How To Automate Content Marketing Research featuring Mick Essex

Michael Levitz (00:00)
welcome to the Forecasting Podcast. It's been a minute since we've been back. Robin and I went into a deep coding and product hole. We have now come up for air and we've brought with us a good friend, Mick Essex. Mick, do want to say hi?

Mick (00:13)
Hey guys, glad to be on the podcast and glad you guys came up for air and I was the first one that you got to talk to. It's awesome.

Michael Levitz (00:18)
We're glad to have you. ⁓ even just in the kind of warm up here, we've learned a ton. So we're psyched to jump into this. Robin, do you want to say hi?

Robin Tully (00:27)
Yeah, hello, excited to be back and excited to talk to Mac.

Michael Levitz (00:30)
All right, so Mick, before we dive in, do you want to just give everybody a quick background run, your background and what you're up to right now?

Mick (00:37)
Yeah, for sure. So currently I'm the head of growth marketing at PowerIO. We're a cloud-based no-code plugin builder. We serve our products on every site builder you can think of. Most recently, Wix and Shopify are our two main platforms, but we've been an old school WordPress user for many, many years. And we like to work with small businesses all across the world to help them grow their businesses online. That's what our mission is, and that's what I try to do is...

head of marketing is to get that message out to as many small businesses as I can. Most recently, I've been in marketing and outside sales for 20 plus years at this point. I like to joke that I've been around before Facebook. The way I like to put it is I'm an old school marketer using new school tech is kind of how I frame it. It's worked pretty well so far.

Michael Levitz (01:24)
And head of marketing, you know, can kind of bend in different directions into different skill sets. You have an amazing and super interesting kind of take on that role. Can you talk a little bit about, you know, your toolkit that you bring?

Mick (01:35)
Yeah, Head of Marketing is such a fluid title and list of responsibilities. I've functioned in so many different roles. mean, from the traditional standpoint, you know, we have all of the common levers, social media, email, organic, ⁓ SEO, all of those different pieces. What's interesting about Power is that we're very product focused and product centered. So it was incumbent on me to learn as much about coding as possible.

I always have the disclaimer I am in no way a programmer, coder, engineer, whatever you want to call it. I'm not that guy, but I know enough to be able to read it and to understand what is happening. Looking at HTML and looking at JS, I know what I'm looking at now. And what that helps me do is it helps me understand the back end of what people are seeing on the front end. And that helps me tell a better story, I think. So I don't know a lot of marketers that have a lot of product knowledge. So I feel like that

makes me somewhat unique in my role. But other than that, ⁓ we don't have a sales team. So marketing has to carry the entire load, which presents us with its own challenge. Right? We're completely product led. And that means we have to rely on the UX and what's on the page to do all the heavy lifting. So it's made me a better marketer, honestly. We don't spend a lot of money on advertising. So I have to think more intently and be more deliberate.

But what it also does is it allows me to take chances and take risks and make moves that otherwise might not be able to happen. that's just, I guess, the basic rundown of how we run things. We're obviously getting into AI quite a lot. Several different things that we've automated recently, a lot of GPTs that we've made and we're exploring in 8N now as well to try and automate a lot because we're a small team and we don't have a huge budget. So we have to be creative and make AI work for us, if you will.

Michael Levitz (03:22)
That's a great point. We'll get into the N8M stuff. We'll get into the AI stuff. I think team size is super important because a lot of teams have shrunk. A lot of new teams have started and they haven't kind of scaled the way an old marketing team would have. How big is your team and can you talk a little bit about what are the roles?

Mick (03:39)
We're a whole team of two. whenever I first, including me, so whenever I first started, was, I actually, did not realize this and they did not divulge this before I said yes to the job. But I replaced two people. So we had like a traditional marketer like myself and then we had an actual product marketer as well. And I essentially replaced both of them whenever I started. So there was...

Michael Levitz (03:42)
Including you or not including you?

love it

Mick (04:03)
One young lady in our customer support team that had gone to school for marketing in Europe, but had never actually gotten the opportunity to kind of flex those muscles. So she started on with me part-time 20 hours a week, the first that was in 2023. And then last year, she moved over to marketing full-time. We actually hired a second marketer last year, but unfortunately for her, was about the same time that AI really kind of took off. And we really

We really didn't have the headcount budget to add a third person, but there were so many different things that we were doing that ⁓ working seven days a week for me just wasn't, I couldn't, I wasn't going to keep doing that, right? Like that's not healthy. So we brought on the third person and then we quickly realized that a lot of the tasks that she was performing, we could use AI for, so it's back down to just myself and the other young lady. So it's just two of us. We run the whole ship, just the two of us.

Michael Levitz (04:58)
That's, mean, I think it's more and more common. And I think, you know, as I talk to people, you know, what they're really interested in is, hey, we, we don't have time for any fluff. We don't have time for kind of, you know, bad decisions or just wasted vanity stuff. You know, how do we make the most of one or two people? So I think, you know, one of things we'd love to get into today is how do you, you know, do all the amazing stuff you guys do with just two people plus, you know,

Mick (05:01)
Mm-hmm.

Mm-hmm.

Michael Levitz (05:26)
automation and AI.

Mick (05:27)
Yeah, we obviously had to take a hard look at all of the different bigger buckets, kind of one at a time, and really see what is the output, what is the outcome of, and let's just put them into huge buckets. So let's look at SEO, social media, email marketing, and then product marketing. Are there any of those that we can eliminate? Are there any that we don't have to have? And I made the decision at the beginning of this year that we were cutting out social media.

And what I mean by that is from our business pages and from advertising. It takes a lot of time. Social media just in and of itself, right? It's obvious, especially on platforms like LinkedIn, where it's clear that the algorithm is prioritizing posting from people and not businesses. Like that's made obvious. That's why you quote unquote see CEOs and founders posting all the time. No, they're not. There's somebody that's posting that for them, right? Like I think we all know that. So

That along with ads, we really looked at how much time is this taking and what are the results that we're getting from it. And it wasn't that we weren't getting anything, but we saw a better opportunity with the likes of email marketing and SEO and obviously, AEO. Our time would be better spent focusing on those pieces than we would social media. So we dropped social media as a bucket altogether. I still post, we still post as individuals.

And then we funneled that time into equally across the other three. So with us being product led, with us being built on a freemium model, we rely very heavily on organic traffic and organic growth. We have almost three million web pages that we have in the world. So that obviously takes a lot of time and we can't look at those pages one at a time. So how can we automate that? How can we scale that? How can we take our time and put it

there to make that better. So most of the time we were spending on social media, we diverted that into SEO and organic growth. Email marketing was kind of its own thing in the big picture. We've been using HubSpot for a very long time. We have over 200 workflow automations that are going. We redid every single email and workflow that we had. That was a Q1 OKR is to completely revamp our email marketing. Since then, we've doubled open rates, we've tripled click rates.

And we've actually been able to eliminate some of those workflows and roll them into one. So there's not as much to check back up on. So big picture, we consolidated the places we were spending time. We focused more intently on the things that we knew were moving the needle. And then the final piece was, and this was from the founder all the way down, for every task that we do, can AI do that? And that was something that we had to put together for every single task.

We're doing this, can AI do that? And what we found is there obviously is quite a lot, right? One of the things that was taking a lot of time from my partner was ⁓ blog editing. Just taking the time, we have built an army of guest authors around the world, but we have to edit every one of those articles and make sure that they conform to our content guidelines to make sure that it is properly set up for SEO and now AEO. So... ⁓

built a couple of GPTs that will automatically scan those drafts, will send the author back a list of things that they need fixing, and then they send the draft back. Instead of her having to do that, AI is doing that. Instead of doing three article edits a day, we're doing 30 article edits a day. Just small pieces like that. What are the different things we could do to put AI on the job and just make things quicker?

and get everything out faster. We're in ship it mode right now.

Michael Levitz (08:55)
I love that. a couple things. So just going back chronologically for a minute. So you're still doing social as a strategy, but not from business accounts, just from individual accounts.

Mick (09:05)
Right. Yeah, basically. Yeah.

Michael Levitz (09:07)
that makes sense. And yes, we have definitely seen the same thing. I think that's just a great example of making difficult decisions that go against common sense. Like what? We're turning off social. But it sounds like you ran the numbers. You saw the amount of time it takes to kind of weigh the algorithm is throttling those posts. They don't want those business page posts. And why just keep doing something just because it feels like historically you were supposed to.

Mick (09:18)
Ha

Mm-hmm.

I should say that I'm also a big believer in turning things off to see what the results are. You build these things and you put them in place and you have them running for such a long time, but what happens if you turn it off? And I think there's, it's kind of a couple different things that can happen there. First of all, you're terrified that things are going to fall off the cliff, right? Well, things, if you set them up correctly from the beginning, turning something off and back on again is literally a toggle.

So turn it off, watch the numbers. Did things plummet? Did you see a monster drop off? If you have your data and analytics set up and you're able to check those things and you have your different projects set up correctly, turn it off and see what happens. And the shocking thing, and I think the other part of that is like, what if we turn it off and nothing bad happens? We've been doing this all this time for what?

There's also that because you lobby for those projects two or three years ago, and it's almost like you're afraid to turn them off and prove that you didn't need that anyway. We're doing that with lot of our email workflows right now. We just turned off a review workflow. We rely very heavily on good reviews on different app marketplaces. We turned it off a month ago, and we're still getting just as many reviews as we were without it.

Then it's like, okay, then everybody gets kind of excited. Like, well, what else can we turn off? know, because one more thing turned off is one less thing to track, right? Like that's one less line on the graph you have to pay attention to. So now people are actually getting excited to turn things off, crossing our fingers that it doesn't go badly. But I think, you know, it can kind of go both ways because there are those times where it does have a huge effect.

I cost us a lot of money in July. I'm a big believer in admitting failure. I think that we learn a lot and grow from failure. I had the idea of targeting a specific keyword for one of our products and it worked incredibly. Revenue, acquisition, everything went through the roof. So obviously whenever that happens, how can we replicate that on something else, right? So we tried the same approach with a different product and within three days,

It was like a cliff dive. And in just three or four days, it was a few thousand bucks I cost us in revenue. But I learned, know, what works for one thing doesn't work for everything. But if I never tried that, I never would have known that the first thing was going to work. So I'm fortunate to have a CEO and to have founders that encourage us to move swiftly, be curious. And if you break something, just fix it. You know, it's OK to break things.

as long as you fix it and you learn from it and don't do it again.

Michael Levitz (12:07)
That's a great point. Do you know the case study with Uber and their app download advertising? I'll shoot it to you and I'll put in the notes here. They had turned off something like a hundred million dollar ad budget globally of app download ad placements with a bunch of different ad networks and nothing changed. Their app downloads did not go down.

Mick (12:14)
No.

Michael Levitz (12:31)
And that just launched this whole, I think the guy who actually ran that program like became like a click fraud, uh, kind of expert and open, opened a business just around that. So yeah, I mean, they, they just said, just turn stuff off and see, you know, you think you're spending all this money responsibly and it may not be doing anything.

Mick (12:42)
Wow.

Yeah, exactly. just, don't know if you don't, if you don't try. it's always worth, exploration is how things are discovered, right? That's the beginning of discovery is exploration. So yeah, I encourage anyone listening to explore, do things, turn it off, turn it on, try something wild. You just never know whenever that one right thing is gonna really click and you just uncovered a whole new revenue stream possible.

Michael Levitz (13:15)
That's totally true. Robin, feel like, yeah. And Robin, I feel like you're constantly experimenting with this. You'll come in on like a Monday morning and just be like, hey, we're swapping this service out for this service. What do you think?

Mick (13:17)
or save a ton of money.

Robin Tully (13:28)
Yeah,

I mean, the sunk cost fallacy is a strong thing and we should kind of actively push back against that. And then, know, as Nick was talking about earlier, like when you have to juggle so many different work streams and be the both kind of literally and professionally be the man of many hats, you know, you have to figure out where is the best investment of time gets back to that multi armed bandit allocation of resources we've talked about before. Yeah, I mean, it's always like,

the kind of scientific experiment and experimentation of all of that is always valuable.

Mick (14:00)
and

Michael Levitz (14:01)
Yeah, the funny thing with Robin is he's like, ruthlessly decisive, at least from a, you know, from an outside perspective, things that I would like dawdle on. Robin just like, screw this. It's taking too much time. I'm just going to like dive in, fix this whole thing and swap it out. And then you come back the next day and you're like, what? How'd you do?

Mick (14:17)
Aya.

I fully

support that. I often take an approach that most marketers disagree with. I think A-B testing is a waste of time, 90 % of the time. I'm not saying don't A-B test. Let me be clear. But A-B testing, everything is insane to me. Make a choice and go with it. You'll find out real fast if it was a good choice or a bad choice. But what happens is you learn more

after execution than before it, right? Like you can over plan. I forget the old mantra in we're just working in SAS in general. If you're not terrified to launch, you launch too late. I think that goes with anything, frankly. Don't waste so much time A-B testing everything. Waste two months and then you're ready to go. Well, now you're two months behind your competitors, in my opinion.

I'm not saying never A-B test, but I think most A-B testing is a waste of time.

Michael Levitz (15:14)
Yeah, I think generally like it's the, what do they call it? Like the Japanese, like a shipping container or whatever. It's like, we'll suggest AB testing just to, Japanese inspection. You know, we'll just suggest AB testing. So this guy gets bored of his own idea and like forgets about it. You know, we won't really have to do it because we'll just wait for more data.

Robin Tully (15:33)
But I think that's interesting too with what you were saying earlier about the experimentation that you had done about some keyword worked very well and then take a big leap and then kind of fail hard.

I think one of the, in data science, there's this, lot of models will have that thing called a learning rate. And it's just a number of how fast can the model adjust its weights. So this like, I think with a lot of people, the AB testing, yeah, they're just going from, know, A to B back to, there's no actual exploration there. So leaping around, you will kind of find the actual kind of optimal path faster than just the kind of micro fluctuations that people do.

Mick (16:10)
I think I came to this realization whenever I had run like my sixth or seventh A-B test in a row and I followed all the protocols. made sure that I had my sample size was big enough and I made sure that I let it run for long enough to get to statistical relevance. I did like six or seven in a row and the result every time was it really doesn't matter between your A or B variant. And I'm like, are you kidding me? Like I've done all. So after that happened multiple times, I'm like, you know what? If it doesn't matter.

at the end of the day, in most cases, that I probably shouldn't bother A-B testing in most cases. So that doesn't affect or apply to everything, obviously, but now I go through this more intentional way of thinking of does this need to be A-B tested or not? And nine times out of 10, I decide that it doesn't, let's just do it and see what happens. So I'm quick to pull the trigger. I have no fear.

Michael Levitz (17:03)
We're in the Shippet era. So question, so you mentioned you have 3 million pages on the site, is that right? And help us like kind of visualize what portion of that is like content, what portion of that is, you know, product stuff? How do you divide it up?

Mick (17:04)
Yeah.

Those are all external pages. I forget exactly how I should say this. It's basically programmatic SEO is what we did. Whenever you take the number of, and this was math done before I started, but we have over 60 different products altogether, anywhere from a form builder, which is our number one, all the way down to a brand new tool. It's an email app, actually. I don't even think I've told you about Michael.

Michael Levitz (17:40)
Hmm.

Mick (17:41)
Basically, it's an email marketing platform within Power that you can do all of your emailing from inside our editor. You don't need MailChimp, don't need Klaviyo, you don't need any of those things. We're still in beta, so we're still testing, but ⁓ that's just something that's very new. Anyway, for all of our different products, we also integrate with basically every site builder you can think of. I think we're over 1,000 now. I don't even know there was 1,000 site builder platforms, but...

Whenever we started doing that math, we've got 60 products across a thousand platforms, start doing the math. Then we programmatically set it up to where all of our different web pages would auto publish in other languages. I think it's 12 languages altogether. So now you multiply that again by 12. So once you start doing that, the pot gets really, really, really, really big. The pages that make up the majority, I would say 80 % of our page profile.

are tutorial pages, how to install a form builder on a WordPress site. And programmatically, we have the app type and we have the platform that are interchangeable with the variable. So you can see how to install a form builder on a Shopify website, how to install a pop-up on a Squarespace website. So the majority of our pages, all of those pages are all external pages that are served on Google. So none of those are

internal pages or within our editor those are all external external pages.

Michael Levitz (19:02)
That's amazing. And what is the distribution of, if you can tell, know, customers, people on the platform versus people that become leads to those pages?

Mick (19:11)
We don't track it that granularly to the degree I think you're asking, and mainly because we are a freemium product. To this day, 96 % of our users will never pay us a dollar. But every one of those have value because every one of our apps that are used for free has our watermark at the bottom of it that will then just exponentially grow even more. We average over thousand new users a day.

and we spend zero dollars on advertising. Everything is completely organic. So to that degree, we don't truly track leads. The main metric that we track is our free to paid conversion rate. And that's going to be have like an overall number altogether. So in general, for every 100 users, depending on the platform, it could be anywhere from two to nine of those users will turn into a

paying customer. The other good thing about the way we're set up is our lifetime value is very long. It's almost seven years on average is how long we keep a customer. So one customer, although our MRR might be kind of low in the big scheme of things, most importantly, our net recurring revenue is actually quite high. So we don't have to spend a lot to get a lot. Our CAC is 21 cents.

So it's a very low risk is a, something's like a dangerous way to put that. But because of that, we are able to take more chances and we don't have to track numbers directly to an MQL, to an SQL. You we don't have to bother with all of that.

Michael Levitz (20:38)
That's incredible. all right, so you've got these 3 million pages. And we had talked previously about Refresh, which I think is also really interesting. But how many new pages, what volume of new content across the things you've prioritized? I think it was, was it email and page publishing or those are two main channels?

Mick (20:57)
Pretty much,

yeah. Those are the two main buckets. The cool thing about our page content is they don't have to change a whole lot, the existing pages, where that might come in. This is kind of a really neat way that I think we've built this engine. I'll just kind of walk you through it a little bit at a time. Let's say that you host a site on Wix and you want to add a social media feed on your website. You want...

everything that you post on Instagram or TikTok or whatever, you want that to show up dynamically on your website. So you find a social feed app, you find ours, you download it, you go through the Wix editor and you make it look like you want and all those little things. And then whenever you click publish on our publish screen, which I mean, truthfully, all you have to do is click publish and because it's Wix, it automatically happens. But because of how we were set up years ago,

It actually shows the publish or the install instructions on that publish screen. So anytime Wix makes a change to their installation instructions, we make a change internally on our page. So obviously everything will match one to one, right? So the way these tutorial pages are set up, the content on that web page for how to install X app on X platform, it's being dynamically pulled from the publish instructions.

So whenever we update our publish instructions, it automatically updates it on the page. So that essentially functions as an automatic content refreshment right there. Whenever we add a new platform, like just recently, I've come into mind, we just added Ghost as a platform that we integrate with. And we come across these all the time, and usually for me, because I'm a perennial researcher, and I found Ghost, and I thought it was a really cool platform, and it looked really...

Michael Levitz (22:36)
It is.

Mick (22:37)
very functional. I actually have my own site now on Ghost. But the first natural question is, can we integrate with that? So paying the engineering channel in Slack, like, I found this new platform. Can we integrate with that? And then they do their engineering magic, whatever you guys do. And they figure out that, yes, we can integrate with that because our proprietary software, maybe I should have started with this, is our internal software will automatically write the HTML code for you.

So basically, I want this app and I want it on this platform and PowerWrites the code for you. So once we realized we could integrate with Ghost, then we write new publish instructions for Ghost as your host platform. Well, as soon as the engineering team creates that functionality on the backend, all of those other pages are automatically created. So now we have a new page, how to integrate or how to install a form on your Ghost website.

And because everything works programmatically, all of those pages are created in all of those languages, in all of those places. fortunately, I don't have to create a lot of content for our web pages directly. The majority of time we spend on content production is on our blog, different articles that we write.

And then I've started to work in Lovable recently to make my own landing pages because I don't want to wait for engineering and design. So I've started to create my own pages in Lovable that I just want to see if I can get some SEO juice out of them. So that's been a new development recently as well. I've learned what GitHub was for the first time recently. So that's where most of our content marketing time is spent is on articles that we publish that are

you relevant to whatever is hot this week.

Michael Levitz (24:19)
So that's interesting. So what's an example of a recent one, a recent article?

Mick (24:23)
Recent article that I did, the name of the article escapes me. And I think you and I may have talked about this. I read that white paper from that one professor that was talking about how AI actually does what it does and how an LLM works. I mean, the root of it is it's just basically giving you the best option for what the next word should be in that particular response, right? So we kind of broke down.

how AI works and how a large language model functions. So I took that 80 something page article and went into chat GPT and I'm like, hey, I want to write an article that a normal person can understand. So that was the one that I did most recently was I think it's called how AI and LLMs work is basically the title of it. So that was one that I did most recently. I read recently about

I love to do keyword research and look for keywords that have low difficulty and see how I can possibly spin that. One that's worked really well is shadow banning on Instagram. The number of times that that's surged are in the millions. For why did I get shadow banned? I don't believe shadow banning is a thing, honestly. I think it's just a fun buzzword that an influencer made up. And people just like to complain and bitch about that's why they're... Your posts are not doing well because you suck at posting.

get down to it. But people search for that a lot. Well, we have an Instagram feed product, right? Like the social media feed I was mentioning. We have one that's called Instagram feed. Well, if I write an article about how to not get shadow banned on Instagram, I'm getting a few thousand clicks a day on that article. And guess what? I'm plugging my Instagram feed product in that article. So that was one that worked really well.

I actually started ranking for a new keyword on that one just last week and I wrote that article six months ago. So that's been the fun of me that I honestly don't want to use AI to replace because I love keyword research and analysis. I think it's really fun to look through that data. We use Ahrefs. And it's just really fun to find those little nuanced keywords that

may have a keyword difficulty of three, but 7,000 people a day are looking for something related to that term. How can I take that, turn that into an article, first of all, that's useful and actually gives good information. But then, obviously, the marketer is the salesperson and me is how can I leverage that to promote a tool that we have that fits within the content of that article? And that's just really fun. I know I can make an AI GPT to do that, but I just really like to do it.

Michael Levitz (26:50)
So we also use Ahrefs, love Ahrefs. And I think one of the things that we're focused on is how do we take those keywords and make them relevant to like a brand point of view? I call them like disembodied keywords. You know, you're looking for like, you want to win something on let's say, you know, how do I automate competitive intelligence? That's something we look at a lot. And you just get these kind of like super generic

automate competitive intelligence and then the word, the words will be kind of like reshuffled or slightly tweaked competitive landscape. You know, how do you go from those keywords that are kind of empty to something that has, you know, a harsh point of, not a harsh, but like a, an edgy point of view that people actually want to read.

Mick (27:31)
That is the challenge, and I think that's where we're looking at, not at potential, that I think will be a problem in this new age of content development, is how many things are just going to be the basic, same regurgitated stuff in a different way and on a different domain. That's, to me, what I find to be most important. And I actually just posted on LinkedIn about this recently, and ⁓ that is giving your AI a voice.

is what the actual LinkedIn post is about. I have that, of course you have to have the paid version, but I have my tone of voice and my idiosyncrasies baked into my chat CPT. So whenever I take this guy's article that was very difficult to read, and if you don't have two or three college degrees, it's not gonna make sense to you. So the first thing that I do is I, as I said, I ask it to convert all of this high technical language

into something that has, and I'm sure you're familiar with the readability index, that can be read by someone in middle school. And that's where I'll start. Then in Chai GPT, I'll give it a prompt to say, imagine that I just read this and I have to give a lecture to a senior class of graduates on how to use AI in their real life. How would I explain this article in my tone of voice and from my perspective?

and then it gives me the equivalent of a speech at that point. Then I'll take that speech content, and I actually used Notebook LM to make that into a podcast episode, believe it or not. I didn't publish it anywhere. I just wanted to see if it would work, and it does. Then I'll ask it to take that speech and convert that into an article. Now I've turned this super high-profile, very rich-languaged article, I made it into the content that an eighth grader could understand, then I converted that into

what would be essentially me giving a speech to a group of people back into a blog article again. So I've basically I've rinsed that article four times at that point and understand this all happens in about 30 minutes. Like this isn't several days. And then that gives me a completely rewritten article that is understandable to the general public and in my tone of voice.

Michael Levitz (29:40)
So not a lot of time left, so much to get through. So tell us what you're doing with N8n and then we'll wrap up with some AEO stuff.

Mick (29:47)
Yeah, yeah. N8n is a new beast that we've just recently started to... This is one of those things we were going to tiptoe into it, like how can this apply? And it blew everyone away so quickly that we spend so much time in it now because it has so much functionality. So the place where I started, the first one that I built was a Reddit intelligence bot. We were paying someone $750 a month.

to basically scrape Reddit for us, like physically go through Reddit subreddits that were interesting to us and find posts that made sense. Now, there's a lot of Reddit scrapers that are out there, but from what I could tell and what I had seen, it only gave things on a very high level. So, the quick and down and dirty of that. basically, I set up notes within N8n that will scrape Reddit based on the keywords and the subreddits that I want.

It throws that into a mixer and it looks for posts that combine one or the two in multiple variants. Then it goes through a washer that basically filters out everything based on an engagement score. It's an algorithm that I gave it that will score. It'll give me a hundred posts and of those hundred, once it runs through the renter, there are 29 posts that look like they're probably relevant to what I want to comment on. Then it goes through a bunch of other nodes that I won't bother you with.

But the end result is every 12 hours, I get a Slack notification of the top 20 posts over the last 12 hours that I should comment on that's relevant to Power or relevant to no code software or whatever. Where I think mine is different is that in Slack, first of all, it'll give me a link directly to that post. It'll tell me the basic gist of what that post is about, why I should comment on it, how confident it is that I should comment on it.

And then it gives me the upvotes and the likes and the comments. And then I just basically go through that essentially twice a day. And I just scan through those really quickly. The first time that I ran this, I commented, think, out of the 29 that I got to take a comment on, the five of them. And then it sends me an email as well based on the current sentiment within that subreddit, what I should post.

In this example, it suggested that I post about the high cost of email marketing software in the email marketing subreddit. I posted that I got 181,000 views on that post in 24 hours with 67 comments altogether. And we've already gotten three customers out of that one post. So that that in a way that I call it the Reddit intelligence bot.

It does all of that. It took me about a week to build and I failed several executions before I finally got it right. But now it works flawlessly. I don't know if you heard my Slack notification, but that's my afternoon Reddit digest just now of what it is that I need to comment on and it's already bearing fruit. So that's where I started with that. We're working on building a HubSpot replacement for email marketing. We're working on

repurposing instead of paying for Zendesk, we're going to build our own AI agent to answer support questions for us. It's actually building a knowledge base for us. If it searches our knowledge base and doesn't find an answer, it will suggest an article that we need to write for our knowledge base. Because this information that this person was looking for, they could not find it in the knowledge base. Here's an article you need to write, it's already keyword optimized. And then I run it through my GPTs to produce the knowledge base article.

All those things are functioning that would have taken two or three people a week to do. We're doing it in an afternoon with a couple of AI agents through an ADIN.

Michael Levitz (33:26)
That's incredible. Yeah, we need to do a deep dive or you should just do like a webinar where people can just literally join in and I will join and you should just kind of walk through screen share what you're doing.

Mick (33:29)
Hahaha.

If you want, as a takeaway for anyone listening, I can send you over the JSON files of those N8M workflows. You upload that JSON file and it builds the workflow for you immediately. And you basically just swap out the different APIs and keys for your business and you can have that running on your own platform in just a couple hours.

Michael Levitz (33:57)
Yes, please, that would be incredible. And thanks for open sourcing of those weeks of hard work.

Mick (34:03)
Yeah, I think it's super cool and I want people to know what I did, not knowing a single thing about JSON, by the way. I just want to make that clear. I have no idea what I'm looking at or what it does. But yeah, I mean, have no engineering background and I'm building full on AI agents in a few days.

Michael Levitz (34:19)
and your own landing pages and web pages and level. That's amazing. All right. So AEO, so we started talking AEO. You know, I feel like the type of content you're creating is kind of so perfect for AEO because you are actually, you know, just directly answering and educating people on very specific tasks. So you're, you're kind of like uniquely positioned to kind of win this war straight out of the gate.

Mick (34:21)
Yeah.

Ha ha.

Michael Levitz (34:42)
Can you talk just about you have an incredible like workflow on this? Can you talk about how you're approaching it?

Mick (34:47)
Yeah, I was very excited whenever this first came out. I actually talked with an agency in New York that they're a content marketing agency that we're kind of shifting their entire focus to answer engine optimization. And I hope that one day we can decide on what term is, by the way. Like I've seen so many. AEO is mine. I'm sticking with that. But basically, it's just like what you said. We all are familiar with what Google Search Essentials say, and we know what...

having actual content that checks the E, the E, A, and the T, but how do you show up in a recommendation inside an AI citation, an answer in ChatGPT, or an AI overview? There has to be some structure that's there. Trying to uncover exactly what that structure is, I still feel like is a moving target. It is getting much better, and there are some clear things that are coming up to the forefront. For example, making sure that all of your header structure is proper.

a good use of bullet points, having a summary section, having a clear and defined FAQ within that particular piece of content. All of those things are what AI likes to consume because it gives a very direct answer to a very pointed question. Well, if an AI tool, if most people are using AI tools to answer questions and not build things, then it makes sense to write

articles and web pages to answer those particular questions. So the way I approached it is that, and this is both twofold, feel like. So what's worked for us, and we have the data to show, we've 20x'd our AI citations in the past eight, nine months or so, is there's both on-page and off-page considerations. obviously for me, it was easiest to build an AEO optimizer GPT.

that I run things through, it looks at the page source. So it's not only looking at the content on the page, but it's also looking at how the article or the web page is actually structured. Like everything's about structure, right? And that starts with the framework of the page. So the first pass is to make sure that the content on the page is useful. actually has an expert point of view to it. And it kind of presents out something that can outwardly be consumed as good content.

The backside of that is proper structuring to make sure that everything on the page is structured correctly up to and including software application schema, FAQ page schema, product schema. So that's how we basically approach it is we take the page source from that entire page, we put that into the GPT, it scans all of those things and then it gives me a spit out of what's missing, what I did well, which I think is important, but also what's missing. And then we just adjust the page from there.

and it's worked really, really well so far.

Michael Levitz (37:29)
That's fantastic. And I know there's more to dig into, question, are you writing pages specifically for AEO or are you primarily focused on taking pages that are already existing or that new content you're creating and optimizing it for AEO?

Mick (37:44)
We're still going through repurposing content that we already have. It is kind of two-fold though because we do have an entire process for new content intake. So all of that new content that we get in from authors externally, we automatically run it through that whole automation. So any new content that we post will automatically be optimized for AEO. So what we're doing on the flip side of that is fixing a lot of the pages that we have out there.

The good thing is from our website's perspective, because we have so much stuff that's automated already, we can make changes to a certain section of code and we can impact a million pages at one time. So that was the one thing we just did very recently is we added FAQ page schema to all of our product landing pages. We had an FAQ on the page, it was SEO optimized, but we didn't have that structured schema there. And of course, then it's just a sit and wait game to see if it actually makes a difference.

But we are seeing, I get the report just like anyone else does, we are seeing us rank for keywords we've never ranked for before. We're still not ranking number one for a search term with 130,000 queries, but incrementally, we're improving our total keywords that we're ranking for, and we're incrementally coming up in top three, and we're seeing a lot more AI overviews as well. So we know it's working.

It's just we have such a huge volume to go through and there's so many different keyword comparisons. Will it work a year from now? I have no idea. But for right now, a lot of these posts you see on LinkedIn where people are posting their search console with all the numbers. You can look at mine. I don't see it. There are blips, but nothing, especially with the amount of page volume that we have. I'm not seeing.

Impression losses, I'm not seeing click losses. It's there. It's some, you know I think that we are losing some where people are getting answers without clicking like I'm not saying that we've avoided that But I'm not seeing these monster drop-offs at all. Like it's just not happening for us

Michael Levitz (39:47)
Yeah, and I would say that that's because you are serving that kind of like low funnel substantive content and the people that getting hit are the kind of a higher funnel, you know, kind of general content that's just supposed to get you in. And that's the piece that, you know, is just getting answered automatically. Do you agree with that?

Mick (40:06)
I would, I would. luckily for me, I'm not in a position where my software is $5,000 a month. And I have, you know what I mean? Like most of our users are free, right? So it doesn't affect us a lot in the same ways that others, that it might with others. We have a very low average order value. We have a really healthy lifetime value. So we can offset that with not having the best free to pay conversion rate. We can offset that because we work in such high volume.

At the end of the day, turns out to be, it put us in a really lucrative position.

Michael Levitz (40:36)
Yeah, I mean, for me, like the two kind of primary rules of thumb are one, answer questions, you know, very deeply, and two, choose questions that you have a core authority within. And, you know, you're just kind of in the sweet spot of that. You're not, you're not kind of, you know, just kind of taking a flyer on some random search term to try to, you know, bring clicks into your site.

Mick (40:48)
Mm-hmm.

Right, yeah, mean everything that we post, anything that we put out, we do have a point of view on it. I'm not just looking for clickbaity stuff. It only works best if it is somewhat relevant and I can give a point of view on it. So I think that's an important piece as well. And we have a big team from all over the world. So I have a lot of different points of view to pull from, which makes it nice.

Michael Levitz (41:22)
That's fantastic. All right, we are 10 minutes over. I apologize, but this has been incredible. Thank you so much. There's, I think, awesome stuff. Definitely want to put the JSON, the N8N stuff in the notes. A couple examples of your articles, I think would be great. Like a couple different flavors. You know, like a...

Mick (41:27)
no.

Michael Levitz (41:40)
a deep how-to content article as well as the one you mentioned around, sure, I'm forgetting the exact topic, but the AI pipeline that you were talking about.

Mick (41:48)
I'll tell you what, also give you one, an article that I used AI to write for me that came from a screenshot of a LinkedIn post. Yeah, I didn't think that would work, but it absolutely does.

Michael Levitz (41:55)
That would be cool.

Love to see that. And yeah, I think just the kind of spirit of kind of how you're doing this with a two person team, how you're making incredibly, you know, kind of tough decisions, what to do, what not to do. And I think turning stuff off is like the, big takeaway from this. Turn it off, don't be afraid and then see where the, you know, see where this stuff's really adding value.

Mick (42:18)
Yep, think if nothing else, that's the big takeaway I think is a good one is turn it off, see what happens. You'd be surprised how much time you can save on the other side of that.

Michael Levitz (42:29)
Alright, so where can people get in touch with you or follow you?

Mick (42:32)
I'm on all the socials, but LinkedIn is where I spend most of my time. It's Mick Essex. It's just like it, well, don't know if it's, everybody spells both of my names wrong, so I don't know how helpful that'll be, but that's where I'm on LinkedIn. All the other socials are Mick underscore in LA.

Michael Levitz (42:45)
hat extraordinaire. We didn't even get to that as a bald person. I have lot of respect for your hat collection. And, you know, just all around amazing guy. You've been incredibly generous with us. So nice. And, you know, just we love having you kind of in our network.

Mick (43:00)
Yeah, it's a pleasure. I'm happy to talk anytime.

Michael Levitz (43:03)
Please set up a webinar that we will join walking through your N8N setup. That's, think, going to be high demand.

Mick (43:06)
Haha.

I've got one more agent that I'm working on and I think I'll be ready for a webinar. can show how it all works.

Michael Levitz (43:15)
I will be there. Have a great one. Thank you.

Mick (43:16)
Yeah, thanks a lot you guys.

‍