In this Episode
- [00:30] – Stephan introduces Aki Balogh, the co-founder, and CEO of MarketMuse, an impressive marketing tool that’s every entrepreneur’s secret weapon to competitive marketing.
- [05:13] – Aki talks about the importance of expert knowledge input for self-driving content machines.
- [11:31] – Stephan discusses how thin content is not just small word count, but it’s also about being very surface level and shallow in addressing the topic.
- [16:38] – What are the factors in identifying content gaps using machine learning or AI?
- [22:41] – Aki discusses the approach in reaching the optimum flow of link equity, good context, and anchor context in content strategy.
- [29:07] – Aki explains the term topical authority and how it works in content strategy.
- [34:04] – Stephan explains the acronym UPSYD by Keith Krance and asks Aki which stage of the customer’s journey do they focus on?
- [40:41] – Stephan talks about the AI technology being used by Lumen5, a software that creates videos from existing content.
- [45:07] – Aki recommends Madkudu, a predictive lead scoring software that helps companies accelerate growth using customer data.
- [48:35] – Follow Aki Balogh on his social media accounts, and visit MarketMuse’s website at marketmuse.com to learn more on how you can transform how you research, plan, and craft your content.
Welcome to the show, Aki.
Thanks for having me.
First of all, let’s talk about the intersection of content and AI. Some people, I’m sure, read the articles and kind of Wired and Fast Company and that sort of thing that says, “We’re not going to be writing content anymore, it’s going to be AI that’s doing it for us. So all these people who are content creators are going to be out of a job.” And that’s coming very fast like GPT-3 is out now from OpenAI. Although in private beta, some people have had access to that already. And they’re creating entire articles, essays, ebooks, and things that are not written by a human, and humans can’t tell. So I’m curious to hear what’s your view of the near-term future is in regards to content and AI.
Yeah, absolutely. It’s a great question because it points to, to take it back a couple of steps, the evolution of search and content and how all of this evolved. When the internet first came out, people were just putting content up and putting keywords, and search engines were indexing based on that. It was very rudimentary, then people started hacking that and just pushing out low-quality content at scale. So search engines, Google, in particular, came back and started putting in quality measures to read the content and understand the content of the content or one technology they use is called knowledge-based trust factors. So can you trust this article because is it factually true? And understanding what a concept is and what subcomponents comprise a concept. Or another way to think about it is what questions are people asking to get to the content, and then what do they want to understand once they’re there, and how they navigate that learning journey. That’s been Google’s mission all along.High-quality content drives high-quality traffic. Click To Tweet
Google is just a company, but they are driving the technology of this versus something like Facebook, where they are just pretty rudimentary at this, I would argue. So you had this kind of evolution of machines reading content. On the writing side, in the first, you had keyword research tools and suggestions, “Hey, you might want to use this based on this word, it has more demand or something like that,” that evolved to content optimization software, such as MarketMuse, where you’re reading, or you’re helping a writer understand what’s topically relevant. In order to describe a topic, what kind of components do you have to have in there or should you have in there to have essentially information value. How do you increase the value of information as a human, and then obviously, article generation is just the next step of that. The reason I go through all of that history is because the generation of the actual article is not just like one endpoint. There isn’t just one way to write an article. If you put in a title and have it generate, multiple interpretation steps happen. And there’s either a human operator involved with those steps, or you just press the button and presto, and it spits it out.
Even though there was an article in The Guardian that was interesting, that was an AI writing to humanity like it’s kind of cool. But even that article, they didn’t just put in a topic, hit the button, and spit it up. They generated a bunch of articles, and they cut pieces and compiled it. So there’s an editorial step that happened. So in terms of self-driving content, I tend to tell folks, the difference between self-driving content and self-driving cars is there’s only one way to correctly drive a car to the destination, but if there’s content, any given topic could have thousands or tens of thousands of ways you’d want to write about it. Or maybe more than tens of thousands. There are different angles you can take, and there are different kinds of tributaries you can go into. Of course, for every topic, if you add expert knowledge, your knowledge to it, it becomes more valuable, and it’s that new information. And so the machine can synthesize large amounts of information and do pattern matching, but it can’t bring that wisdom that somebody was doing a job for quite some time and was able to instill.
In a way, it is kind of a discontinuous product or evolution, but in a way, it’s kind of related to the previous layers. And so where we see this kind of tactically evolving is, you’re going to want to use the machine to give you maybe components of an article, and then you can assemble the article from components. We’re not yet at the point where you’re going to want to have the machine just write everything, and then you just publish to the web. The level of AI required for that would be kind of a strong AI. It would have to be artificial general intelligence, which we don’t have to worry about for at least a couple more decades, maybe, hopefully.
At least a couple more years. Although GPT-3 has fooled many people, I think there was a Guardian article published. Maybe this is the same one you’re talking about that was written completely by GPT-3. And the headline of the article, which wasn’t written by GPT-3 was, This article was written by an AI or something along those lines.
That’s right. That’s the same article. It’s great because it gets people worried but excited about the future. But that article was generated n number of times. I think, eight or more times, and then they cut pieces together, and then they kind of edited it. Although the words were technical. Yes, the words were written by a machine, but the humans compiled the article based on the goals that they wanted to achieve, and that worked.
Gotcha. Okay. And so, where does AI fit into MarketMuse? Because I know that’s kind of one of your selling points of the software, is it’s got machine learning baked into the product, and it makes it a better technology than some of your competitors.
Absolutely. To answer that question, I’ll just give a 32nd kind of how did I even get into this because for me, I was mentioning before the show, it’s a huge honor to be talking with you. You wrote the book on SEO, I read that book, and I learned a lot from it. But I did not come from the SEO world. I came, as you pointed out, from the machine learning and investment, and kind of startup tech world in a way. So I wanted to take an AI or machine learning approach to content marketing. I want to do it to help society, I want to build an AI technology that helps society, and I thought content would be a good way to do that.
In that spirit, the first product that we built, which was called Content Analyzer, but now we call it Optimize, basically takes an article and gives you suggestions for what topics you might want to include to make that article more impactful and more information. That first product, we basically invented that, there were other similar prototypes of products that were looking to do similar things, but the way that we built that product and the quality of data and the software worked, and it had not worked for others before. That was our first foray back, I think, gosh, 2015, maybe prototype in 2014, but certainly to market in 2015. To build that technology, we built what we call our Knowledge Graph System, but it’s basically what’s called a topic modeling system. So it’s a system were given a topic, the machine goes to the web, reads 10,000 articles on a topic, distills the essence down, and gives you the human outline of how to cover that topic comprehensively. So it’s a kind of large scale pattern matching and distillation that’s done through topic modeling. So that was our core technology. We now have a patent on it. But essentially, we’ve put a lot of engineering and data science focus into strengthening that.
And then, on top of that, we’ve just built a system of products essentially. So we went from optimizing an existing page to researching new topic ideas to what are the questions that people are asking that are topically relevant, what are the linking connections? What other pages on your site should you link to, to what is your competition doing, so competitive analysis. So those are what I call page-level applications. So for optimizing an individual page, you can use those. And then we went to the next level, well, in late 2018, we launched a system that will analyze your entire site and connect all the topics and pages and things together, and it can show you essentially, the breadth and depth of your topic clusters. In layman’s terms, for a given concept, given topic, how many pages do you have that relate to that topic, and where are you strong and weak? And where do you have the opportunity? And what should you work on next? What topics should you work on next? And what articles should you create that are gaps, like, where do you have content gaps and topical gaps? So that was a whole nother level to our offering.
Now, we had just launched a few months ago, the third level, which is, we’ll generate it for you using AI to fill. Now that we have these outlines, we know what topical gaps you have. You pick some, and we can estimate the value of how much money you would make if you fill those gaps, which is good for getting a budget to resource the project. So then you make your hit list of articles, and then we can generate the articles. And to my earlier point, we have sort of several editorial steps to make sure that the other end of the production line meets the original goal that you want it to meet.
Wow, that’s quite a comprehensive product suite. And so if we were to look at this at the page level, let’s say that you have an article about lawnmowers, and you don’t talk about related topics, what Brian Dean of Backlinko would refer to as LSI Keywords. Which is not the best term, I think, because LSI is not a thing that Google is using. But it’s a way to refer to related topics in a very SEO kind of acronym type of way. So these LSI keywords or these related topics, if you have an article about lawnmowers, and it doesn’t talk about grass, or lawns, or lawn care, or gardeners, or landscapers, or landscaping, or weed whackers, or weeds or gardens or anything else, it’s just lawnmower. A lawnmower, you have a very surface level, thin piece of content. And your early iteration, which has evolved quite significantly, could identify that your articles are thin in that way because thin content isn’t just a small word count. It’s also about being very surface level and shallow in the addressing of the topic.People consume content to learn. When they're more informed, they tend to trust more. Click To Tweet
That’s exactly right. Essentially, all of these acronyms TF-IDF, LSI, LDA, Latent Dirichlet allocation, neural nets, rank brain, whatever all these like technologies that people use, they’re all ways to model how our brains learn. Basically, content is learning. So if you’re looking to buy a lawnmower, what are the types of questions you ask? And it could be the use of it, so will this cut this type of weed or grass or whatever, that’s one thing. It could be the product features, so is it like, is it the one we push it, or is it the one where I sit on it? Or there are lawnmowers that operate with no electricity, maybe in a rural area, or is it gas or whatever. So the product features the value props and then the different audiences. So are you a gardener? Is this for a home? Or is this for commercial operation? And so essentially, you want to cover what you’re about, what is your company about, what are your products about, what audiences do you serve, and what value do you provide, and then how differentiated from other lawn mowers or whatnot. You want to write all of those down. And all the technologies do is just help us get there.
I had this conversation with a CEO of a very large writing network the other day, and he’s like, “Well, I don’t know about this AI, I think this is not right, I think it’s gonna hurt the quality of content,” and I’m like, “Well, the AI is specifically natural language generation technology, which is different from what we were doing earlier topic modeling and content analysis.” Now, generation technologies are just going to be another productivity boost. They’re not going to completely replace humans because as a lawnmower business, you have a lot of knowledge about lawnmowers that you have not put pen to paper. That’s the constraints. So you have a lot of subject matter experts that would love to write articles, but they don’t have time, because they’re like designing lawnmowers and stuff, right? So what if you could take that knowledge and that science and product knowledge and commercial insights and put it into your content and scale? You would need an enabling technology.
Right now, the enabling technologies are agencies where they’re like, “Well, yeah, we’ll do it for you, we’ll write all the best lawnmower stuff in the world,” but they never built a lawnmower. So then you might have an entry-level marketer writing about stuff, and you’re reading, and you’re like, “Oh, gosh.” There’s no way they wouldn’t know, but actually, it’s not quite accurate, and the idea is to inject the knowledge into your content onto the internet because that’s another thing people assume. People assume that the internet has great content; it doesn’t. It doesn’t. When you analyze the internet articles, some content is very strong, like Justin Bieber or credit cards. Lots of credit card search content on there, of course, a lot of money to be made, but there are a lot, like try searching for like, Mongolian cultural reference, it’s just very rare to find. And of course, some of those are because it’s less commercial.
It’s just that there’s a lot of content gaps; there are a lot of commercially viable topics where the quality of content on the web is low. There are housing systems software, SAS warehousing systems is one that I was showing somebody recently, and almost all the content on it stinks because people just haven’t written it yet. But there are a lot of people in the world who know about warehousing SAS, some company software, and there are a lot of people who need to know because they have a warehouse, and they would benefit from learning more about how to optimize their warehouse.
Right. There are lots of different ways that we could go down this rabbit hole. First of all, let’s talk about identifying those content gaps, like how do you figure out using machine learning or AI what the content gaps are. Like, I know there are tools from SEMrush and Ahrefs, for example, that will look at competitors, and compare you with the competitor and see what keywords they’re ranking for and you’re not. So that’s a content gap analysis that can be quite helpful. But I think you’re talking about something a bit more advanced than that, right?
Yes. The way we look at it, you have a couple of factors, you have how much demand is there for topics, how much search volume for a topic, and the related topics and stuff. How big is that market, what’s the competition levels? Is mesothelioma highly competitive, or is Mongolian culture not competitive? So that’s another factor. The third factor that we’ve added is relevance. What is relevant to that topic? To your earlier point, lawnmower, gardener, and weeds are somewhat related, so that should measure into that calculation. It’s not just a two by two. It’s a three by three, or some sort of maybe some other geometric figure. But essentially, once you have that model, that map, you can essentially read all of the pages, all of the keywords, topics, everything you’re ranking for, everything you mentioned in the article itself. Just create this big collection of concepts and then start to connect the dots, and the machine can do this because it works in what’s called a high dimensional vector space. So it can model your entire site in high dimensions in a way that a human brain could not do.
Let’s say you want on your site and 10,000, or even a couple hundred or a couple of thousand dimensions, and it can basically make these connections, and just run a bunch of math and say, “Oh, there you go. That gap, that article is something that you’re missing, but it’s highly relevant. It’s very attractive from an SEO demand generation standpoint. And it’s very core to your message.” And it’s something that you can differentiate on versus competitors, and it’s kind of untouched by others. Write that first, and when you write that, it does better in search, and your CFO is looking at you saying, “Wow, we’re doing so well, in demand, and how are you doing this?” Well, the machine just finds these ideas, and that a content strategist, or SEO kind of evaluates the idea one by one, and just bangs with the content.
Right. So this is as much about creating new content. So the content gap is looking at where you have not written pages of content that you should be, but also, you have existing pieces of content that don’t cover in-depth the topic well enough. They don’t refer to like in the lawnmower example to lawns, or grass, or clippings or yards or yard care, like those are very closely related topics, whereas gardening and patio or deck or landscaping would be a little less relevant or closely aligned to the keyword or the topic of a lawnmower.
Yes, so the way the system does its big crunching thing, and it gives you a list of basically new articles you should write and existing articles you should optimize, and then you can kind of pick your strategy. So some agencies and companies like to optimize, optimize, optimize, not publish so much, others like to just publish and never optimize; some do 50-50, whatever. But basically, you can pick your battles as to what you want to do, plus what are some campaigns you might be running, what are some seasonality things, the different aspects to your business that might be relevant or important at that particular time. Then also you have to think about what level of resource or you’re playing. If you don’t have a lot of authority on-page or off-page authority or topical authority on something, you’re going to have to spend more resources; time, money, people, and so on. You’re gonna have to spend more resources to hit that goal. So you’ve got to kind of do this cost-benefit, but essentially, yes, you make your hit list of new articles to create an existing to optimize, and then you link them together.
Yeah, so how do you interlink these different pages? Because if we were to look at an example, let’s say like the New York Times, they have these topic pages that they can rank quite well. Like, for example, for the keyword “Iraq,” there on the first half-page one with a topic page, which is pretty impressive. And that topic page doesn’t even have a description of Iraq, and like an overview of the geopolitical issues and that sort of thing. It’s just recent news articles. And so if you look at how they’re linking to the topic page internally, they’re not just doing it in a brain dead sort of written approach, they’re doing it in a way that adds value to the user to the reader, as well as to Google. So, for example, they won’t take just the first occurrence of the word Iraq in the article and turn that into a link to the topic page, especially if there’s another link in that same sentence that goes to a much more important recent article about a topic that or an issue, a current event that that happened, that they don’t want to take the emphasis away from that by adding a link to the topic page. And they’re not going to just have a list of tags are topics at the bottom of the article with 50 different keywords, and they’re all linking to various topic pages because that’s another pretty popular but not very effective SEO approach that you’ll see a lot of.Good content providers want to increase the amount of knowledge, information, and wisdom of their readers. Click To Tweet
I’m curious, like what’s the secret sauce behind interlinking these different new pages and the existing pages that you’ve optimized so that you get an optimum flow of link equity internally and good context as well with the anchor text, but not over-optimized, and not too over the top. Let’s talk about that.
Absolutely. I mean, that’s part of the art of SEO and writing for the web. The first place you start is what are the user intent or the searcher intent to your point. When people search Iraq, it seems less likely to think about the touristy things they could explore and more likely to think about conflict, which is, it is what it is, right? We wish it were better, but that’s what it is. So most of the intent Google points toward news articles and stuff. But that may or may not be the case. When people search for a lawnmower, are they looking to learn about lawnmowers and abstract? Are they looking to buy a lawnmower near where they live? Are they looking to compare a lawnmower to a weed wacker? It could be any of those, right? So as a business, you do have to think through all the stages in your buyer’s journey, your different audiences, and then go to the topic layer. And that’s part of the human interpretation that you do as a content strategist.
Once you have your content, and you’re looking to link them, one of the most common questions as well as, “So I have these pages, and I have all these snippets of information, do I put it all in one article, or do I kind of like spread it out? And if so, how?” And it depends. So you do want to the most popular, or I guess the way to do it is a hub and spoke model, where you have what’s called a pillar page, which is typically a long, I haven’t seen it, but probably the New York Times is like a long explainer page on the rack, and then you have subtopics and different places you would navigate to learn more about, I want to say Kandahar, but that’s in Afghanistan. But like what’s the state of the military in Iraq, what’s the state of oil, whatever, and you navigate to those pages, constructing that structure is what you do as a strategist. The machine can give you ideas and say this information should be somewhere, but then how you compile it is more up to you. We can suggest basically, and anyone doing this by hand, but we also do it through AI or the machine.
But we can suggest, “Here are pages on your site that are related to this page.” So lawn mower, weed wacker, weeds; these are pages you should link together because of the topics, and you would drop it in as anchor text. So like, “the lawnmower is rated for this type of weed,” and then you click on weed, and it goes to the weed page, and then I don’t know, it’s another user intent problem right there. But basically, we can surface those latent connections, but then you would pick from there and use that to build out your topical hub in the way that you want to. Basically, there’s a difference between like reading a book with a table of contents, and you typically go front to back, versus the web, you arrive somewhere, and then you bounce around, and then you click on this ad or something, and now you’re on the Twitter page. You just want to follow the journey that you want your customers to have.
Yeah. How do you create this content that explains the topic? And that’s something that used to exist on the topic pages on the New York Times, but they’ve removed it. So there used to be a paragraph or two about Iraq on the Iraq page, and now it’s gone. And it’s just linked to all the different recent articles about Iraq. And I think that’s going backward in terms of value for the user and SEO. But I wasn’t involved in that process; they’re not a client or anything. So I just happened to notice that. So how would you create that description that that explainer copy to go on this topic page?
Yeah, the way that I would think about it is how do you create a Wikipedia type entry for your company or your product or the message you want to deliver? When you search on things, obviously, Wikipedia typically comes up on page one. And they’ve done that because they have a lot of original information and it’s very high quality, and it’s very well-linked. It’s 100% humans doing that, to my knowledge we should probably talk to them. I’ll do that after this. But to my knowledge, they’re just completely crowd-sourced, but they’ve done a good job. But that definition of information, what is Iraq? What is the state of the Iraq conflict? What is the history of the conflict? That’s one way of doing it, which is great, and it works well. But if you are an Iraq researcher or an entity that may be an NGO that has unique information, you should seek to create your own. What is Iraq? Or what is the history of Iraq? Or what is the history of the US military in Iraq? You might want to create that yourself to essentially compete with Wikipedia. And to the extent that you can do it for certain questions, or if you do it well, you can outperform Wikipedia.
Like for us, if you search, what is content strategy, we come up because we wrote a lot about content strategy, and it’s unique and novel, and it’s highly factual information. The last part of what I just said is sort of where we are now with the tech as we’re trying to figure out how to make it more factual. And that’s sort of the challenge with GPT-3 and NLG, and everything and just information today are like, how do we make sure that it’s factually true. We tried to do it by looking at a lot of content on the web, but there isn’t an easier way. But anyway, as long as you have high-quality information that is new in some way like if you just basically take the Wikipedia page and make a poor quality version of it, it’s not gonna add anything to the discourse. But like I said when you analyze these pages, and what’s ranking, and what are the top pages on Google, there’s so much whitespace. There is an incredible amount of stuff that has never been written about because people just haven’t spent the time, and it’s just there for the taking.
Okay, so let’s take the example of what is content strategy. How have you seen that whitespace or your technology seen that whitespace identified this opportunity to create something that adds to the discourse, and what exactly is that novel or unique angle or additional research or perspective that you’ve added around content strategy?
Yeah, for us, it’s topical authority. It’s still kind of a niche term, but it’s starting to gain traction. We invented the term topical authority, to my knowledge, we invented that term about five years ago. Because we’re trying to describe what we are doing and just saying, adding information value to things was a lot of words. Topical authority, it’s what I said earlier about your volume, competition level, and relevance. So from a relevance standpoint, which topics do you have a presence on? For each of those topics, how strong are you, and how well have you covered it? What’s essentially the breadth and depth of your topic cluster, your topical coverage against that one specific topic? And then how many topics do you have?
So, measuring that and plugging that into your formula, your SEO opportunity formula, that’s what we’re pioneering essentially. Then the trick is then also talking about it in kind of non-SEO terms because the reality is that there are millions of people writing content on the web to drive demand and to educate people about the products and services. And that’s amazing. Only a fraction of them are like technical SEOs, and most of them are editors, and writers and they’re making human judgments on things. So, how can we have machine intelligence help them achieve their goals so that they write great stuff, but we don’t have to teach them how Latent Dirichlet allocation works. We’ve threaded that topical authority into these kinds of general high-level explainer pages on what is content strategy.
Got it. Okay. How are these pages doing in terms of traffic generation? Are they getting a lot of visits, or is it just kind of longtail stuff?
It’s always relative. It’s been great. It’s powered our whole business, and it’s made us who we are. We’ve had hundreds of customers, many of them are companies, you recognize many of them even better, or startups you don’t recognize, and they need content to drive demand, and we’re enabling their success. And we’ve been called “the undisputed leader in the industry,” which I’d be happier if more people knew about the industry. That’s our bigger challenge. People don’t know that you can optimize the content through AI, or they don’t even know what content marketing is, sadly. Or they have heard of it, but they’d rather just buy ads. There’s a lot of those challenges that we face. But for our footprint, and the amount of demand and leads, and inbound we need to generate, we’ve been able to do it 100% on content. We’re really happy because the proof is in the pudding.The difference between self-driving content and a self-driving car is that content has thousands of possible destinations upon destinations while driving a vehicle only has one. Click To Tweet
Even if you search many of our competitors, competitor A versus competitor B, we come up because we’ve written a better page about them. Of course, we compare both of them to MarketMuse. So there’s just a lot of knowledge we’ve put out there. When I was just starting a few years ago, I had zero dollars, I couldn’t buy software, but at least I could read up on stuff and educate myself, and then kind of do it by hand. And that’s totally cool, too. So for this segment of the entrepreneur, people starting businesses, they don’t necessarily have to buy what we do. I mean, we now have an always free version called MarketMuse Trial. But even if they don’t use any of that, at least they’ll know how to do this stuff on a higher level, and then just kind of map it by hand. That’s fine, too.
But yeah, we’ve added a lot of kind of original information and ideas to the web. And for us on the business side for a company, we’re not a massive conglomeration. We’re a small company. For the amount of traffic we need to drive, that’s been enough. We’ve been able to just kind of achieve multiple goals, and we drive enough demand to power our sales channel, we drive enough awareness to educate the market slowly over time. I wish it would happen quickly, but it happens slowly, but that’s okay. At least it’s happening. It’s going to continue to happen. It’s compounding. And by writing it, we’ve learned it ourselves by having to teach it. So it also pulls a company together, and people learn up quicker, and they onboard better and they provide better service in our team. So all of that can be accomplished through content. It’s like we’re sneaking education to people, which is great. We want to increase the amount of knowledge and information and then wisdom that comes from that.
Gotcha. And so when you’re creating content, or you’re getting machine learning assistance in creating this content, are you thinking in terms of people in the different stages in the buyer journey that they’re like—I’ll use an acronym I learned from Keith Krance who is also a guest on this podcast—and it’s called UPSYD; Unaware, Problem aware, Solution aware, You’re solution aware, and then finally Decision. So if you’re completely unaware that you have a problem, how do you attract people to your site and sell them your solution? That’s an interesting problem to have, right? If they’re problem aware, but they’re completely clueless about how to solve it, maybe they have symptoms of a certain disease, but they don’t know what the disease is. It’s kind of undiagnosed, or maybe it’s restless leg syndrome, and they don’t know what the solutions are. Like, “Oh, they should have more magnesium in their diet or take a supplement.” So they’re problem aware, but they’re not yet solution aware.
And then solution aware, they know there are solutions out there, they know that there’s topic modeling, for example, or there’s a topic type of explorer tools, but they just don’t know that your solution exists. And then the Y for your solution, and then finally D for decision, where they’ve got all the information and they’re at a decision point. So creating content for those different audiences, I think, is smart because not everybody is at the decision point, and they just want to see your testimonials and case studies and whatever kind of sales collateral that will push them over the edge. Maybe they’re just completely new to this space, and they don’t realize that there’s a better way.
Yes, most companies, unfortunately, and we are where we are, but most companies have focused on that last stage where it’s like buy now, or here’s 20% off to buy, like surely it’s the price.
Free shipping, satisfaction guarantee, it’s like, and people buy what? So to your point, it is those four stages, and so if nothing else, you need to have the consideration content; should it be company A or a competitor and why. That’s pretty important. But if you’re going to do a good job, you need to have that awareness of what is the thing in the first place and why would you use it. And if your legs are moving at night or whatever, maybe you don’t know the term restless leg syndrome, like my legs are moving, or I wake up sore in the morning, or whatever. I haven’t had it, so I don’t know what the symptoms are. But whatever those things are, you might just be searching for weird symptoms, and then you learn about a term. And then, once you can put a name on something, it’s a big step in the learning journey. Once you can name something, then you know what to put in Google, then you can read about it, and the rest kind of flows from there.The machine just finds ideas. It's the human- SEO expert or content strategist- who evaluates the idea one by one to give life to content. Click To Tweet
Right to your point there about restless leg syndrome as an example. I don’t have it, thankfully. But I think one of the symptoms is like if your legs feel prickly and it keeps you up like it’s very uncomfortable. And so, if you’re searching for legs or prickly at night while trying to sleep, just finding that term restless leg syndrome is a huge milestone in figuring out the solution. Because now you have a name that you can search on, you can look it up in Wikipedia, and you can see what other symptoms go with it, and what are some of the solutions and so forth.
Now, let’s talk for a couple of minutes here. I know we’re running short on time, but personas avatars, like defining what those persona types are, and what motivates them, not just their demographics and psychographics but what is their internal problem? Not just external problem, but the internal problem using the terminology from Donald Miller’s StoryBrand Framework, but also then, what are the four forces, the wants, the aspirations, the frustrations and the fears that trouble them or kind of motivate them, so that once you’ve identified those things, you can weave that into the copy that you’re writing to make it more persuasive, and more relatable and inspire them to take action?
Yes. I mean, one is, to your point, essentially, how do you inject empathy into your content and your messaging. Especially when we work with medical content, and we’re not doctors, but we certainly generate and help and provide insights on medical conditions, and then an expert evaluates it, but the software does a really good job of suggesting things that are technically accurate because it’s looking at a mass content. So we have statistical significance helping us with this, but when you write that kind of content, the people who are searching for it are likely suffering from a condition. And if you get the content wrong, or you forget to mention something that could negatively impact somebody’s health outcome.
So that’s a big part of it. When I say content, just because of the nature of my work, I automatically assume SEO long-form written articles, text articles, but there are, obviously, many other types of content. Social content and videos and stuff like that. Although typically, it’s siloed, so you’ll have one team writing the articles and the other team doing the social media, but those wants and desires and fears are reflected, especially on social media. It does come out, so you want to have some line of communication there. It would be too optimistic to ask that you merge the groups. Some companies can do that. So some companies take different topics or products, and they have like little teams of you have one content person, one social person and one adds one this and that along with each topic, that’s smart. But even if you have a small team, just have the kind of talking. Especially fears, right? People react to fears and anxieties in order of magnitude faster than they react to the good news. And so you want to make sure that that’s injected into the overall knowledge base that you’re essentially building on the web.
Yeah. So a great point to get those different silos, different people talking to each other so that somebody who’s creating video content and somebody who’s creating article content can share their learnings and their empathy for the target audience with each other. And so that gets baked into both formats. There are probably some tools outside of MarketMuse that you think are pretty cool. And I’ll just name a few that come to mind that relates to our discussion here. One is called Lumen5, lumen5.com. And that has some AI to it, where it takes an article or blog post, and creates a social video out of it, that kind of distills the key points from that article, put those key points into different frames, and then adds imagery or stock video to that with music so that you can watch this video that was generated automatically. But of course, it doesn’t publish it until you go in and you tweak it and say, “Oh, actually, that’s a little bit still too verbose. Let me cut some words out of that.” “Let’s change the stock video that it chose to something else that I think is a little more appropriate as a metaphor.” “Let’s add a little call to action at the end of the video,” and boom, now hit publish, and it creates this cool video. It took a fraction of the effort. So that’s one great tool. I’m not sure if you’re familiar with that. Have you heard of Lumen5?
I haven’t used them. But I’ve heard of that type of technology. Yes.
Yeah. So that’s one, and then there are several free keyword research tools that help you to identify questions that your audience is asking. There’s an answerthepublic.com which is based on Google Suggest. And then there’s also asked.com. Have you also tried Asked? Are you familiar with that tool?
I haven’t tried it. But to be honest, we have one, and it’s also free now called MarketMuse Trial. I mean, you can run seven queries in it a month. So it’s a little more limited, so I just use our stuff.
Yeah. Answer The Public, it’s pretty cool, too. It was an inspiration.
Yeah. So I like both of them, but they’re just different because also Asked is based on the People Also Ask Box in the Google results. So you’ll get this cool visual that you can save, you can also save a CSV file, and you can just click on one of the questions in the flowchart thing or the, I don’t know what kind of graph it is, like a tree graph or whatever. And then it will reorient to that question being the center point, and then all sorts of additional questions come from that. It’s cool, and it’s free. So that’s another tool that our listeners might want to check out. And then finally, TubeBuddy, which does AB split testing of titles and video thumbnails for YouTube. And so I’m curious if any of this has spurred on any particular favorite tools or things that the best practices, questions, topics of whatever that the listener should think about as they’re doing their content planning?
Yeah, absolutely, there are quite a few. In fact, in a few weeks, I’m presenting at a CEO forum about just showing our entire business architecture stack and all the things that we do, and all the layers. There are so many layers, so many great app software platforms out there. Two that immediately come to mind that have made a big difference for us; one is a community, the Marketing AI Institute run by Paul Roetzer, and they’ve done a great job at basically just defining what types of AI exists in marketing. Content, video, social, they gather all the software platforms, and they kind of explained what each one does and how do you use them. It’s just a great community. I don’t know if they have a Slack community that would be valuable. We have a Slack community for anyone interested in content strategy, of course, shameless plug, but Marketing AI for like high-level of what are all the things out there that AI is doing, that’s one thing.
Another platform that we found helpful, there’s one called Madkudu. They build a predictive machine learning model for lead scoring. When you write a bunch of content, you have all these inbound leads, but they’re inbound, so you have all different types, big companies, small companies, random rolls, whatever, and geographies. And so they all go into your Salesforce or your CRM, but then you don’t know like of the thousands of people who do I talk to first and in what order. So what Madkudu does is they run a model, and they look at who’s more likely to convert, who’s more likely to buy something, who’s more likely to spend more money with you. They run different analyses, and basically, they give you this API, and you plug it into Salesforce. And so when a lead comes in, it just spits out a number like one out of 100. So like, “This is an 80, chase after them, they’re really important.” “This is a 40, don’t worry about it, just let them just kind of nurture and come back.” So they do the heavy lifting to figure out the lead scoring, which is, for us, it’s magic.
That’s super cool. Is that a paid tool, or is that free, or freemium, or what?
Paid. In fact, probably in the, I don’t know, thousands or low tens of thousands, maybe, but they should have something free, I would hope. I’ll reach out to them because they are so complementary to what inbound that people should be able to access it. So I haven’t looked at their site in a while, but it’s worked for us.
That’s great. That’s very cool. So if somebody wanted to try out MarketMuse Pro, do you have any kind of special deal for them? Or how can folks do the MarketMuse Trial free version, and how can they do the pro version? And yeah, what’s the next step for them?
Yeah, totally. It’s easy, go to marketmuse.com, up on the top right, you can click Get Started Free. Sign up there; that’ll drop you into our “always free software MarketMuse Trial,” basic freemium. It has training, and it’s got a great tutorial, a bunch of training, it’s got a MarketMuse Academy for like 201 and 301 level concepts. So you can just get started using it there. If you enjoy it, and if you want to analyze your entire site, you can put in a credit card, and then you get one month free of MarketMuse Pro, or now we have Plus, which is kind of a lighter version of Pro. But either of them basically will analyze up to 500 pages on your site, giving you insights on it. There’s a promo code for one month, which will give you one month free. By the time you probably hear this, we’ll probably have a slicker way, so you don’t even have to do that. But anyway, if you need a promo code, you can either chat on drift we’ve got chat or email me. I’m at Aki@marketmuse.com. But anyway, you should have a one month opportunity to try it out. And after that, it’s either $179 a month or $499, and it could be $325 and then annual plans. So it’s just a couple hundred bucks a month to analyze a small site. And then if you have a very large site, we work with a lot of major publishers, a lot of major brands, especially financial services, and enterprise software both need a lot of writing about your stuff. And so we have Premium packages available.
Okay, awesome. And I’ll include the promo code in the show notes for this episode as well. And if folks wanted to follow you on social media or connect with you, you gave your email address, but what’s your preferred social platform that you’re most active on?
Yeah, my Twitter is Aki Balogh, the company’s Twitter is MarketMuseCo. And I’m also on LinkedIn. I try to crunch through email often, so email is good, but of course, it can get blocked sometimes. I have an AI filter in my inbox that I should mention too. SaneBox. So sometimes, it accidentally filters out something important. So LinkedIn all of those, Aki Balogh. Luckily not a lot of people are called Aki Balogh in the world.
Are you the only one?
Well, the reality is Aki is a nickname. My legal name is Akush, which is a word that comes from Turkish. It’s just hard to pronounce, so I go by Aki. And then, because it’s a nickname, not a lot of people professionally use that name in Hungary.
Got it. Okay.
So I might be the only Aki Balogh, as far as I know.
Awesome. Well, thank you, Aki. This was great. And I hope and I think it will have inspired our listeners to think differently in terms of their content creation and content optimization strategy. So thank you for that.
Thank you for the opportunity.
- Aki@marketmuse.com – Email
- LinkedIn – Aki Balogh
- Twitter – Aki Balogh
- Market Muse
- Twitter – Market Muse
- Facebook – Market Muse
- LinkedIn – Market Muse
- Keith Krance – previous episode
- A robot wrote this entire article. Are you scared yet, human? – The Guardian Article
- StoryBrand Framework
- Answer The Public
- Google Suggest
- Marketing AI Institute
Your Checklist of Actions to Take
Think about the searcher’s intent before starting with content creation. This enhances the relevance of my posts among my target audience. On top of that, consider different types of user intentions. People use the Internet to learn, research, and buy.
Whenever I’m about to write, ask myself, “what are people asking?” “What are they thinking?” The answers will serve as my guide on how to choose the best angle for my readers.
Be more proactive with my planning and content marketing. Create a spreadsheet of content ideas and be dynamic in the way I present them. Aside from blog posts, infographics and videos are great examples of compelling content.
Learn to take advantage of data science. Optimize my business’ operations based on data gathered in and outside of my company. With enough information, business decisions will become more precise and less wishy-washy.
Research more on AI and Machine Learning. Some of its features and tools may be crucial for my business success, primarily when my company relies significantly on my customers’ data and the digital industry.
Run a competitive analysis. Finding out what my competition is and isn’t doing will give me a better edge on filling the gaps and becoming the best option for my target market.
Don’t be surface-level. Add depth to my content and make sure I always give value to my audience in everything I do.
Be consistent. Create a trustworthy brand and stand by my mission and vision no matter what. Let it reflect everything my business stands for, starting from the top down to my team’s last member.
Evaluate my metrics. Run analytics at least quarterly to see how my content marketing strategy is doing. If something works, keep doing it or find ways to improve it. And if something doesn’t work, find another way.
Check out Aki Balogh’s website, MarketMuse, to learn more about creating content that gives businesses a competitive edge.
About Aki Balogh
Aki Balogh co-founded and is CEO of MarketMuse. Prior to MarketMuse, he worked with the CEO of InfiniDB, a Series B startup that made an analytic database for Big Data engagements. Before joining InfiniDB, he was an Associate at OpenView Venture Partners where he looked at Series B investments in Big Data & Machine Learning. Prior to that, Aki was a software developer and a management consultant in the Data Science practice of Diamond Management & Technology Consultants (now PwC Diamond Advisory).