The Future of SEO: Accelerating SEO Growth with AI Workflows

All right, everybody. Welcome to the webinar. Thank you for your attention on this beautiful Tuesday. I think it's still Tuesday for most people. And so for this webinar, we're going to be talking about how you can accelerate SEO growth through AI workflows. Before we get going, before I do any introductions at all, go ahead and type in the chat where you're coming from in the world. It's always good to know where people are coming from. Eoghan, where are you coming from? You sound like you're in the UK somewhere.

I'm from Ireland, but currently located in cold Chicago. So Midwest.

Yeah. And how long have you been there?

I've been here 7 years.

7 years. Okay. All right.

Keeping a small tech community going here. Yeah.

I love it. I love it. All right guys, type in the chat where you're coming from while we let some people in. We'll probably get started in about, call it 20 seconds or so. And this is meant to be interactive. So as you have questions, go ahead and ask the questions. The way this is set up as well is I'm going to share slides on my side and then I'm going to pass it over to Owen to share on his side and then he's going to have a demo as well. So, and then I think we'll move it back over to me. So anyway, it'll be a little janky, but it is what it is at the end of the day. So let's see. People— okay. Anthony, Anthony, we got you coming in from Vegas. That's great. Anthony, I might be in Vegas next week. We got a LinkedIn user coming in from Houston. We got people coming from Colombia, Ireland. Okay. There's your hometown.

Nice.

Right? Or your home country, I should say. And then Michigan, South Africa. Michigan, congrats on beating Ohio State. LA, New York, Miami. So a lot from the US, but still people from around the world as well. And Phoenix. Okay, cool. All right. So let's go ahead and get rolling on this. I'm going to share my screen and then I'm going to do a quick introduction of myself and a quick introduction on Owen as well. And then I'll get into my slides and then we'll jump into it. And by the way, go share this with your friends too. This is going to be helpful because a lot of people talk about this stuff, but a lot of people aren't doing it.

Slovenia.

There you go. Another European. So I'm going to share my screen. I almost hit share on my, my, on the wrong screen. Let me just do this real quick. All right. All right. Do you all see this? Yeah.

Good to go.

All right. Perfect. Thanks, Owen. All right. So again, we're talking about the future of SEO here where we're almost towards the end of the year, less than 30 days now, right? So accelerating SEO growth through AI workflows. So the first thing that's worth talking about, and actually before I even do that, I'm going to do a quick introduction on myself. So my name's Eric Siu and I have an ad agency called Single Grain. I do a podcast called Marketing School, another one called Leveling Up. And those are both on YouTube as well. So we nerd out on this stuff. And the Marketing School podcast typically will hover around, you can see number 1 for marketing and then top 10 for business. That's when we get to our high point, but it kind of just jumps around. And so we're happy about that. And then we get to interview some cool people and that's what it is. I just love nerding out on business stuff. And so Owen, I'm gonna let you introduce yourself here because you are the special guest here. So go ahead, Owen. For sure.

Thank you. Yeah, so I'm Owen, currently Head of Growth at AirOps. We'll get into a bit about AirOps later on. I'll show some cool workflows, as Eric said, that we currently have on our own website and helping Single Grain with as well. Previously, I have quite a technical background, computer engineer by trade, and I was a growth engineer, ran RevOps and marketing at my last company, which was called Telmex. And on the side, when I have a little bit of free time, I'm building a community called Build AI First, where we try and help other builders who are trying to think the right way, think, I guess, in the 21st century way with all the tools that are available.

I love it. Great. How can people find that community? They just Google Build AI First?

Yeah, go to buildaifirst.com.

There you go. There's a community for you all. Awesome. I'll run through my slides quickly because I really want to hear what you have to say, Owen. For us, maybe we should define these workflows. We have an initiative at Single Grain called AI Workforce. And so for AI workforce, we think about, we're thinking about, okay, what are all the jobs to be done per role and how do we break those responsibilities up and then break them into workflows? So let's say the job of an SEO is a very thankless job. There's a lot of jobs to be done where they have to think about, oh, how do I do keyword research? How do I do internal linking? These are all workflows that can be created and then you can daisy chain them together. And then there's a way to do it with Airops. There's a way to do it by hand as well. A lot of the work that we're doing, we've leveraged Airops, I think, in the past. And I think now we're mostly doing it on our own. But just to help you all think about this in terms of, from a high-level standpoint, from a theoretical standpoint, let's look at the— this is like an AI-driven SEO manager. Okay? So again, what does an SEO need to do? They need to think about keyword research. They have to think about on-page SEO. They have to think about off-page SEO. They have to think about technical SEO. They have to think about Long-tail keyword discovery or trending analysis. There's all this stuff. It's stuff that they need to do, right? Every job is just stuff that you need to do, right? But the key thing is you need to map it out first. Again, in software, people like to call it jobs to be done, right? What are all the jobs to be done first? And then you should prioritize which ones are the highest leverage. And then you can work backwards from that and figure out which workflows should you think about even prioritizing, right? And again, when we get to AROps, I think Owen's gonna be talking about how they're doing it in very smart ways. And I'm keen to see what happens there. But for us, we built an AI consolidation agent. So one of the key things with SEO is when you think about all the content that you're creating, you have to make it easier for search engines at the end of the day, right? So what works really well? When you delete your content, well, you free up crawl budget on your website, which leaves more opportunities for your other pages to be discovered, which means more traffic for you at the end of the day. So deleting content works. Okay? Now, updating content also works really well. Wikipedia, this is why they One of the reasons why they rank so well in the top 3 for a lot of keywords that you search for, like these informational queries, is because they do a good job. They have so many moderators or they have so many editors that are constantly updating the content. And I remember a long time ago, I looked at the Abraham Lincoln Wikipedia page. Don't ask me why I looked at the Abraham Lincoln page, but I did. And in the very beginning, that page only had, I don't know, 3,000 to 5,000 words or so. But now when you look at it, we're talking tens of thousands of words, right? So it's just been updating quite a bit in the past, right? That works. Now, in this case, we built an AI consolidation agent. Consolidation agent here. We use this for ourselves. We use this for our clients as well. This helps us just move a lot faster. What this consolidation agent does is you can see at the very top over here, we just might start fetching Google Search Console data. Then as this agent starts to work through these tasks over here, it's like, okay, let's look at the metrics. Let's look at the URLs that are— let's look at recent URLs based on GSC data. And then you can move all the way over here where you can see it's going up and down over here. Now we're identifying the closest vector match. We're trying to find similar content that maybe should be consolidated so they're not cannibalizing each other, right? Because oftentimes, and this happens on our site as well, we cannibalize the content and we forget that, for example, I remember this in the past, we were ranking really well for the keyword digital marketing agency, but we actually had 3 pages that were ranking for digital marketing agency. So it kept switching from time to time. And that's bad, right? You don't want that to happen. You combine them together. So those of you that have watched Power Rangers, that's how you make the Megazord. Or if you watch Voltron, that's how you make Voltron. You have the 5 Tigers combined or whatever it is. And then at the very end of this agent over here, we provide the URL consolidation recommendations. So we keep a human in the loop because we don't believe that AI is at the point where you can just let it do everything. I think it's smarter right now to have a human in the loop. Maybe it's a different story in the next 2, 3 years or so. So this is a screenshot of us using AirOps actually in the past for this. So refreshing PSEO pages. So we create a lot of these programmatic SEO pages or fetch page summary from URLs. So you can create a lot of these and I'm not going to steal Owen's thunder here. He's going to talk about this very soon. We'll get there in a moment. But this is what AirOps looks like. And this is a workflow that we built internally that will help us find internal links. So this is just another Little— well, sorry guys. I don't know why my screen was— I don't know why that's happening. Sorry guys. But anyway, this is finding internal links over here. So you can see it's fetching different internal links and then it's saying, hey, you should add these internal links over here. And then a human in the loop can improve it. This is just done for ourselves internally and this makes it easier to work for ourselves and also for clients. There might be a way to do this in Airops too, but I just wanted to show you what we have. going on over here. And our thinking here is that in the future, a lot of people are going to build these different agents and we want to have a lot of these different agents available to the public and hopefully they can just use them. So that's how we see it. So before we continue, this is going to go over to Owen in a second. But before we continue, if you scan this over here, you can get on our newsletter. It's got about 30,000 marketers on it and we share a lot of this stuff as much as we can. We create a lot of content, whether it's our marketing school or me speaking at things or people from the team sharing articles and things like that. So I'm going to share it for 5 seconds, 4, 3, 2, and 1. And without further ado, I think we are moving to you now, Owen.

Love it. I will stop and give it to you. Yeah, let me share. All right, everyone can see it. There we go.

Yep.

And so I, I love the way you set it up, Eric, of kind of going through Like there's so many jobs to be done and like there's intelligent ways to set it up. But I think taking it back even one step, and I like this slide here, which kind of just defines where we all need to be or what we all need to think about in the next year. And probably a lot of people on this call are ahead of the curve already, but definitely as knowledge workers, we need to think about how can we best put these tools to use that are available to us today. So I think a lot of people have gotten over the stigma of these tools are here to replace us.

Not at all.

There's so much in the realm of marketing and growth and SEO in particular that we just don't get the time in a day, in a week, in a year to actually achieve. So it's like, how can we put these tools to work for us as companions rather than interns that need a ton of management? So it's about setting them up correctly. wants to move. People often ask us the question in particular, does AI content actually work? And I guess it depends. If you're gonna go to ChatGPT and say, give me 100 articles on X topic and you copy and paste them into your CMS. Well, I am sorry, but you're not gonna get much results out of that. Google's gonna punish you. You're just not gonna engage with your audience. So you're shooting yourself in the foot there. But if you do take a strategic approach and you use tools kind of like we have in the bottom left here, there's a lot of different building blocks, as you call them, Eric, in your like assembly there. You want to make sure that you use the tooling to best help you on the task where it can be put to use. So in Airopt, I'll give everyone a demo in a couple of moments on a few different workflows where we're putting the tools to work internally and similar for our customers. But It's a matter of finding the end-to-end job that you're looking to automate and then finding the right tools to put in at certain moments. And for a lot of— for the end part of that, it's going to come down to an expert in the loop. You still want a human to gut-check it, to review it before you ship it live. But there's a lot of elements and building blocks out there that can help do your work that you're doing manually today better. or at least to the same level. So that's what we'll cover in a bit more detail of how to think about how to find the right tool at each moment. So some customer highlights before we get into it that actually do show that we deliver value. Windly are like an e-commerce platform. They've used some AI workflows again with experts in the loop to bring their high-quality data in front of their customers. So they're writing longer-form articles and listicles that are not top of funnel, but more middle of funnel and bottom of funnel to convert users interested in their products. We also have companies like Bardeen who are taking more of the programmatic approach Eric spoke to, where again, they have unique insights and unique data attributes internally that they're able to create programmatic content around. And again, it's enriched with their insights that Google appreciates and will rank really high. their users actually get value from it. So again, it's not proliferating spam, it's strategically picking the right content and scaling that up and showing success. And then Deepgram, they started from a net new base with their directory. They had this idea internally, Marcel, who's become a super user on AirOps, of there's traffic to be captured from these new AI tools that don't have a huge following themselves of a bigger website footprint. So he was able to use some internal knowledge, build really intelligent workflows, and create landing pages for each of these tools that gave tons of context, how-to guides. And not alone did they over 20x their traffic, but they actually drove revenue on their paid plan at Deepgram. So they were targeting a cohort of users that were interested in new AI tools. They created the pages at scale and were able to convert them. So great end-to-end success there. So kind of we think about you're dealing with the marketing team now, you're in a different function like SEO, and you're trying to say, how do I put these things to use that Eoghan and Eric are going through? And I think from my side, it's a combination of all of these components that's needed to have a really successful approach. So all are valid on their own and kind of all are necessary. to have really good success in end-to-end automation. But it really starts with your strategy. You need to understand at the most basic and fundamental level what it is that you're trying to achieve. There's going to be parts of that where AI steps like a Perplexity, a ChatGPT, a Claude are going to fit in under different needs. Claude is going to be better at writing. GPT-01 is going to be better at longer form thought. And something like Perplexity is going to be better at web searching. So there's different tools even within each of these. Expert in the loop, I mentioned it there. Bardine or Windley or Deepgram would not have had the success that they did with their approaches if they didn't add in that expert in the loop, which was able to provide the initial context or review the page before it went live. And then last but not least, you want to have end-to-end automation set up. I loved Eric's example of going from page to the meta description. And that's where you need to tie all these blocks together and in the end get execution. So part human, part automation. So on Aerops' side, we help you build and scale AI workflows, particularly around your CMS. We've invested the time into the building blocks around website management. Content creation and optimization. So our core principles are bring your own data. It's what Google wants, it's what your users want, and then you're adding value and overdelivering on the promise. That's what you wanna always satisfy your users with. And then just like those examples I ran through, you wanna make sure that you're scaling with an expert in the loop. You have this context, it's getting created and researched in the right way by these automation steps. It's getting enriched with your brand context, but then again, you don't wanna set it live without giving it that final gut check. So input and output, humans should definitely be involved. So how to do SEO right? I feel like we could probably have another 99 versions of this slide, but on my side, it breaks down into 3 different buckets. One is your strategy. Second is the realization of that and the publishing of that content, which gets created, and then you wanna optimize it. So Eric's example around content refresh or content that overlaps with others, that's something you always need to keep top of mind as you're pushing out new content. But yeah, start with creative and proven ideas based on your research. So in-house, you're gonna have topics you might not have prioritized yet. You might have had ideas someone else brought up you just haven't got around to doing. So it starts with what is that strategy? Is it topical? Is it keyword? Is it a new feature? You need to start with what's the crux of what you're trying to achieve. Then you can think about execution and creation. So you wanna execute with AI workflows for each step enriched with your brand context. So bringing your own data and then bringing your style and your tone of voice as well. So I'm sure everyone has experimented with GPT, write a letter for me on this or write a blog post on that. It just doesn't sound like you. So out the far side of one of our workflows, we want it to sound like it's coming from our CEO or our marketing team, or if it's a sales pitch deck, it should sound as if it's actually Nick on our side who's speaking. And then finally, content optimization. It is the one thing that's always deprioritized when we talk about jobs to be done, but it's a necessary evil, I guess, in some ways, and there's significant value. So from the first 2 steps, you've invested in content that is driving clicks, driving value, driving conversions. You cannot forget to go back over time and make sure that you keep your content updated and refreshed. And internal linking is super important as well. So the key crux of all of that is Making sure that end to end you're able to bring in your data into a tool and then be able to refresh it with all of that context combined back into your CMS as well. So I think, yeah, I kind of knit it all together here of AROps and your CMS. So just like Eric's workflow or stuff I'll demo, it all comes back to you have an idea, you've executed on it, you wanna make sure that then you can push it to your CMS. And then close the flywheel and bring it back in then for analysis. So a lot of that comes to being able to enrich it with Google Analytics data to see on-page engagement, Google Search Console to see, oh, maybe I have a ton of pages that have really high impression but no click. There's an opportunity or low-hanging fruit. It all comes from linking your operations back to your CMS and closing that Closing that loop. Yeah, just before I go into demos, I know when I started experimenting kind of like 15 or 18 months ago, I wish someone had presented me with this slide, which along the way I've made a ton of mistakes, but this I feel is good. So where should you start? I will definitely say do not try to go from 0 to 100. Start with something small. And then there's a load of prompts here, but What are time-consuming tasks? What are things that you're doing a load of copying and pasting on? What's like a task that you always go to GPT for and always use the same prompt for? Start with something really basic and then build your automation around that then. So start with a very defined goal. I've seen some of our best builders work out of FigJam or a sheet of paper logically breaking down what is the sequential set of steps in their manual workflow and then trying to say, oh, this 10% here, I can use a tool for this and then rinse and repeat. But it's a matter of iteratively building up these solutions over time. That's how you have the best success. And yeah, I kind of broke it down here. I'll actually demo a few of these today, but top level is some of the strategy ideas you can use. So And when it comes to writing social posts internally, I have a good workflow that helps me go from a hot take down to different forms, like a different format for our leadership, different one for the company page, and then one that I can use when it comes to content creation in that middle section. And we use it a lot for the same starting point of maybe a podcast output or a newsletter we wrote. And then turning that into a content brief that someone can then take and enrich all in AirOps. We also use an internal criticizer, we call it, for scoring our pages versus different personas. That just helps us again align and always keep the target audience top of mind. And then one more that I'll show is around optimization. We can't forget that. And it's oftentimes where there's the most value. And the easiest starting point when it comes to SEO. So be it URL to improve meta description or crawling your site and then automatically suggesting internal and external links. There are many, many use cases that you can get going out of the box. So I think it's time to do a demo. As questions come up, Eric, as well, you might like shout them out to me in case people want me to double-click on anything. But what I have loaded up here is one thing I spoke about, and it's the one thing like once a month I'll always do when it comes to like Search Console, which is what are the pages for us that drive huge impressions but our click-through rate is just like so, so poor. So most of these are like under 1%. What you'll see is they're actually within striking distance. So this is pages from our blog that are in positions 5 through 15, and they have less than 1% click-through rate, which is pretty poor even with those positions in mind. So what I'll generally do is manually review it. There's examples here of content overlap. So we have like a, we just Answer Series A about 2 months ago. We have a previous announcement here, so I'm gonna probably need to do some manual work there to figure out which one I actually want to have Airops funding ranking for, which would be the more recent one. But end to end here, what I wanna do is main thing that'll improve me or improve the click-through rate for me is updating the meta description. So What I've pulled in here from just a simple connection to Google Search Console are these pages, the target keyword of that page. And what I can immediately do then is just run this across all of them if I want. But what it's going to do is it's going to go to this page and I'll show people edit workflow. It's going to go to the URL I provide out of the box. It's going to do some cleanup in case it was incorrectly formatted. It's going to pull everything on that page. It's going to also get the markdown, which might have other context, and it's going to start pulling some of the header tags. So to understand the context of what are we actually calling out in H1s, H2s, and H3s, that's really important. And then it's going to generate some recommendations for me that I get to decide then if I like or not to push to my CMS. And then yeah, goes back to the grid. So if I go back here, we'll see like title-wise it's giving me suggestions and then for description as well. So I can edit this as I want. It's probably better than what we have, so I'll want to go ahead and use it. And so that you can scale across all of your pages. If you've never done anything like that, this is like a really easy tweak to make. And you can get a lot of value from it. The one that I do really like though is our workflow here called Improve On-Page Content. So I'll actually set this one running and then I'll explain what goes on behind the scenes, but very close to what Eric had in his nice slide. This does a lot of research behind the scenes to really uplift the quality of content that you have for SEO purposes. So you'll see here, it takes in a ton of context from my site that I have mapped. It's taking in the target URL, it's taking in the keyword that we want that URL to rank for, and then it's going to start to assemble a load of information on our brand. So we have internal documents around how we want to speak to different personas, how we speak about ourselves, like some of our customers, some of the proof points. And we link all of those in here as inputs. And then last but not least, our writing persona and our writing tone should be included as well. So kind of for our blog, it is like the company voice. For other workflows where it's more like thought leadership or for social, obviously the format, the persona, the tone is going to change. But what we then do is First of all, we're ranking lower down the page than other people. So let's learn what the successful folks are doing. So we look at the target keyword on Google and we're going to scrape the top. I think we take the top 5 ranking pages for that keyword. What we then do is go to each of those pages. We scrape them and process the HTML and do some other formatting. But really we're asking asking GPT then to do the tough work for us. So we give it the prompt of like, hey, you're an expert for this company. You're given the top X number of pages ranking for a given keyword, and you're there to analyze for us what are the traits and techniques that these pages are using that are working in their favor. So I could spend days talking about how we've crafted this prompt. But we're giving it a user-assistant pair example. And that's really where a tool today like Claude or GPT just doesn't have the starting condition or starting context to know how to operate. So what you're doing here is you're feeding it a sample question and then a much better answer out of the box. And then it's a little bit more trained on what you expect from it. And then what we do is we're giving it the top 3 results of what we found that we think are most relevant to the type of content that we're trying to create. We give it all that context. It is then going to come out and put it in Markdown format for us. Then like I said earlier, all of these tools are great, but at some point you need to have the human input. We get it to a point here where We need to manually review it. So if I jump back to this section here, oh, of course it put error out for me. Well, I had one ready earlier. So we have this section and what it's doing is recommending changes to different parts of the page for me. So it's telling me here, like in the introduction section, we want to add a definition of like AI-powered growth, which is actually what that blog should be focused on. There's other recommendations here around trying to add in additional scope. We're going to add on extra steps that will enable us in a clean and concise way to get more keywords in there. In the wrapping up section, we want to improve our CTA. It's able to adjust to all of these different dynamics, which is really nice. Last thing I want to show here is, oh yeah, this write improvements. Taking all that context and what just got shown to me was the output of this big prompt. These things are scary. There's a lot going on here. But thinking back to what I mentioned earlier, you don't want to try and go to this immediately. We didn't go to this immediately. We actually used our generate with AI feature for a lot of it where we could write in, oh, I want a prompt that helps me take top 3 outputs from a SERP result and help me generate recommendations for my page based on it. It would give me something that's like 80% close to this, and then I'm able to tweak it from there on. So we're using humans to create the prompts, review them, do the first pass. to make them really good. And then AI can then follow on with the work and then generate the recommendations for us. So that review step we just looked at, we take the input that's provided, which I should have shown you all, which is like this one. And I click, we'll say I want to accept these ones here and no more. I can go accept output. It's going to continue on its work. And this one's also ready for review. But what it then does is it puts all of that into practice. It will then format it as I want. So depending if your CMS is in HTML or in Markdown, I think we give both work— we give both as outputs. And you will then end up with a page that you can ship directly to your CMS. or what I would recommend you update and reformat before you do that, before you send it to, for in our case, Webflow. So what I would do if this wants to complete and I wanna ship it back to my CMS is we use Webflow. I'll click continue and then we'll say it's a blog post. I will just map my columns here and it will basically go live. Hopefully this one finishes in the next second or so because we can show the Markdown editor. But I am left then with a revamped piece of content. In this case, it tailored it to a specific keyword. We also have out of the box some other workflows that will do internal and external linking for you. How that one could get used is we'll say once a month, once a quarter, you're pushing 10, 15 pieces of content, you want to come in and make sure that everything is interlinked to the new content. And you could just come in, bring your full CMS, run that workflow over all of them, and then push them back to your Webflow, Contentful, Strapi once you're happy with how everything looks. So they, they do take time, I will say, because they're detailed and they're doing a lot of work. But yeah, if anyone is interested in this quick workflow afterwards, I'm happy to share it out and get it set up in your AirOps.

Owen, how do you track results here? I mean, this is nice, but how do we see— I think it's not even a problem with this. I think just problem in general we have internally too is we don't do a good job of tracking results with this stuff.

Yeah, I'll say on my side, I've been a sole marketer for the last 4 months at AirOps, so I could definitely do a better job. And what I've generally done is within a tool like Semrush, and then with Google Search Console at the end of every week, I have a quick script out of Zapier that just pulls down those results and then pushes them into another grid. And then we're just trying to look at the delta. Like for these ones that I would've run, I guess back on 10/29, I wanna see like, have they A, improved in rank and then B, improved in click-through rate? So A little bit manual on my side because I'm working solo at the moment, but generally trying to set up automation to pull the metrics for you on a recurring basis. That then gives you at least when you do get time, the opportunity to go in and see like, oh, at least these things are moving, or sometimes they haven't moved. And so I think just committing to the automation and committing to some analysis, that is kind of the— that's the last part the humans are involved in.

Got it. Cool. So there's a couple questions. So what CMS system does AirOp plug into? WordPress? What other CMSs?

Yeah, Contentful, Strapi, Shopify. Those are the ones I can name off the top of my head. And then generally for WordPress, and then for anyone else that we don't cover today, we're able to hook that up for you With an API step, which is pretty simple.

Great. We got one from Nikhil here. Are there templates for these workflows? I'm guessing yes, there's a bunch in Airops.

Yeah, for sure. There's a ton of templates. If I go to here, once you log in, you can go new, you can start with a template, and we have some here. They're across the different buckets of content creation, strategy, Even imagery. That's another big one that people often overlook when it's meta images or alt tags. So these are some that we provide immediately out of the box that if you sign up today, you can get going with.

Cool. And then another one here, how are these configured in the knowledge base, brand kits, et cetera?

Oh, great question. And I might lean into my next post with that. But, or my next demo. But if we look at my data section, I'm in the wrong one. Social post. Let's go your data. So how I have it set up is one, I wanna keep our sitemap always updated. So this we can set on a recurring basis. Then for some of our other knowledge that is related to company, and I have other customers in here, We're able to push that context directly to a brand kit. And we have our product marketer starting this week, so I'm gonna get her help in enriching this with a lot more context. But brand kits then are— that was the input we just saw to that last workflow where it's getting your tone of voice and the right style that you're aiming for. And then depending on your exact workflow and your objective, I'll show you my social post generator in a second. But maybe you wanna bring in different formats. So here I've connected up to formats I've liked that are in my Google Drive, and that's just super, super simple of— this one's called the Industry Insight, but it's just generally posts that have succeeded. I bring them into AirOps' memory stores, and then I can reuse that format on a recurring basis. And then similarly, writing styles. I personally don't have one that I can define very well, but I like an Adam Robinson, like a Gary Vee or Kieran Flanagan. And what I've done is trained another workflow given 10 or 20 different LinkedIn posts. Can it output for me someone's writing style? So all of this to say, like your knowledge base and your brand kit can hold such like a diverse set of materials. And then it's up to you as to how you then wanna use it.

Great. One final question here. So David Dodge asked this, Neil Patel wrote and told me that his human-written content was outperforming AI-written content by 5.44x. What is your strategy regarding content writing and what are your thoughts on that?

I'm sorry, Neil Patel's example, is that his AI writing or just in general? Do you know?

He's saying in general. Okay.

We've seen the opposite. Some of those customers I've shown earlier where they're like an Angie or a Thumbtack or like massive MasterClass, massive content engines, they have not just improved the quality of their output and that's like using a human evaluator, but also it's starting to rank much, much better as well because There's always that last 10% of context that you just can't extract from an engineer internally and get into a blog post. So they've workflows set up to increase the detail they can insert. So I'll say depending what sample size you take, you can probably prove either case. We're obviously working with people who want these workflows to work at scale. But when it comes to Creating content that works. I think going back to my slide here on how to do it right, you should not just set up a one-step prompt in an Airflow or go to GPT and say, write a blog for me. Someone who asked that question about brand kits and knowledge bases has a bit of knowledge about AirOps, but really you wanna bring the detail, the unique insights, The data that makes your company you and then use that in your workflow. So just like if Neil Patel is writing something on a given topic, he has domain expertise, probably has data points. When you get to the point of writing that with AI, you don't want to lose that essence. So you need to come with your call transcripts from a sales call, or you want to come with a pitch deck, or you want to come with an interview. That you've had with some sort of prospect. So I would say how to do content right, load up and structure all of your best data, make sure that you have your product marketing in order, which can be as simple as what your tone of voice is, who your persona is, and then yeah, what audience you target and then knit them together in the way that you would if you were going to write it yourself. So Create your end-to-end writing process, but with automation in the right areas.

Cool. All right, I'll take it from here. I got some more slides for you all and I'll share some prompts with you all that you can take some screenshots of as well. And then we'll open up for more questions. You stay till the end, there's a special offer from AirOps. There's also a special offer from us as well. So stay till the end. And if you stay a little longer till you get the prompts too. That's what it is. All right, so let me share my screen again. Here we go. And let's do this. And let's do a slideshow. And I think we're good again.

Okay.

All right. So I'll try to go quickly because we all got a busy day here. So we— Owen talked a little bit about programmatic SEO, right? So just to give a little refresher to everybody on how it works. If you look at TripAdvisor over here, you look at where my cursor is. 131 million visitors per month, right? And this is from Ahrefs. And traffic value is about $66 million. So if they're— it's not the perfect metric, but if they're paying for that, the Google Ads, they'd be paying $66 million a month. So that's their traffic value, right? And then they rank for about 24 million keywords. So what do they rank for? Well, it's things to do in Seattle, things to do in Japan, things to do in Ireland, right? Fun attractions in whatever state. There's a lot of geolocations, a lot of— it's When you have a bunch of geolocations and a bunch of activities, it works perfectly for programmatic SEO. And they have millions and millions of these pages, millions and millions of these keywords, right? And so the question is, I mean, you can definitely use something like AirOps. You can choose to do it yourself or not. The AirOps version is certainly there. And then Owen will kind of give you the special offer towards the end. The way we did it is this was our initial push into programmatic SEO. So for us, it's agency-type keywords, right? So San Francisco digital marketing agency, Chicago digital marketing agency, Chicago SEO agency, right? There's a lot of permutations there. Not as many as like a TripAdvisor, for example, or an Expedia. But when we started this initially, you can see it was like, it was kind of slow going. It's trying to get going here. It's not that many clicks, right? It's like, right? It's like kind of petering off over here. Now the cool thing is with PSE, we have a story. There's the growth. There's decline and there's the bounce back, right? So the story with this graph in the very beginning is we just took too long. We didn't double down enough. We didn't bet hard enough, right? Now, then it started taking off. It's like, and then an update happened. It's like, oh, well, okay, it's starting to go down, right? It's up again and down again. And then it shot up again. So what happened here? Well, one, the initial growth, we didn't bet hard enough. Okay? So when you have something that's working, whether it's in business or whether it's in your personal life or whatever, you're probably going to want to double down on that, right? Not financial advice. That's the first thing. Second thing is when it's declining, you shouldn't let it. I mean, sure, there's an update that happened, but where we didn't put enough effort in the very beginning was we let the prompt do its job without QAing as much as we could have, right? So quality assurance, whatever you want to call it, quality control, QCing. We didn't spend enough time on the quality at the end of the day, right? Google has said themselves, they don't care if it's AI content or not. They care if it's providing value to the user. So if you focus more on quality, you're going to do a lot better. And so that's what happened over here. The uptick over here is because we started focusing a lot more on quality. We started deleting a lot more content. We started deleting a lot of pages that after 45 days got zero traffic or got zero links, right? And so for us, it's like you can probably create an AROps workflow for that as well. And so for us, we're just like, let's focus on quality. Let's also focus on upgrading these pages too. So a lot of these services pages, it would link to a relevant podcast or embed a relevant YouTube video where we're talking about that topic, right? Or we'd internal link to some relevant pages as well, or some recent blog posts that are similar, right? And that way we're updating the pages at scale. And so this all comes down to workflows at the end of the day. And again, you can do it manually or you can use a tool out there. So you can see we talked about slow start, Google update recovery, continuous audits, a lot of pruning, right? So that's what happened at the end of the day. I'm oversimplifying it, but again, you can use a tool like an AROps for the workflows, or you can do it yourself manually. If you're going to do it yourself manually, take out your cameras over here, take out your phones. Here's a prompt for PSEO. I'm not going to read this stuff to you right now. This is what we used initially, right? And this can get you started. And all we're using is OpenAI's API. Okay? It's not exactly rocket science. Sure, it's artificial intelligence, but it's not exactly rocket science. So that's the prompt for PSEO. I think there's another part 2 as well. So take a screenshot of this one too. So this is above the fold. So these are for our service pages, for example, right? So for example, you're a content writer for Airops. For background purposes only, here's an overview of the company's activities, right? Company description. And we pull all this information and then that way we're able to Modify above the fold and then also modify the meta description, the writing style and the content, right? So these are the prompts that we use for PSEO. For us, it's a powerful and scalable tool. It's not the only tool, right? Just because you have one hammer, you don't want to keep using that hammer for everything, right? So for us, pretty quickly, this is in the initial months of investment, but we invested about $26,000 and made $105,000 back in revenue. And this is only in the very beginning. There's compounding effects to this. This, right? So about 300% ROI. That's not bad. And so all I would say to you again is, you know, just because people, you look at TripAdvisor, you look at Expedia, you think you can pump out a million pages. No, don't do that, right? Maybe just do like 50 pages. Maybe just do 100. And if it's working well, maybe you do a little more, inch it up a little bit more and more. I think that the problem with a lot of people is they want results yesterday and they're willing to sacrifice their entire website just for some quick results, right? Don't do that. I've seen that happen a lot in the past. I've tested this on a lot in the past, in my early SEO careers. Don't do that, right? Or else you're going to understand what it feels like to get your website torched. So that's a case study on us, right? Like this stuff works. And I'm going to show you an actual case study that we did with a client, a wine client of ours, where they're only getting an average click-through rate of 0.8%, which is not high. And this is on a section of their website. So they sell wine, right? WineDeals.com. So for them, we wanted to scale content production while driving more conversions. And the process here was we just created 200 new pages. Again, you don't need to do tens of thousands, right? 200 pages. So all we did was let's just focus on high-intent pages that are likely to drive conversions. And so all we did here was we said, okay, the PSEO topic, programmatic SEO topic that we targeted was best wine types, so best Chardonnay, Under $10, under $20, under $30, under $40. These are perfect for templated pages, right? And so each page featured dynamically inserted product listings along with AI-generated copy to support it, including FAQs and general information about the wine type and price point. Okay. Very, very simple. The results here, again, this is for a very small section of their website, 6,300 additional clicks, right? 184,000 impressions. And then average before was 3,400 and 100K, right? I guess we're talking 3,000 more clicks or so, right? And this is over a— it wasn't over a super long period of time. Looked like it was 3 months after publishing and it's still compounding. It's still moving up into the right. The average click-through rate on these pages was 3.4% versus 0.8%, also because they're higher intent as well. So this just goes to show you that again, PSEO is a tool. It's not a panacea. And even though I love to talk about it a lot, there's a lot of other tools I have under my tool belt, right? Batman has a lot of tools in his tool belt. It's not just PSEO. Added about 1,123 new keywords and 111 of them are in positions 1 to 3 and 445 of them are in striking distance of 1 to 10. 105 product adds and 155 cart views. Average engagement session of 40 seconds per session. And that's how it works. So as we get towards the end here, so there's a lot of things that we're building internally, but this is like, okay, how do we build a single-grain AI where you can just command it to do work inside of your Slack instance? and tell it to check on your workflows, ask for the results from your workflows. These are just things that we're constantly messing around with. I'm just excited to see what people build in the future, what AROPS builds in the future, what we're building in the future. I think the future is very tech-enabled and I think that's a very exciting future. That's that. Owen, I'll move this over to you for a second. Actually, you know what? There's a slide where it's on both of us. If you guys go to airops.com/signup, there's a discount code over here, AI-GROWTH. Again, it's AI-GROWTH. And you have this training opportunity over here, which I'll let you speak to. So go for it.

Yeah, pretty short and sweet. We've had a ton of success with some of the customers we highlighted there. And if anyone wants exclusive training opportunities, not a sales pitch, we'll just help you and your team maybe break down one or two different workflows you're thinking of or different strategies, email team@airops.com, mention that you were on the webinar. And yeah, I'll personally help you out afterwards.

That's great. And someone asked for the QR code again. So here's the QR code for the email list over here. I'm gonna leave it up for 5 seconds 'cause we have one more slide before we get to the end. So I'm gonna leave this up for 5, 4, 3, 2, 1. I just realized we need to change this Twitter bird to an X actually. So it's a little old, but yeah, look, you can scan these over here. So we, on our end, the offer is there's an AI workflow strategy call that you can have with the team. And it's more about just, we build them from scratch. And we also use AirOps as well. AirOps is great. But if you go to singlegrain.com/workflows-webinar, or you just scan this over here, you can focus your phone on that and then you can put yourself in for an AI workflow strategy call. And then on the right side, again, if you email team@airops.com, there's a 10x exclusive training opportunities. And again, you can go to airops.com/signup. I'll just let you guys take a screenshot of the screen over here, or you can choose to scan the QR code. But aside from that, Owen, as we work towards wrapping here, do you have anything else you'd like to add?

Yeah, I think as people dip their toes in the water here, start small. So as you said, Eric, if you are focused on programmatic SEO, please don't go and create 200, 300, 400 pages. Start small, do the right things that you've control over to begin with. And then as Eric said, double down where you do see success. We have had customers as well that just as you showed, Eric, had that decline, some algorithm hit or whatever. But how they turned it around as well is double down on what works. And then they started to add more of the context they started to see their audience engage with. So how they saw what people engage with was look at those top PSEO pages that are driving time on page, clicks, and then click-through rate further into their funnel, and then just learn from those ones. And if you give the data to a GPT and ask it to analyze it, it'll probably do a pretty good job for you. Yeah. Double down where you're working, use the tools you have to figure out what is working, and then use that to iterate further.

Great. All right, everyone. So that is it for today. Thank you for your time. Thank you for your attention. And we hope to see you around for the next webinar. And Owen, thank you so much for the collaboration. This has been great and looking forward to one in the future.

Yeah, likewise. Thanks for having me.