So What header image

So What? How to use AI for SEO

So What? Marketing Analytics and Insights Live

airs every Thursday at 1 pm EST.

You can watch on YouTube Live. Be sure to subscribe and follow so you never miss an episode!


In this week’s episode of So What? the team discusses the best way to use AI for your SEO tactics, as well as the changes that have occurred.

Catch the replay here:

So What? How to use AI for SEO


In this episode you’ll learn: 

  • What SEO tactics you should be using in 2024
  • How to use AI for SEO analysis and planning
  • What changes in SEO tactics to be aware of

Upcoming Episodes:

  • TBD


Have a question or topic you’d like to see us cover? Reach out here:

The follow transcript is AI-generated and may not be entirely correct:

Katie Robbert 0:26
Well, hey everyone, Happy Thursday. Welcome to SWOT the marketing analytics and insights live show. I’m Katie joined by Chris and John. Hello. I was like, Oh, we going for the high five.

John Wall 0:38
Standard High Five

Katie Robbert 0:39
In case you’re wondering if this show is live. And if we rehearse ahead of time? The answer is yes, it’s live. No, we don’t rehearse Clearly. Clearly, we would have nailed that by now. This week, we’re talking about how to use AI for SEO. So this came up. Because we were having a conversation about what we’re doing for ourselves, and what we’re doing for our clients. And it occurred to us that, yes, generative AI is taking over a lot of the conversation. But that doesn’t mean that the basics of marketing still don’t apply. So this week, we’ll be covering what SEO tactics you should be using in 2020. For how to use AI for SEO analysis and planning and what changes in SEO tactics you should be aware of. So with generative AI on the scene, there have been a lot of conversations and questions around, well, how is that going to change organic search? And the answer is a lot. However, there are still basic foundational things that you should be doing with your SEO in order to stay competitive in order to still get good rankings. So Chris, where would you like to start?

Christopher Penn 1:43
Alright, well, let’s start with, I guess, the SEO landscape. So we know that, as of right now, generative AI in all its incarnations is, is having an impact on search. There have been a couple of different studies and pieces on this topic. So one study from late last year by Search Engine Journal said that from a for one B2B firm, genitive, general AI, specifically, Google Search generative experiments, aka the the summaries that Google automatically makes a search results would have a negative impact of between 18 and 65%. On organic search, were in simply unbranded organic search. So for folks who maybe are not as well versed on SEO, there’s two forms of SEO of search right. There’s branded and unbranded branded search is people searching for you by name, looking for a Trust Insights, looking for Katie robear, et cetera, unbranded searches, people looking for things like consulting firm near me gas station near me, things like that, where there is there’s no brand call out. But there is some intent. And for a long time, content marketers have been working on how do we capture unbranded search, because if you capture search on brand search early on, you’re basically capturing people earlier on in the customer journey. So if you can provide helpful useful content to them, they will presumably become, eventually become customers, right? People need to know who you are, in order to search for you and thinks that the impact that generative AI is having is that generative AI is undercutting that by saying you know what, you don’t need to go through 40 different search results, you can just have the summary, a nice summary prepared for you by AI. And that’s that the challenge is for a lot of generative AI engines, there’s not a whole lot to click on it, which means that there’s not a lot of ways to, for us as marketers to get that traffic. A second result recently, from the Wall Street Journal, by news publishers were saying they’re seeing they’re estimating 20 to 40% losses in traffic. Because of such generative experiments, the ability for people to again, just go right in and get the summary. Now we’ve all had that experience. Anytime you have gone through and search for say, like a cheesecake recipe. There’s a gazillion to search from, and when you get to the search the cheesecake recipes, it’s like 48 pages, like this person’s grandmother’s second cousins, dogs, roommates, friends, like just tell me how much sugar to put in the cheesecake. And it makes the recipe for the gender of AI cuts all that out.

Katie Robbert 4:37
What you know, and you’ve told me this before that, you know, you can use generative AI as sort of a standard search engine but I feel like and maybe you know, maybe I’m in the minority, but I feel like that means you have to know exactly what you’re looking for. You know, so if I go to if I pull up a search engine, and I look up a cheesecake recipe I don’t necessarily have a specific one in mind. And so I’m someone who’s more interested in looking through the different options, you know, the, the ones that take three hours versus the one that’s take, like 45 minutes, like, what is the difference between those? You know, or what does it look like if I substitute ingredients, but I don’t necessarily know that I want to do that yet. And I feel like if you go to generative AI, you’re not getting that variety of responses. And, you know, both I think are applicable. But I feel like that sort of the downside is you’re gonna get one specific answer, versus with a search engine, you’re gonna get a variety of answers. Is that a true statement? It depends.

Christopher Penn 5:42
So this is one of the more well known AI based search engines. This one’s called perplexity that a lot of folks seem to enjoy. And I said, Write me a good recipe for cheesecake to make it home for for people who are gluten free. One of the things that gendered AI makes possible that you can’t do in traditional search is have that longer conversation even up front, like, hey, I want I want this now that you can have a conversation with it. As opposed to how we’ve used we’re used to doing search, which is writing, you know, cheesecake recipe, gluten free, or some some very contorted version of language. The conversational aspect of generative AI, particularly if you’re using voice interface, is allows you to have much more thorough queries, especially when you’re, you don’t like you said, when you don’t know what you want, like say, Well, I’m thinking about this. And here’s kind of my ideas, stuff like what do you think, and these generative tools are much more able to handle that as opposed to a traditional search engine.

Katie Robbert 6:41
What’s interesting is though, I’m looking at your prompt here, and you’re saying write me a recipe, versus I would be looking for, you know, gluten free cheesecake recipes, for example, or I would be looking for gluten free cheesecake recipes, you know, premade crust or without premade crust, or like, whatever the thing is, you know, so it’s, it’s interesting to see how this is going to start to evolve in terms of how people are using I mean, we we went through this as marketers, not that long ago, or maybe it’s longer than I’m remembering. Regardless, we went through this, when smart devices became part of the search engine, sort of have that ecosystem. Because when you’re talking to a smart device, you tend to speak in complete sentences. Like, you know, hey, smart device, I don’t want to trigger mind, hey, smart device, helped me find a recipe for gluten free cheesecake. Versus if you go to a search engine, if you open the browser, you’re set, you just sort of truncate your sentences and say, gluten free cheesecake recipe. And that’s all you you know, you put it in. And so we’ve, as marketers, as we think about SEO have had to change how we’re creating content to accommodate those short and long tail keywords.

Christopher Penn 8:06
Exactly. And so even in regular search, you’re gonna start seeing the ability to use generative AI. So what does this mean? So it means a couple different things. One, it means that, from a marketing strategy perspective, generative AI will consume a lot of the unbranded search, in some fashion, whether it’s Google, whether it’s Microsoft copilot, whether it’s Bing, whether it’s for Flexi, that’s kind of not intended, kind of baked in at this point, which leaves Rand branded search. So one of the key strategic imperatives, and we’ve been saying this now, for four years, one of the key imperatives for marketers is you must invest heavily in your brand, you have to invest time and effort, building your brand, so that people know to search for you by name. Because even if they go to a, an AI based search engine, that will still get you to your website and stuff, which is really, really important.

Katie Robbert 9:13
So what are some of the basics that we still need to be doing within our own content within our own websites? You know, I feel like you know, when someone says SEO, they’re thinking about, Oh, well, do I need to have a target keyword anymore? Do I need to worry about the metadata? Do I need to worry about, you know, the h1 tags and you know, canonical URLs and those like, Where does where to all of those best practices fit in to 2024? What are the things that if I had to prioritize, I need to make sure that I’m absolutely doing all the time and not skipping regardless of what generative AI is doing?

Christopher Penn 9:53
So that’s the same four pillars that have been part of SEO for 20 years now. Technical Write your website has to be hosted somewhere that is fast. And that the it’s your site is functional, it’s operationally it’s quick. And the place to look at that is in places like Google Search Console, right? So if you if I were to pull up Google Search Console, you want to look at okay, well, is my site fast? Let’s go into our mobile report here and core web vitals. Let’s look at pages here. And so this is the Trust Insights website, it is flagging that we have pages that take longer than two and a half seconds to load. So that’s technical. That’s that’s pillar, one of SEO. Hill, a two is on site, which is things like, are you using schema? Are you using metadata are using it? Or is your site? Is the content itself structured properly? Does it have an h1 tag does have a title tag? Does it have, you know, the essentially, the scaffolding for good content? Part three, is content? Like is the content any good? Do people actually like what you have to offer? And number four is off site? Which is how much? How much? Are you creating expertise and authority and trust? through things like links and and signals out there that search engines like Google know, to say, like, Yes, this is a reputable site. So those are the four pillars, and that has not changed. And that probably will not change. In fact, depending on this is kind of interesting. Depending on the AI system you use, some of them do what’s called grounding. This is you will see this in Google’s new vertex AI system, it will actually go to Google Search catalog to validate that what the AI is saying is correct. So having good SEO and having your content in Google means that AI systems themselves, we’ll check Google search to see if the information they were repeating back is correct.

Katie Robbert 11:57
All right, John, get out your pencil. And we’re about to make your to do list longer again.

John Wall 12:01
Right? Yeah, yeah, all the Google tools, the whole pack is there, you know, I’ve done that PageSpeed Insights is great. That gives you a little more insight than this graph here to tell you exactly what’s wrong and where stuff is going. And yeah, there is the Lightspeed plugin, too, if you want to get to those PageSpeed stuff, having a plugin where you can just click and with that, you can check out any site. So yeah, that’s definitely it’s funny how there’s a bunch of stuff you can do on the content. But it’s like, if your site is just not working that well, you, you know, you gotta get that taken care of before you start digging into everything else.

Christopher Penn 12:37
Exactly. So this is PageSpeed Insights, as John mentioned, you feed in a you a URL. So look at if you want it to be super fancy, you could take one of the URLs that Google was flagging as maybe an issue, feed it in here, and then it will give you its analysis of your website of the things that are fast or slow. So one of the things, and this is an important use of generative AI, one of the things that a lot of SEO practitioners and a lot of lot of marketers who don’t live in the SEO world will can do is ask, Hey, I like what does this stuff mean? Like, what do I do? So for example, you have here, reduced JavaScript execution time, and there’s a bunch of stuff here. If you are a marketer, and you are not an IT person, this probably is not super helpful. But one of the things you could do is you could take this table in some of this data and say, Hey, gender debate language model of your choice. What does this mean? What am I supposed to do about this? So if I go to Gemini here Hi, I have some SEO questions. I got this report from PageSpeed Insights telling me to fix my site to make it go faster, what does this mean and where should I start? So, even just something as simple as this, for the non technical SEO person or the non technical marketer is is a good place to ask so so here’s prioritize the problems scripts, look at the report closely. The highest tool CPU time and your biggest culprits focus on those first optimization strategies. You can minify compress, defer display third party scripts evaluate the necessity so I go back to this and say, Okay, well, what do we got here? That’s first party that’s Tag Manager. I could continue asking questions over and over again, like well, now what do I do? Well, now what I do, it’s a great if you’re familiar with the term pair programming, where you have a developer and you have someone who’s sort of a manager kind of working together. This is in a way your pair programming but your pair marketer your pair SEO expert where you can say, hey, I don’t know what I’m doing. And for a lot of people this is actually can be very, in some ways comforting. Because you don’t have to worry about asking questions to someone who might judge, you say, Oh, well, you should know that what’s wrong with you? You’ve been in marketing for 20 years. Now, you should know what that means. AI is not going to do that to you.

Katie Robbert 15:23
Well, for example, I don’t know what minify means. Like, I look at that, and my first question would be, like, cool. How the heck do I minify? Like, what is that? What kind of tools do I need? Because I think that that’s where a lot of marketers sort of throw their hands up is like, yeah, we get the list of things we need to do to optimize our sites to make our SEO better to increase organic search. But half the stuff like you need to know another language in order to even do it. And so we sort of go back to the pieces of SEO, that we feel a little bit more comfortable with, which is keyword research, making sure that my content has the key word mentioned three times. And it’s like, okay, those basics I can handle. But when you’re asking me to minify, and w three total cash, and you start to lose me,

Christopher Penn 16:21
do you and again, that’s one of the things that’s useful about this tool is you can say, what is that have a WordPress site? How do I do this? What is this thing? How do you do that? How do you install the plugin? And you can just keep asking over again, I’m still confused. What is this? Again, these are great conversational tools. And once you’ve given them the basics, either using, for example, the Trust Insights race framework, or the Trust Insights, pair framework, either one of those would be a good place to start with this. But you can just keep asking questions over and over again, I don’t understand this. Tell me what this means. What should I do about this?

Katie Robbert 16:56
So that’s the technical side of things. But where does like so where does keyword research fit into all of this now? And 2024? How can we use AI for SEO keyword research?

Christopher Penn 17:10
So there’s a few different ways you can you can tackle that. Number one is to take the data out of Search Console and say, Well, what are we being found for? So let’s go ahead and do that I’m gonna bring up let’s see up there as studio. And well, first, let’s go to Google Search Console, close this tab here, I don’t need this anymore. I don’t need this anymore. So this is our Google Search Console. I’m going to export just the queries out of here, which is the terms that you get found for in Google Search Console. And you always you end up with a file on your desktop. And so I’ll just take a quick peek. These are this the terms and the volume that you get found for right now. So let’s go ahead and put this away. And say, Today, let’s talk about some keyword research, and topics for SEO. I have a list of queries, where which my website is found for from Google Search Console, I need your help as an SEO expert to understand the broad topics that my website is being found for. What do you know about this topic? In general? What are the best practices? So just turn off the safeties. I’m starting off with part of the race framework, roll action, context execute, but instead of the context being provided by me, I’m using the Trust Insights pair framework to prime the model with the model’s own knowledge about SEO and keywords just to get the right words and phrases into our conversation. says, Okay, great. I’m going to give you my Google Search Console queries report. Now, please analyze it and tell me broadly what my site is being found for. And now let’s go ahead and upload our file.

And now it’s going to take this real big file the CSV file and parse it and understand what’s in here. Trust Insights brand and people a significant portion of searches for your brand name, marketing, analytics and data science, Google Analytics 4 AI and machine learning. Data Quality other marketing topics additional observations. A substantial number of queries are longtail keywords include indicate specific user intent, potentially less competition, while branded searches significant this healthy brand mix of non branded queries, informational versus transactional, the majority of queries seem informational suggesting your site primarily attracts users seeking knowledge and resources, recommendations, content optimization, technical SEO, and so on and so forth. So this is what we’re found for right now. Art of the question that you might have is, is this what we should be being found for right, is this relevant? So the question we can assets a great, I’m going to give you our ideal customer profile. Next, based on our ideal customer profile, ICP analyze our current search performance results versus our ICPs needs and pain. What things might our ICP be searching for that we are not satisfying? Or creating enough content? So, Katie, you want to talk about the ICP while I get this loaded?

Katie Robbert 21:18
Yeah, absolutely. So a few months back, Chris, and I put together our ideal customer profile. And so we took a blend of our CRM data, our marketing, automation data, our Google Analytics data, and then our also just general knowledge of who we want our customers to be. And we started to put together Oh, and LinkedIn information, LinkedIn, data as well from people who are current and past customers, and then also who we would like to target as customers. And so we put all of that together, we used generative AI to start to put together a profile. And so we focused really heavily on not only the firma graphics in terms of company size, or where they’re located, but really trying to dig into the pain points. What is it that is keeping these particular individuals up at night? What are the problems that they’re having, as it relates to the types of services that Trust Insights offers. So it’s not just Well, what’s keeping them up at night is the fact that their roof is leaking? It’s what’s keeping them up at night, regarding the future of generative AI in their marketing practice, or, you know, how secure their data governance is, those kinds of pieces. And so what we did was we were able to put together a very comprehensive, ideal customer profile that really dug into those pain points. What would be the triggers for making decisions, the types of things that they’re looking for, and so it’s very comprehensive.

Christopher Penn 23:00
Exactly. So we fed in the ICP, and it said, Hey, here’s your ICP versus your search results, pain points and goals lemon demonstrating ROI. While your site ranks for terms like B2B marketing, analytics and marketing analyst consultant, there’s limited visibility for queries directly related to proving social media ROI or value. Consider creating content that specifically addresses this pain point such as measuring social media ROI, or proving the value of social media marketing, data overload and insights your site performs well for data quality related terms. But there’s room to expand on content that helps marketers translate data into actionable insights, topics like data, storytelling, data visualization, or turning data into insights might be valuable. workflow in process improvement, keeping up with trends, content, creation, pressure, alignment, clarity, limited resources, etc. So this now highlights, hey, we know from Search Console, this is where you’re what you’re found for. And this is what you’re not found for the it should be based on the ICP.

Katie Robbert 23:57
Which is incredibly helpful, because I know, historic, I know, just from doing the work, we have not refreshed our keyword list in quite a while. Definitely prior to generative AI being part of the everyday conversation. And so this, I would take this information and start to use that to start to do that keyword research to say, Okay, what topics, you know, should we be really talking about, because this is a really good high level list. And the thing that, at least that I think I know about SEO, is it’s not enough to just create one piece of content for one keyword. If you really want to be known for it to rank for it. You have to have a whole cluster of content with related keywords. And so maybe that’s, you know, you pick one piece of content and that’s your Cornerstone content, and then you have everything related to it. So let’s say our quarter stone content was predictive analytics, which actually is one of our pieces of kohner, Cornerstone content, you would start with what is predictive analytics? That’s your cornerstone. And then every other related piece of content that refers back to that piece of content is, how do I use predictive analytics best practices for predictive analytics, predictive analytics for marketers, predictive analytics for agencies? What are the downsides of predictive analytics? When can I not use predictive analytics, all of those become individual pieces of content. But that takes planning. And so I’m looking at this, and this, to me is the high level topics. It’s not the keywords themselves.

Christopher Penn 25:37
Exactly. So let’s dig into that let’s drill down into let’s we can say, Okay, this is great. This is a great and helpful high level overview of our content gaps, dig into. Please give me a concrete offensive list of keywords that our ICP might search for, on this topic, limit keywords to four words, or less. And let’s see what we get. And the reason I specified four words less is because yes, there’s a lot of longtail stuff, but we want sort of the top level stuff. Because in a minute, we’re going to validate this with an SEO tool. So Wow. This is great. Oh, my goodness.

Katie Robbert 26:42
John, what I should have done is have you worked side by side with generative AI and see if you could generate this list just as fast off the top of your head?

John Wall 26:51
Right, good keywords.

Christopher Penn 26:53
Okay, so let’s go, I’m going to start a separate text document, because there’s just too much on screen there for me to capture. We have all these which are comma separated, which is annoying. So let’s blow that out. And let’s go into these which are also place that when now let’s go into our list of solutions and tools. And go for here into challenges and pain points. So now we have this lovely, yeah, let’s take out process lines containing cones because you don’t want them take that out. Alright. So now we have a starting point of that 57 different keywords. Let’s go ahead now into and you want to pull open the SEO tool of your choice, I’m going to use h refs here, I’m going to go into Keyword Explorer, I’m going to take the seed list here and paste it right in and say show me the search volumes on these topics. So we’ve got a bunch of stuff. And what we want to do is we want to filter so a keyword should have it. For us for it to be worth your time, you should have at least 10 searches a month. Less than that, not super helpful. This is this is particularly a good start for a lot of these terms. So we’re gonna go ahead and check this off here. And I’m going to add oops. Now let’s add this to the Trust Insights list. And so that will add those terms to our broad list and now click on matching terms. And see if there’s anything like data analysis down business intelligence analysts data lake versus data warehouse, what is data analysis? What is market research, Master Data Management? What is data governance, exploratory data analysis, what is the data lake, and so on and so forth. So there’s a lot of excellent matching terms here. That would be that should be on our keyword list based on what we’ve been talking about so far. So let’s go ahead and add those in. And now, our next step would be for for these terms to figure out what can well what content should we create for this. So let’s go back to our thing here. And of this list when we look here, we see data visualization, really high up on this list. This great data visualization seems to be really important. And so I think it makes sense to create some Hub and Spoke content around that. Are you familiar with the Hub and Spoke content marketing strategy?

This is essentially, as you were talking Katie, the your Cornerstone content and you have all the the ancillary pieces.

Katie Robbert 30:13
And I will say, you know, if you’re some if you’re someone who’s using these tools, and you’re asking these questions like, Are you familiar with this? Or what approach should I take for this? It’s topics that you as the end user should know whether or not the generative AI tool is giving you the correct information. Because they still hallucinate, you know, so you could say, are you familiar with the Hub and Spoke content marketing strategy, and it could say, Sure, and then give you a gluten free cheesecake recipe. So you need to make sure that the information that’s giving you back is accurate before you go down the road of like, using these tools to then go ahead and create the strategy?

John, have you do you use the Hub and Spoke content marketing strategy? Much? Have you used it historically?

John Wall 31:14
Yeah, I mean, we used to do a ton of stuff with software development topics is huge, right? Because you can pick different methodologies. And then you can drill down to the tools and techniques. So yeah, in fact, there was a whole software configuration management community that was built and just turned out money by cranking out Hub and Spoke content. You know, if you, if you find the right space for it, it can be incredibly effective. But the challenge is, yeah, people need to the people in the space need to be able to follow that Hub and Spoke themselves. You know, if there’s certain parts of it, they don’t understand at all. It can break down and then you have some glitches in it. But yeah, for the right play, it can definitely be fantastic content.

Christopher Penn 31:57
Yep. So we’ve got that we’ve told it create the hub and spoke model. It says here are your two hubs, data visualization for marketers, data visualization for business leaders. These are the two hubs, data visualization for marketers hub, that’s one pillar pages, The Ultimate Guide, and so on and so forth. It’s up with three different headline options, data visualization, business leaders, for strategic decision making your Hiller your pillar pages, headline options there. And then, and then we go into the spoke content. So tools and techniques, top visualization, Data Tools, for marketing teams, choosing the right charts and graphs, your marketing data, and so on, and so forth. So now, using the data, we’ve gotten front and validated from an AI model, validated with an SEO tool, and then brought it back into generative AI, because we know what we’re aiming for, we now have two hubs. And we have nine, two pillar pages and nine spoke pages for each of the those two hubs. So this would be the sort of the content that we’d want to create, your next step here would be taking something like this one here, say, Great, this is an excellent start. Next, keeping in mind, our ICP and their needs, and keeping in mind what we are and are not found for, please create a an outline for this spoke page. And I chose which one to choose here, building a data driven culture. And let’s see what Atlanta comes up with.

Katie Robbert 33:40
While it’s doing that, so it just gave you the outline of the hubs and spokes, could you take it that step further and say, you know, can you give me the outline for the hubs and spokes with the target key word for each individual piece of content? Because I think that that I mean, that, you know, for, you know, full transparency, that’s where this conversation internally started with Chris and I putting together content and me saying, me asking the question, you know, well, what’s the target keyword for this? Because that’s, in my mind. That’s not how people search for this particular topic. So how do we make it more SEO friendly, by giving the pieces of content titles that we think that people are actually going to search for versus using all of the jargon that we tend to fall back on and use?

Christopher Penn 34:39
That’s a very good question. So let’s go ahead and ask it. That exact thing. Now, if you wanted to, at this point, I haven’t we have not organized if you want to, you could even add back in the exported CSV from the initial backup. Actually, let’s do that. Let’s go ahead and do that because I think that’s a good To exercise, let’s go ahead and get our overview here. And let’s export our list that we built. I have provided the initial list of keywords and their volumes in the attach file. So we’re gonna go ahead and attach this file, upload it. Find out where that where my desk that went.

Katie Robbert 35:30
Chris is adding these things back in, because he knows that post livestream, I’m going to be asking him all of these questions and ask you to put all of these things together so that I can execute them. Sorry, so that John can execute them. John, you’re getting assignments today? Yeah.

John Wall 35:47
Alright. Getting the heavy copy job.

Christopher Penn 35:53
So now it has this the required keywords based on the keyword volumes attached to the hubs and spokes. And while it’s doing that, I’m going to I want to scroll back up here to where we were just a little while ago with our outline. So this outline page has a target audience content goals, and then your outline. And so what I would suggest doing next for this is with your subject matter experts, whoever they are at your company, give them this outline, give them the voice transcription app of your choice, and have your subject matter expert talk through this and record the content. And the reason I’m suggesting you do that, rather than having gender to AI make it is because you want it to have your unique perspective, as opposed to the generic stuff that the model will spit out.

Katie Robbert 36:55
The thing that I like about this and the thing that we’ve been including that I think a lot of people are forgetting to do, even if they’re using generative AI to build outlines is who is this for? What are they getting out of it? What is the goal of this piece of content? And that to me is something that is so like, that’s again, that’s an SEO basic that in 2024 should be included, regardless of the tool that you’re using is? Who’s the audience for this? So the audience is business leaders. What is the goal of this piece of content? What is it meant to do? It’s meant to educate, it’s meant to provide actionable steps, okay, great. Those are things that I as the person writing, it can wrap my head around and keep in mind so that when I’m building the content, I can go back through and say, Is it educating business leaders? Is it providing actionable insights? If not, I need to go back and do it again.

Christopher Penn 37:52
Exactly. So we’ve also got now all the different hubs and spokes with their various keywords attached to them. So that’s, that’s available now, with this outline, great. This spoke content outline you created for building a data driven culture with data visualization is very helpful. I want you to transform it into a series of interview questions that we can give to one of our subject matter experts to turn into a podcast or podcast series, or a recorded interview. So that we capture their unique specific point of view. And so now, we’re not talking about not just making the content and outline and content, we’re actually talking about converting this into something like you could do on a live stream or podcast, because one of the things that is sort of a foundation principle of modern content marketing in 2024, is don’t do the same work over and over again, do it once in one format, like a live stream, and then repurpose it over and over and over again. So if we wanted to, for example, I could take this and I could have Katie, I could have you be the interview subject and we could start an episode of the podcast say Hey, Katie, can you explain what a data driven culture means to you and why it’s important for organizations today? Right? That is a great interview question. And now you have you created your content for your your hub or your spoke. But you’ve also created a YouTube video, and some shorts for Tiktok and, and Instagram, and content for your newsletter, and audio for podcasts. You’ve done all this work with just one spoke.

Katie Robbert 39:53
So, for those of you watching, don’t be surprised when you see this show up in our podcast next week. But because I mean, that’s the whole point of this is, so we’re using our own information so that we can figure out what are we missing? What is our gap? What are we not addressing? For our ideal customer profile? Why are they not coming to us? Because we’re not answering their questions. Well, in order to answer their questions, we need to create content. In order to make content that is reaching them, we have to have good SEO practices. And so that’s sort of the full circle whole reason why we’re doing this in the first place is generative AI or not, you still need to create useful content that answers questions that your audience is asking exactly what to use, you can use generative AI to help you but you still have to create the right content.

Christopher Penn 40:50
And your cornerstone, your your North Star, your whatever you want to call it, is that ICP that ideal customer profile, if you have that, then you can validate all of this stuff. Because again, SEO is search is great. And getting traffic is great. But the traffic is not your audience that you want, it’s not your buyer, then you just essentially consuming a lot of resources. So at every stage, if you’ve written out your ICP, every stage of this process of any SEO process can you can say, validate this against my ICP, check that my ICP would like this. I asked here, verify these interview questions would also satisfy our ICP. And it goes through and it says yeah, these are these are the things that are aligned with your ICPs needs and pain points. And therefore, this interview that we built, is given the thumbs up, we are greenlit to make this content. And I think it’s really important to say, this is not like a weeks long process. We’ve literally done this in 41 minutes, and we’re ready to take this content framework could build it out more obviously. But then we could start writing the podcast episodes we want to do right from this. If I had a template I do, but I’m not gonna pull that right now. Write a template for making an episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris to person in new show, say I guess then say, hey, take that same spoke content, and reformat as a series of things for Katie and I have talked about on our podcast, and now I can make outlines of podcast episodes that I know I need to make to satisfy our ICP. And, again, it’s the same knowledge just transformed into different aspects. One of the six use cases of generative AI is rewriting you’ve got this content, you’ve got this model of a hub and spoke have to rewrite it, rewrite it as a podcast rewrite as a newsletter called open, rewrite it as this as this as you could go on and on and on, and turn and get and knock down all those pieces in that outline relatively quickly.

Katie Robbert 43:02
What do you think, John? We’re going to overhaul our SEO with this?

John Wall 43:06
Well, it’s always a work in progress, right? I mean, that’s the thing is you’d come up with the fresh content, but everybody else is generating stuff at the same time, too. How about are there any tricks for using AI? For post analysis? You know, after you do a bunch of stuff, can you go back and get an idea for what’s working. And then the other way we hear stuff talking about, but I haven’t seen much going on is actually writing content so that you get right into the Gen AI search, you know, skipping traditional search altogether.

Christopher Penn 43:34
So there’s a bunch of different ways to do stuff like that. But let me show you an example of this. So here is I’m gonna take out this week’s data diaries that we wrote from this week’s newsletter. And I could actually say, Okay, let’s take a moment to validate a piece of content against our ICP. This is content that will go on our website, determine how our ICP would react to it. And what if anything we could do to improve its appeal to our ICP and improve its ability to attract search traffic. So again, depends on whether or not you are doing just content for like, an outreach to an ISP directly, which case you wouldn’t actually need to worry about keywords and things. But if you’re doing for SEO, you want to include that provision. And so I’ve pasted this week’s data diaries in and it’s going to go through and say, Okay, well, let’s see what kinds of things you wrote about and what could be better. I actually have a three page long prompt called Thomas, the critic that will, you know, go in excruciating ly painful detail about all the things you did wrong in your writing.

Katie Robbert 45:01
Here’s what I think’s important to note here. So Chris, you were saying, you know, a couple of minutes ago, like, this isn’t a days long process, we did this in 41 minutes. Because we had all of our foundational data ready to go, we’ve already gone through putting together the ICP definitely took us a few hours to do. Yeah, it was, well, it, the actual document took a few hours. But prior to that, we were doing all of the research, figuring out what goes into it. So I want to make sure that I’m setting the correct expectation is that when, and this goes back to, you know, basic things like, oh, I don’t know, maybe the five P’s, having your requirements, so that you have your purpose, I want to build an ideal customer profile, so that I can write better content so that I can target the correct people who needs to be involved? Well, Chris and I were involved. We got feedback from John, we got feedback from Kelsey, we got feedback, you know, we just sort of took a look at everything. What is the process, so we went through the process, the platforms that we needed to gather data from based on what the goal was, and then, so we had to gather all of our requirements first. And then we can go ahead and build the ICP, similar to using AI for SEO. You know, we’re we’re doing it on the fly. But had we sat down to do it. And because we’ve talked about it extensively prior to this livestream, we knew we want to use AI for SEO for this reason, these are the problems we are trying to solve. And so we’re walking through those steps now.

Christopher Penn 46:37
Exactly. So John, to your question, How did how did I do? Well, I got some, I got some good feedback on on what our ICP would like about this piece of content. And there’s an L by the way, you have no keyword optimization space. So if if we were to post that we’d want to come up with, you know, putting that in there. It’s not really industry specific. There’s no visuals in it, and there’s no call to action. So those would be the things that it ding me for it said, Okay, these are things you could do to make this better. So again, as we repurpose content, one of the things that we could do in the process is have our account manager, Kelsey, run through a piece before it goes up to the website with a series of questions for a generative AI and generative AI would say keywords not so good titles not so good. Go ahead and make these changes. And you know, it can it can add those changes in like code almost. And then now it’s ready for now it’s ready to meet the needs of the ICP. Now it’s ready to meet the needs of SEO. But it starts with something that we already wrote.

Katie Robbert 47:37
Sounds like that’s the process that we’re going to be building for her next week. Well, because that’s, you know, and that’s again, sort of, you know, full transparency, that’s where the conversation started, was, well, what is the target keyword of this thing that we’re building? And the answer was, well, there isn’t one. And then I net, you know, I asked, Well, shouldn’t we be doing that? Is that still a best practice for SEO in 2024? And the answer is, yes, you still need to have a focus for your content. Otherwise, the search engines or the generative AI, you know, models are like, I couldn’t really tell you what this is about. It’s about a lot of things. But it’s not about one specific thing. And so you want to make sure you’re still having that focus. So basic SEO practices still apply. Good keyword research, making sure the you know, content is structured in a way that’s useful and readable and valuable, that you’re having the, you know, the what, why, how segments of your content.

I know we’re coming up on time for this episode. But this is definitely a topic that we could clearly keep talking about for a long time, because it is a good SEO takes a lot of work. You can’t there’s no real good shortcuts. Once you get all this information, you still have to do the thing, you still have to build really good valuable content, you still need to put it into your website or your content management system, and make sure that all of those different pieces are put in correctly.

Christopher Penn 49:17
Yep. So here’s what we’re going to do at the end of the session. Now based on everything we’ve talked about so far, and all the work we’ve done, I’d like to create a prompt written for a large language model like Google Gemini, that will help someone audit a piece of content to understand how well it’s aligned against our ICP. And what could be done to help perform better from an SEO perspective, right in the second person. And so what this is going to do now is going to sum up the 33,000 tokens that we’ve generated in this session so far, and condense it down into a prompt. What you would do with this then is you will then you could do one of two things. You could hand the prompt itself to some like calcium, for example. So hey, paste this in to start a new session. Or you could, if you’re using ChatGPT, you can create a custom GPT. And this is your starting point for that custom GPT. If you’re using Google’s Gemini in their, in their development environment here, you’ll notice there’s a little chick one up here called system instructions, right. And there’s a little little thing here. So what I could do, like pull up a instance of this clean instance. Let’s wait for this to finish generating.

Okay, so it’s finished generating this prompt, I’m going to, I could go into the instance that our team member Kelsey uses and call this SEO auditor. Save, go into System instructions, and paste in that hit Save. Now, this is like a custom GPT. But in Google’s Gemini, and all Kelsey would have to do is say, all this content attached the content and those system instructions that we just boiled down, will stay in the background and run it. So now we’ve built a tool that will make the process more efficient.

Katie Robbert 51:27
I like it. Check out all that John, you write it down?

John Wall 51:31
Yeah, like I haven’t played around with AI studio. It’s great that you can save all these, you know, queries and have them on demand as they go now. Oh, it’s just it just this use Gemini to and that it verifies these responses against search to or does it just… this is what the Gemini 1.5 models?

Christopher Penn 51:45
So yeah, this is the most advanced file, you can see we have a couple of GP T’s that it basically brought over from OpenAI right in here so that I can use it in that context. So yeah, that’s, that’s how you that’s how you scale the stuff a little bit more instead of just being something that someone copies and pastes randomly. Now, it’s you starting to build tooling for people who maybe aren’t as fluent in generative AI yet, but you can you can give it that shot in the arm to get going right away.

Katie Robbert 52:16
Well, and it still covers basic SEO practices, because again, sort of the you know how to use AI for SEO, the topic of this live stream. The point is that the foundational pieces of SEO still apply. It’s how we’re getting to executing them. That’s changed.

Christopher Penn 52:35
Exactly. Looks like we did get a comment. Adam was saying on why roles were congenitally I promise you mentioned, provide the character card in your blog post. I couldn’t find any way. I’ll have to go back and look for that. But if and what we can do is we can put that in our free slack group go to as analytics remarketing, I’ll put I’ll remember what I said. And I’ll put the character card in the slack group. Any final parting thoughts?

John Wall 53:02
User Experience, make sure it’s great. That’s it.

Christopher Penn 53:09
All right. That’s gonna do it for this week, folks. today. Thanks for tuning in, and we’ll talk to you on the next one. Thanks for watching today. Be sure to subscribe to our show wherever you’re watching it. For more resources. And to learn more, check out the Trust Insights podcast at trust AI podcast, and a weekly email newsletter at trust Got questions about what you saw in today’s episode. Join our free analytics for marketers slack group at trust for marketers See you next time.

Transcribed by

One thought on “So What? How to use AI for SEO

Leave a Reply

Your email address will not be published. Required fields are marked *

Pin It on Pinterest

Share This