, they're all recorded. You can search through them. You can derive information like what do people typically mention who kind of convert as customers? What are the objections of people who don't convert? This is amazing. But for things like recording team meetings and one-on-one calls and creating a list of things that have to be done, Automatically, I don't like it. I feel it removes agency from people themselves.
So let me tell you what I like about having all this data. So let's say we do all these one-on-ones throughout the organization, right? And I can't be there or the leadership team can't be there for everything. But there's a certain way to do things, just like there's a certain process you have for doing your podcast, right? If someone came in and started like I started messing with your process, like you don't like, I can tell if I start to go off in a different direction, like you try to like You know, it's— I know that's how it is right now. If that's not managed throughout an organization, it starts to become like people are like, oh yeah, let's move the one-on-ones. Let's not do the one-on-ones. And then we usually, if you like, let's say I'm reporting to you, we have a working document where it shows the primary KPI. Is it red, yellow, green? Am I on track, off track, whatever, right? Where I'm going with this is that people start to forget the purpose of that meeting. And if I come— if I keep coming to you in these one-on-ones, I'm like, Tim, oh yeah, I have this problem over here. I have this over— Here you go. Take my monkeys. Take my monkeys. Take my monkeys, right? That happens so much and that has to be managed. Now, because we have all this data now, in the future we're going to be like, hey, Dragon, here's the process right over here. Are we following this process? Who is not following this process right now? Who needs to be monitored a little more? Who needs to be kind of moved in the right direction? And that's where I see the use case being very powerful.
First of all, again, context for listeners with the monkeys is there is a great article. I think it's called Who Has the Monkey? Right.
I don't know, but yeah, go ahead.
Yeah, I think it's called Who Has the Monkey? Google it up. Amazing article about delegation and management. So yeah, just— but I didn't quite get what you mean there. So you're on one-on-one and it takes notes and how does it help? Once again, maybe I zoned out for a second.
It's okay. I'm boring sometimes. So How this helps is it ensures that we're following our one-on-one process. So here's the thing. If I'm bringing you all the monkeys all the time, we're actually— you as a manager, you're actually not following the one-on-one process in terms of enforcing and saying, hey, Eric, we actually expect you to come with 3 solutions every time you come with a problem, right? That is the expectation now. And we expect people to have something 80, 90% solved before they bring it to their manager, because most people like to bring their problems to their manager and then their manager ends up taking everything on more and their manager ends up not doing their job. That is— that happens much more often than we'd like it to happen.
Okay, that still went over my head. People in comments, please tell me if you understood.
Let me go back to your note-taking thing real quick and then we can move on.
Yeah. Wait, wait, wait, guys. If you understood what Eric was saying, please explain it to me in comments.
You want me— I could explain in a different way.
You want me to explain? No, no, no. Let's move on.
So the note-taking thing, and then we can move on. So I've been doing that 5-minute journal doc. You know what that is, right? So every day you're filling out what you're grateful for, what you're able to accomplish today, and then 3 amazing things that happen at the end of the night, all that. I've been doing that for about 11, 12 years or so, very consistently, okay? I never go back to those notes, but the act of writing it down, like there are papers out there that show that you writing something down, actually, you actually remember.
But you don't write it down with AI. AI does it for you.
No, that one I actually write it down.
But my, my whole point of this note-taking—
When AI writes it down for you? Yeah, you're right. But my point is there is value in going back to the note-taking piece. There is value in the note-taking, physical or even me using Granola. There is value because I do come back to it from time to time. When it's physical, sometimes I never come back to it. So there's different use cases.
Okay, guys in comments, who won the debate on note-taking?
Because I don't think there's anything to win. I don't even actually think we're debating on that.
Let me get my engagement, please. Okay, okay.
Let me get my engagement. Guys, give him his engagement. Tell him he won.
Scheduling calls and meetings. You, in your videos, you mentioned a bunch of tools for scheduling calls and meetings. But again, I'm throwing another curveball at you. That's, that's the phrase in English, right? To throw a curveball when you try to ask a tricky question.
Go ahead.
Yeah.
Yeah.
I looked up the correspondence between Michelle, my executive assistant, for booking this very interview. I didn't feel there was any AI involved.
This one, there was no AI. No.
Okay. Tell me, tell me when and how you're using AI to schedule your calls and meetings.
Yeah. So one simple— I mean, it's like a Calendly link. I use Reclaim.ai. That's, that's just like another Calendly app, but it is a little more intelligent where it doesn't move my high priority items. So you can use a scheduling link for that. The second thing I use is called Howie.ai. So I just started using that recently. And that one is where I might say, let's say you might say, hey, Howie, can you schedule this with Eric and I? Keep in mind he's in Pacific time zone, right? It will actually factor that in and it'll work on scheduling that. And it actually does the back and forth. The thing with Howie is it is AI, but if it can't figure it out, it does have a human in the loop there. So that I pay $29 a month for that. booking assistant, and that's actually done a pretty good job. And I've tried a handful of those.
So why didn't you use AI to schedule this interview?
This interview was scheduled a couple of months ago.
I think you didn't use the tool before. No. So if I email you right now and I say, Eric, let's schedule part 2 of this podcast interview, I would bring—
I would bring Howie in. Yeah.
What do you mean, I would bring?
I would bring Howie in because I would still need to be the one saying— because this is still This is very much done with you.
What does it mean I would bring? You would, uh, like, is there a button to click to bring Howie in? It doesn't automatically pick up the email.
You have an assistant, right?
Yes.
Okay, how does it work with your assistant? I want to schedule something with you.
Uh, she would look at my calendar, uh, she would find a spot. We already talked with her a lot about kind of the—
but do you need to tell her?
No.
Okay, she has access. She has access to your email.
She has access to my email, she has access to my calendar.
Mm-hmm.
And it was a process of us discussing previous cases, what I want to book, what I don't want to book, how I prioritize things. For example, if something is very important, she knows that she can schedule it at 6 AM Singapore time because Singapore and the US, they don't work well. If something is not that important, she would be very kind of defending my time. And she would say, only Singapore work hours if you want. So like, she has all those frameworks in her head because we've talked with her a lot about it. So when I'm asking you if AI is going to pick up, so you get an email, you see that it's like a request to schedule a meeting. What happens next? How do you, how do you engage AI in this?
Oh, this one's simple. So I just hit reply, I respond myself and I say, Howie, find a time for us. Right? That's all I say. Found a time for us. I might have to say you're in Pacific time or whatever, but there's a little work there, but I just have it do that.
So, so it's kind of like, hey Siri, once you mention Howie in your email, it picks up that you are talking to it and it would take over conversation. It has access to your email and it would continue talking.
Yep. And then I'll also ask Howie sometimes I'll just say, hey, what are my, what are my meetings for, for next week? Right? Like, what types of meetings do I have? Or I might say, hey, Howie, Cancel all my meetings next week except for my, my, my Ahrefs interview, my Tim Soulo interview for next week. It will do all of that.
So if I'm not sure how many views this interview would get, I'm hoping a lot. So if I would ask my audience to troll you and just inundate your inbox with requests for a meeting.
Mm-hmm.
You said you still need to work.
Yeah.
You still need to ask Howie to work on it. So you just ignore all these emails.
And by the way, something I want to call out too is It seems like you're okay with having like an executive assistant look at your inbox, which I think is fine. That's just a personal preference.
Yeah, it's my preference. I have personal inbox, which is at Gmail, and the professional at Ahrefs. So yeah, like I separate those two.
Yeah. And so where I've come to this, and Neil's the same way too, is we used to have assistants looking at our email inbox the whole time and managing all our calendars and moving things around. I think even, you know, love or hate Elon, I think he had an assistant too. And I think he no longer has someone managing his time like that. My philosophy now is I don't— maybe I don't want someone controlling my time like that. I want to be managing my own time. And so I've had probably— I've had like 7 executive assistants and like 3 of them had complete control. I really didn't like that. So it's just a personal preference thing at the end of the day. Yeah.
Well, that's, that's a whole different conversation on personal assistants. Do you want to bring up any other cool use cases of AI for marketing specifically, but maybe also for running business? But please, no legal or blood work or those kinds of things.
Why don't you like those things?
Because they're not marketing. This is a marketing podcast. I'm not Andrew Huberman or something like that.
Yeah. So I'll give you one thing. This is more tied to business. So, and we can tie this in with marketing as well. So let's say I have— you have a— you guys have a lot of marketing SOPs at your company. Yes. You guys have a lot of documentation, a lot of courses, correct?
Yeah. I'm not sure where this is going, but let's say yes.
Okay. You don't always have to know where it's going. Just participate, you know? So with a lot of the SOPs that we have, we know that when we create SOPs for our— so standard operating procedures, documentation for our company, nobody goes back to those things, right?
True.
And also when you want people to follow these, it's also not fun for people to have to follow directions, like, because humans want to do things their own way, right? Now what I do, what I've been doing, what I've been playing around with is I've been dragging these SOPs into Gemini 2.5. And so let's say it's a marketing SOP or a content SOP. I just say, make this into a game, right? And it doesn't have to be like a crazy, like Mario game or something like that. It can just be a game where you're quizzed on things and there's a leaderboard. And then the winner of the leaderboard, maybe, you know, they get X amount of dollars at the end of the month or something like that, right? But in my opinion, because I played a lot of games growing up, the more you can make something fun, gamify it, the more people are willing to engage with it. And I look at life as a game. I look at this as we're putting deposits into our relationship bank over here, right? Yeah. And so I look at any Loom video, any SOPs that you have. If you can collaborate with AI, you can actually make it more engaging. And this applies to marketing, business, or really anything. That's one example.
Make it a game. I actually thought this is going the other way because a couple of days ago I recorded an interview with Nathan Barry. And we discussed a very awesome use case, which again, I fully embrace. When they were rebranding from ConvertKit to Kit, they hired this agency and they created their kind of brand guidelines, visual guidelines, but also kind of messaging and copywriting guidelines. And this is exactly what you said. Whenever you have those big documents like, this is our voice, this is what we do say, this is what we don't say, this is how we position ourselves.
Yep.
People are lazy, so they wouldn't come back to that. And slowly your kind of adherence— is the right word— to those guidelines is fading away, which is why with AI it is now so easy to just upload this document into like a custom GPT or a Claude project. And whenever you're creating a landing page, messaging, flyer, whatever, you can just say, can you check it against my guidelines if it follows? That use case is Brilliant. I absolutely agree with this.
We do that too. That's great. I'll give you one final one. So, and I'm sure you guys will build this in if you haven't built this in already, but we're just doing this for our clients. So with ClickFlow, we have these, people like using the phrase agentic workflows, right? So we have these agentic workflows that help us very quickly find 100 internal links and it will figure out, okay, what pages, where are you missing internal links? What keywords should you be adding internal links for, right? And there are a handful of tools that already do this, these workflow type products. I'm not going to name any names.
We have automatic internal link building.
I'm sure you do, yeah, right? And, but to me, this is becoming much more table stakes now where anybody can build this stuff, right? And you guys have a lot of interesting data and I'm not, I think you guys are certainly more ahead versus the other SEO products out there. But for us internally, we're like, okay, well, Manage can create those blog posts, but can we go beyond that? And can we add in that editor and the loop piece? Can we get it even closer? And so those are the types of things that we've been building and testing on clients right now. We're at the point right now where we're like, okay, we'll just give 25,000 words for free. And 25,000 words before, we used to charge maybe $5,000, $10,000 for that, right? Now we're just like, okay, we'll give it free for the first month and then we'll see what we can price moving forward. But for me now, I'm like, how can we use AI to build really nice offers? Because it's not just an ebook anymore. It's, we're going, you know, up the stack. We're splintering off a lot of different features and giving them away for free, which you guys do as well. So I think we're going to see a lot more of that from marketers.
Using AI within your product, within the software that you're building, is absolutely necessary and it absolutely elevates your product to a whole new level. We even published an article, AI Features of Ahrefs, and we're constantly updating as we add more AI features to Ahrefs. Because yeah, there's so much you can do. Internal link building, we're going to Literally one-click it because we already have a system that helps you identify internal link opportunities. We have AI that can automatically insert links into like proper phrases. And then we have this thing called patches. So you have a script on your website or you have a Cloudflare worker if you're using Cloudflare. You click one button and it just populates your website with internal links automatically.
Yeah.
So yeah. Using AI, like improving AI in your product, using this technology within the product, that is, I'm absolutely a believer. That is, I'm absolutely enthusiastic about this. It's just, it's kind of different from because adding internal links is a very kind of low-level job. You have like the most junior marketer joining you and in their first week, they can follow this thing. So yeah, I'm— this is why that doesn't sound like groundbreaking. But of course, at scale, if you have a product and tens of thousands of people are using your product, you're creating a lot of value by implementing those kinds of features.
By the way, the point I mentioned that it is not groundbreaking, right? It's just a feature that you give away for free. But it is above like an ebook, for example, if your ebook sucks, right? If you don't have really good, interesting data, like these are just tools you can give away for free. And you and I as marketers, we've been playing this game for a while. You mentioned something very important. I— and I've said this a couple of times. Neil and I were recording the other day and I was just saying the future of marketing, you're going to see a lot of one-click marketing where it's just like, can you just do it for me? Like, here's a suggestion. Can you just do it for me? Because at the end of the day, people don't want to do this work. Like, you and I don't want to think about internal links. I don't want to think about keyword research and all these things. Nobody wants to do that.
Yeah, you want to get the result faster, but there's, there's a difference between what can be easily codified, because internal linking can be codified. There are like specific rules that you want to follow. It is easy to explain to AI and it would do the job. But there are things that require creativity and taste and experience, which I am so far kind of not sold on. Let's shift gears just a little bit. I want to talk about personal brand because you spend a lot of time on your personal brand. You're publishing consistently on LinkedIn. You're creating a lot of videos, podcasts, all that. So again, AI, your personal brand, your personal brand machine. So tell me, because when I asked Neil in our interview, how does his personal brand machine work? How many people are working on it? He just said like, I don't know. I don't know how many people work on it. They just send me like, okay, you need to record this many videos. They just chop it up. They do all the things.
He's being facetious when he says that. He knows his number's really low.
Okay. Okay. So Neil was hiding information from me. I should have pushed him more. Anyway, tell me about your personal brand. Tell me about your team and tell me about how AI is changing your content output and your approach of all those things like building in public, creating content, teaching what you know. Yeah, all those things.
Yeah. So we spend— I know just on these numbers are public. We shared this in the YPO event, but Neil spends about $65,000 to $70,000 a month on his personal brand.
That's crazy.
I spend around $25,000 a month on mine. And my team consists of— there's a brand leader. So that's for my personal brand stuff. And then we have a bunch of editors, right? So we have people helping with short form. We have people helping with long form, like YouTube videos. Like this would be a long form YouTube video. And we also have people helping with thumbnails as well. And then we have a social media manager. So that team's probably, call it 7 people or so. They are offshore, most of them.
This is just your personal brand?
Just personal brand, yeah. And then there's an internal marketing team that I help manage for Single Grain and also the personal brand stuff. We also have someone that writes the email newsletter as well. And so for that, here's how AI has helped us. On one end, I don't know if you're starting to notice now with my YouTube channel, YouTube's finally starting to give me credit again. And so I think I was in YouTube sandbox for a little bit because the way YouTube works is it looks at you, Tim. It's like, are you going to be a good employee? Because you're starting this podcast right now. Like, are you going to be a good employee? Like, sure, the Neil one got 4.4K views and all that, but some of these other people have hundreds of thousands of views.
Yeah.
Like, why is that? Because one, it doesn't know if it trusts you yet. 2, it doesn't know if you're going to be consistent. And then also 3, it doesn't know if you're going to be consistent with your topics, right? So for me, 5, 7 years ago, I was talking about marketing all the time and 100,000 views, 200,000 views, right? Then I stopped posting for a bit. Then I came back, I started talking about crypto, right? Then I stopped posting again. Then I came back again. And then now marketing's back, but there's a little AI twist with it, right? And now it's finally starting to give me credit again. Now, all that's to say, let me wrap this up. I used to— we used to not spend enough time on the idea itself. And it is the idea that matters, right? The way you package this talk, for example, or any podcast interview I'll go in, someone would rip, someone would get like 600,000 views, someone would get like no views, but it was one of the best interviews I've ever done. I'm like, why, right? And so now we spend a lot more time, when I workshop with AI, I'm like, hey, I really like, you know, this packaging over here. I wanna make a video like this. And I'll dictate using Super Whisper or something. And I'll just talk for like 5 minutes. I was like, this is the type of video I want to make. Here's some of the thumbnails and packaging that I like. Give me 4 alternatives that you think are likely to go viral. And then it'll give me a bunch of images, it'll give me a bunch of concepts. And then from there, we'll work off of that and create an outline to go after. That used to take us hours and hours to get done. Now, maybe maximum, maybe an hour to get done, but it's just a time savings thing. That's one example.
Any other examples?
A lot more. So we use a product called Overlap. And what Overlap does is— I hate using the word agent because I feel like it gets overused, but it is very much an agent that watches all of my new YouTube videos that come out on Leveling Up or on Marketing School. And it will clip either the short-form or mid-form videos, which are like 5 to 10 minutes or so. I call those mid-forms. And those, I get to choose from that. And then I select those to be posted to LinkedIn, the ones that I actually like. That would usually take people a long time to find, and it's just spoon-feeding those to me now. And I really like Overlap. You can use different products like Opus as well. That's another example. Now, here's a new one that we just started using. You know Descript?
Mm-hmm.
Descript, they made like a cursor version now where you can make edits and you can add B-roll. It's almost like, oh, I want to add B-roll for this over here. And it might use like ChatGPT Images API or something like that. Or I might want to, you know, change, you know, this, this to like a pink color or something like that. And you now have that ability. And so that's— that just came out and that's going to be insane for podcasting.
I know that our own team that works on this podcast, we tried a few times different softwares that can identify kind of salient, interesting bits and make them into clips. And so far we weren't successful with that. So it's, it's a lot of manual work involved. So Michelle, a podcast producer, she obviously needs to listen to the entire interview. She cannot just offload this to AI and trust that AI would, would pick the right shorts. So yeah, so far we cannot— like when we compare what Michelle or I myself suggest we take from an interview as a clip that can potentially generate views on social media versus what AI is suggesting us. And again, I talked to Gail Braden from Authority Hacker in one of the previous interviews. And yeah, what he says is that AI helps him just find like 10 to 15 of those clips, and he will quickly churn through them and he will find the one that he thinks deserves attention. So it's not that AI would be like, here's one clip, this is the best one from the interview. But yeah, so far, judging the output of AI in terms of which clips it suggests versus what we as humans think would be best, it's not there. Those tools that you mentioned, we didn't try. I think after watching this interview, my team would try them out. But yeah, as soon as we would see that there is, I don't know, at least like 70% overlap between like, here are the 3 clips that we think would do good, and here are like 5 clips that AI thinks, and there would be like overlap of at least 2 clips, then I would think that it's usable. So far, I don't think we are finding it useful.
Yeah. And, you know, I know we're talking about the right now with AI and marketing.
Yeah.
But I think that the thing to keep in mind is like, this is the worst that it'll ever be.
Yeah, true.
But the final thing I'll give you is with ChatGPT images. So I can drive a lot more direction now when it comes to the thumbnail and my brand leader knows how I think. But to be able to, for me to like, after I record, I can just quickly dictate to ChatGPT images, here's what I actually want. Here's the image I like. I use the MrBeast tool, view stats, and it shows you thumbnails that do well. And I'll pull just some things that I like and I'll just drag it in there and I'll make a version of it, make an Asian face of me, and then I could just drag it in. That gives a lot better direction to my thumbnails team. And I just haven't had that power before.
Yeah. An important point to kind of belabor is the tools might not yet be there, but you have to keep testing. And this is what we're definitely doing at Ahrefs. Whenever we see that someone's saying, oh, this is like a cool use case, we will test it. Oftentimes it's meh, it's not there. But once it gets there and you're not jumping on it, Yeah, you're definitely going to be left behind. So you have to keep testing and you have to keep finding things that work. Okay. To finish off, I would love to talk about Carat, this software that you launched. First of all, we've already established because it was, it was supposed to be my first question about it. Was it vibe-coded? I imagine it wasn't.
Mm-hmm.
Do you build it in-house or did you, did you hire some agency to, to, to develop it for you?
It's in-house. So my CTO and I sit in the office every day in LA. So we're there at like 7, 7:30 in the morning. He's usually there like 7, 8 PM or so. And we're just cranking on products like Carrot or ClickFlow. So that's where we see the future of our agency going.
What does Carrot do?
Carrot is LinkedIn account-based marketing. It's a—
In human terms, what does Carrot do?
It's a LinkedIn personalization product. So LinkedIn Ads Personalization Tool, just to simplify it.
What does it do?
It's a LinkedIn Ads personalization product. So let me tell you what— let me tell you, let me go into the features and that will probably explain what it does better. So if you're Ahrefs right now, let me tell you, let me ask you real quick.
Yeah.
You guys, let's say you have 100, 200 enterprises that you're targeting, maybe something like that. Yeah, correct. Yeah. Okay. So you're doing account-based marketing in that sense. So account-based marketing is you're targeting multiple people at these companies just for the audience here. If you wanted to target all 100 or 200 of these companies with personalized ads saying, hey, Ahrefs, you know, we made this for you. Hey, Target, we made this for you. Hey, whatever company, we made this for you.
Hey, Single Grain. Hey, NP Digital. If we want to target agencies. Okay.
You would have to make all of these individually and usually it would take you weeks, if not months, to do that. Carrot allows you to do that in minutes. This is a feature that LinkedIn should have. It doesn't have. Right. And so this is a feature that caused us to get a lot of these enterprise pilots initially, and it hooked people in. But we're— that's the initial product right now, but we're having to continue to build more stuff under it. But that's the thing as it sits right now.
I'm not very familiar with performance marketing because Ahrefs has mostly been growing inbound and product-led. So we're only right now, we are going into performance marketing to see how effective it can be for us. So explain me this. How do I target people from Single Grain on LinkedIn to start with? Or like Neil Patel Digital?
Like basic LinkedIn is what you're asking?
Yeah, yeah, yeah. Like if you go to LinkedIn, you need to like, if you have a banner that says, hey, Single Grain folks, we have like something cool for you, or hey, NP Digital folks, we have something cool. How do I know that I'm targeting your people or Neil Patel's agency?
Yeah. So you would simply type in for companies, you would type in NP Digital or you type in Single Grain and then you can target that way. Now, if you're doing account-based marketing, you don't want to just target the entire company. You want to target marketers, right?
Yes.
You want ideally their directors or they have influence in the company, right? So that's— LinkedIn has all the targeting where you can do that. The issue is if you wanted to personalize the ads where it says, Hey, company name.
Yeah.
That is a very laborious process. And that— we talked to an analytics company recently. They're like, dude, we did 50 of these and it took us 4 weeks to do it. And so what we're going— going back to time savings again, Carrot saves you a bunch of time. So instead of weeks, you can launch these ads within minutes. And here's the other thing. If we say, hey, Ahrefs, check this out over here, whatever you can— if you click through, the landing page will be personalized to your company as well. So we have personalized landing pages. personalized ads, and then we have reporting that shows that this stuff is working. And so enterprises are particularly interested in this because their ACVs, their annual contract values, are a lot higher and their sales cycles are a lot longer. So when we talk to enterprises, they're like, we need air coverage, Eric. They use that phrase air coverage, meaning we're not going to mark— we're not going to measure the success of the campaign. We just want coverage.
So what you're saying, again, because I'm not familiar with those advertising platforms, what you can do on LinkedIn is target specific kind of seniority of people in a specific company. So LinkedIn would choose Single Grain, I don't know, VP level, C-level, or like whatever. And it would show ads only for people from Single Grain who have like Single Grain in their account and who have like certain titles within Single Grain.
And here's a real-life example that you can think about. So For example, if you're Salesforce and you want to target all of your customers coming, hey, Ahrefs, are you coming to Dreamforce? You could target them for specific events. That's some use cases people have been doing, or they might have a really unique offer such as, I mean, this is not really unique, but it's like, oh, we'll give you a free Apple Watch or something like that, right? So this level of personalization, people want this and nobody's done it yet. And here's the crazy thing. We launched this and I threw it up on LinkedIn. And immediately a competitor started copying us. The UI looks exactly like ours and they're actually further along than us. They have a lot more personalization features. But my point is, you know, sometimes when you build in public, just expect that there's going to be tradeoffs.
Yeah, this is exactly what happens when you build it public. So yeah, what I wanted to also dig a little bit deeper into Carrot, what exactly does it personalize? Can it— it mentions the company name kind of in text. Or can it automatically find the logo of the company and put it on an image? Or can you give it directions of what kind of— what does it do to the, to the, to the media, to the creative that you attach, or even to the text of the, of the ad?
So the logo is included. The logo pulls— it pulls from LinkedIn's company logos. So those get pulled in and then people can decide if they want to hide it or not. Sometimes it's a legal issue. People just don't want to get involved there. But we have a Canva-like editor in there where you can just drag text around and then edit it and you can choose templates that we already have in there or you can add in whatever templates you have. So it's a very flexible product from an ad standpoint. And then we're going to be adding, you know, conversation ads, video ads and things like that because those convert well for people.
That's very cool. And this actually reminds me of the product from Jong Yong Fook, who I just mentioned, because he has this product Banner Bear. That we've been using at Ahrefs for a while when we understood that we spend too much time creating those OG images for every article.
Yeah.
And his product basically automates it. It has like a certain set of rules. So you take, I know, the title, you want to have the picture of the author.
Yeah.
You have to have like specific icon if the post belongs to a category, you can change color. So you get, you have a set of rules and when you publish an article on WordPress, it automatically creates this image. By the way, I need to check with him to see what kind of, I don't know, new AI capability, given how good generative AI is getting.
Yeah.
Because, because we stopped doing those custom illustrations, it was too time-consuming and we just stopped seeing value in that. So yeah, at some point people developed blindness for those things. But I wonder like where Jon is going to take his product with all the AI visual generation capabilities.
I'll tell you, you want to know where we're planning to take this? Yeah, yeah, yeah. So with Carrot, what I want to see happen is going back to one-click marketing again. I want there to be the ability where we can look at your data and create creative that's likely to convert and maybe new ideas just by looking at the internet or other high-performing B2B libraries. We're obviously going to have a lot of data on what converts as well. So we might just say, hey, this based on the data we have, this is more likely to convert. And we'll look at your brand guidelines and everything. And basically one click, maybe you create 100 ads in these 3 different categories. That's not it. We're also going to have different offers because B2B companies suck at offer creation. That's like when we talk, when we work with a lot of these enterprises doing these pilots right now, their offers just suck. And I'm just like, what are you guys doing?
Right?
And so if we can help them create offers, that's a much bigger lift. I don't know. That's that's not going to be an easy thing. Help them with creative, help them with offers, help them with landing pages. That to me is something that's valuable.
That, that sounds very promising, to be honest with you.
We'll see if it works.
We'll see if it works.
I have one more thing before we go. I just think in my mind, whether you're optimistic or pessimistic, but I think we actually do agree net-net that this is—
We agree on a lot of stuff. Yeah.
Yeah. And I just, I think those of you that are very curious are going to be fine in the long term. I think those that want to just say, that want to just chill, you'll probably be fine long term too, but there's going to be a little bit of pain in the next 5, 10, 15 years. But I think just, you know, someone like you or someone like me that loves learning and loves getting better, you're going to be 100,000x or so. And, you know, this conversation is just the very beginning though.
Eric, thanks a lot for coming to talk to me about this. Thanks a lot for putting up with all my curveballs and hard questions. I hope you enjoyed this conversation.
I did very much so. Thank you so much for having me.
