How Programming and AI Can Improve Your Content Strategy [Podcast Episode]

Amanda Milligan
By Amanda Milligan
September 15, 2020

Programming and AI can be very intimidating concepts when you don’t have a technical background.

 

But this episode is about how it’s more accessible and useful than ever.

Listen to a special episode in which Hamlet Batista, CEO of Rank Sense, and Kristin Tynski, SVP of Creative at Fractl, interview each other about how they came to love Python, why they’re passionate about the democratization of programming and AI, and how listeners can utilize these resources for themselves.

 

EN_Google_Podcasts_Badge_2xUS_UK_Apple_Podcasts_Listen_Badge_RGB

Enjoy the episode? Continue listening to Hamlet and Kristin's discussion in next week's show! And subscribe to the newsletter to receive a bonus video of even more tips from their conversation:

 

In this episode, you’ll learn:

  • How non-technical marketing folks can use this technology to improve processes
  • How this technological revolution opens up countless opportunities for thought leadership
  • The importance of related skills, like communicating with developers and other technical teams
  • Tips on how to not give up when you face challenges in your learning

Relevant links/resources:

Transcription:

Amanda: Hey everyone. Quick note about this week's episode, it features two guests who interview each other, Hamlet Batista, CEO of RankSense, and Kristin Tynski, the senior vice president of creative here at Fractl. It's all about how AI and programming can dramatically improve your content strategy. If you don't know anything about AI, don't worry, you'll still get a lot out of this episode; I don't know anything about it, and I found it fascinating. If you haven't yet, be sure to subscribe to the podcast because this is a two-part conversation over two episodes. All right, let's kick off that intro.

Intro: Welcome to Cashing in on Content Marketing. Each week, marketing experts will explain how to measure your content marketing results and communicate that value to stakeholders. I'm your host, Fractl marketing director, Amanda Milligan.

Kristin: Hamlet, I've been aware of you now for a couple of years in the, although I'm sure you've been around quite a bit longer, but I know your presence say as a thought leader in terms of utilizing programming languages, and especially Python, for applications in SEO and content marketing, I've always been really impressed with the work that you've put out, how on top of staying on top of state of the art you've been, and looking for new and interesting applications of machine learning algorithms and other automated processes for improving content and SEO. So, I'm just kind of curious to hear a little bit about your background and how you came to understand the implications of these new technologies and the democratization of programming languages, especially Python, in terms of content marketing, and SEO, and anything you want to say about that, I'd love to hear it.

Hamlet: Yeah, no, absolutely. You know, and, thanks again for having me here, it's really a privilege and likewise have been, you know, following your work as well as and I think it's been really inspirational. You know, combining the creative aspect of PR with data science and programming. And to talk a little bit of my background, right, so, my background is technical. So, I'm an engineer by training. And I've been, you know, by lock or something and stumble on this marketing world, in 2002. So, I, you know, started as an affiliate marketer, and one of the things that if you start in that, in the performance marketing world, is that it's less about the creativity and psychology and more about direct response, it's more about what works, right? What are the techniques that work? And you know, that's the hardest part, you know, in business, that's the hardest part in marketing, right? Figuring out something that works, right? You can try 100 things, maybe two, if you're lucky that works, right? So, it's less about what works. So, one of the things that I, when I started as an affiliate marketer is that, I will try something manually, like, you know, I will play, "Okay, now I'm a marketer, let me play the role and try to go into forums, learn more of the tactics, trial and error, finding things that are effective.", and then when I found something that will work, even if it's a small improvement, I said, "Wow, this is a small improvement that I accomplished here for, you know, if I'm just doing everything manually, it wouldn't make sense to scale it because it's not-- the ROI is not there for me to do the investment because the gain is so small. But if I had made it, wow, that is something because my cost is-- now, that is the ROI I can see with the automation.", right? So, I've been doing the automation since you know, I started playing in the marketing world, with a mindset as, find something that even at small scale accomplishes gain. So, say, "Okay, this, a gain, this value here, but it's so, it's like an amber, right? It's an amber, it's very small but, you know, I've scaled it, now I have something that is valuable. It's a concept of aggregation of marginal gains, right? Are you familiar with it?

Kristin: Yeah.

Hamlet: So then, that is kind of like the framework and the mindset that you said, "Wow you know, I'm privileged that I am able to go into worlds.", right? So, I have education in engineering so, I see how I can, you know, scale something. And then, I have the, been involved in the industry for so long that I understand all the fundamentals and the basics and how to do the trials and errors right. And when you do that, when you're just trying to scale things, right, you know, I run a lot of different languages but Python ended up being the one that I stuck with because you know, I have different roles to play, right? Run a business, I have to, you know, I'm not writing code all the time or doing, you know, just things. So, I will write something and put it in and then I get into something else, sales, marketing, operations, you know, all these different things, you have to run a business. And then, I go back to the script that I haven't touched in months and when I was doing that, even in Perl, which is one of the first languages I was playing with for script, I wouldn't remember it, I couldn't even get it to work without you know, going through a tutorial again, I said, "Wow, you know, I forgot about all this basic things.". When I wrote it in Spring Python, I got it right away, I would say, "Okay. I understand why I did it.", I 'd get into it immediately. So, that's why, it's by elimination that I ended up with Python because I'm writing all these different things, but I couldn't remember, it was impractical if I am not doing it consistently. 

Kristin: Right. Yeah, it's really interesting to hear how you came from a technical background, had a kind of a scientific approach of trial and error and then finding those pieces that were scalable through Python or other programming languages or automation. It's sort of the opposite for my story, because I came from a content marketing, content creation, creative background, also always trying to find those trial and error pieces that sort of gave us a leg up or improved our processes that we could build on. And then, you know, over time, realizing that a lot of those pieces could also be automated. And although I am not, you know, a long-term programmer, I've taught myself Python over the last couple of years and it's been, in my opinion, kind of a godsent because it's such an easy programming language to learn. And once you have even just a basic amount of information, you can then start to leverage this huge amount of community contributed, you know, open source, Python libraries that have been built so, that you don't even really have to build things from scratch, you can just kind of piece things together and create new things that can significantly accelerate those, you know, those trial and error learnings and make them scalable in a way that wasn't ever possible before. So, I'm really interested in what your thoughts are on how the last technical content marketing and SEO pieces of the community will increasingly be able to use these sorts of technologies to improve their businesses and improve their processes, and sort of what you see as the future of like, the democratization of these scalable technologies.

Hamlet: Yeah, absolutely. And let me, you know, fire back with a few more questions to begin about your experience because I think that's interesting, right? For me, it's because you know, I come from a background that a lot of this stuff is not super hard, but I see, you know, what were your initial challenges, right? So, I get a lot of people asking me about, "Oh, how did you do this?", and for some reason, I found that some, you know, content writers that are getting into the programming, and they find that a lot of the things that they can map from Excel, right? Because you're used to Excel and use the functions and all that stuff, right? But some of the things that are not native in Excel like loops, right? Looping through a sequence, they didn't, you know, it doesn't come to them naturally and then, one of the things when somebody asked me, I said, "Look, it's simple, just print everything that has happened, you know, don't assume that things are working the way you think they are. Just print them. And if the loop has 1000-- if the container, the list has 1000 elements, just limit it to 10 so, you can see that it is doing what you're expecting.". And something as simple as that, that I see, "Oh, that's a natural idea.", you know, naturally verified everything piece by piece now that I see that everything works, I do the assembly and put it together. But--

Kristin: I think for me like the breakthrough that allowed me to get over that hump of understanding some of the fundamentals and the pieces of Python that were confusing was using either like a Jupiter notebook or Google Colab environment, because it allows you to run just those small individual pieces of code and see the output and debug those small pieces of code as you go versus, having to write like a whole script, and then run the script and then be very confused about what the error was and then trying to figure out where the error was and what was causing it. And so, once I started using Jupiter notebook or Google Colab, I really started to get the hang of it and understand the errors that I was making and figure out how to troubleshoot in ways that were more intuitive. So, that, for me was one of the biggest things and that's a pretty recent advancement, in terms of programming, that I think makes it a lot easier, you know, just going beyond doing tutorials and things like that, the trial and error of writing your own code and testing each piece of it makes a huge difference.

Hamlet: And in terms of the motivation question, how do you, because you know, as I said, you know, as a program, right? We geek and we're, you know, just, there is excitement in doing the work and figuring out but for most people, and I like to compare it to the video game, right? So, you know, if you love good video games, you know that there is a frustrating, your game so impossible, right? But there's the reward when you complete it, you say, "Wow, I did it.", right? And you sort of get a little bit of-- and for me, it becomes addictive because it's like I'm playing a video game, right? When I'm trying to solve a problem. But if you don't have that mindset, right? If you're not, this is not a joyful thing, you're more interested in the output, how do you hack yourself to stay motivated through the frustration when you have a roadblock or you don't know how to get past it and you'll keep googling for the answer? 

Kristin: For me, I think it's just been about excitement of trying to get whatever the output is that I'm trying to get. So, like a lot of my motivation for learning Python came from, there's a site called Paperswithcode.com, which is, it's basically just an organized list of the state of the art for pretty much every preprint AI paper. And it's sorted by trending so, you can often see like the new state of the art stuff that's coming out and what the capabilities are of it and you can read the papers or you can read blog posts about the papers and understand potentially what the implications are of some of these new technologies that are emerging like literally on a weekly basis, there's some new, amazing state of the art thing that has implications across many different areas. So, for me, probably the first couple were some of the like, really early language models like word Tyvek models and then later on the transformer models, that to me were just incredibly fascinating because, well, first of all their capabilities, but second of all, because I feel like they are the first in maybe what will eventually become like a general artificial intelligence. And maybe that's jumping ahead a little bit too much, maybe we're not really quite there, although GPT-3 is blowing my mind on a daily basis. It just sort of felt like there's this existing superpower out there that not a lot of people knew about, and I wanted access to it. And so, working through the problems of how to, you know, how to implement somebody else's code and get it to run and get the output, whatever that was, whether it was, you know, a text generation through a transformer model or some computer vision thing or, you know, trying to train a model to recognize intent in search phrases, or you know, a bunch of other things that I found interesting, it was really just about trying to figure out how to get it to work and get the output that I wanted, and then see how that output could be utilized toward either writing an article about what the implications were or actually put into use into some aspect of our processes at work.

Hamlet: Oh, that's really fascinating. Yeah, and that's my favorite place, you know, Papers with Code is an amazing resource. Yeah, and then, to tell you a little bit about what I've seen on that is that, it's what you mentioned, it's a superpower, right? So, it's about how do you do what you're doing, you know, how do you believe in things you didn't think you were possible? Right, that's the fascinating thing, right? And I think, you know, thinking about, you know, the way that I see the opportunity with marketers and content writers is that we are-- we're living in incredible times, right? So, how many times, you know how many eras in the past or people in the past will say that you could do something that could affect millions of people, right? How many resources would you need to make, you know, a difference, right? And now, we have the opportunity to do that, right? So, we have the platforms that are accessible to anybody, right? So, you could come up with something, you know, some of the clever ideas that you know, that you've seen and the articles that you're putting out, you can impact a lot of people because the democratic station of great ideas is what I'm seeing happening, right? And that can be used for good or for bad, right? And at the end of the day, marketers, you know, content, you know, producers, we can write the best content and if it's not getting, you know, attention, it's not worth much, right? 

But one, we can actually have the tools in our disposal, you know, superpowers that you can affect, you know, a lot people, you didn't even realize, right? Think about last year, you know, I started this Python movement before I was doing all this stuff, you didn't hear about me because I wasn't in public, I wasn't doing stuff in a public way. And I wrote that article about, you know, for a certain journal, you know, now 25,000 views, that article, right? I said, "Wow.", you know, that was the first thing I probably should, because I said, "Look, you know, this is a way to do something.". So, now, people that have ideas that can make, can benefit other people, right? Can build incredible brands, movements, and, you know, drive something that people should pay attention to so, that's what I think, the same approach that I took when I started this, riding this Python wave and started consistently doing that stuff, that's something that can be replicated in a lot of different industries in marketing, because of this gap that we talked about that you know, people that have the marketing knowledge, the understanding and the content, they start dabbling into the technology side and start looking at the state of the art, start looking at, "Wow.", you know, this stuff, that state of the art paper with code that you mentioned, this is stuff that is not in any product, it's in any product since research, right? That's research material that nobody has tapped into and especially the part that I love about that is that, these are academics that are producing this, they don't know the business value of that yet, they don't know how to turn that into a business value, they don't know how to turn that into content or marketing or anything. And a lot of people that are being educated about it, don't know either, right? They don't know how to connect because of a siloed situation of academics, scientists, engineers, marketers, salespeople, businesspeople are siloed, and that opens a lot of incredible opportunities, because of that gap.

Kristin: Yeah. That's an amazing point and throughout my whole history as a marketer, and someone who's written for Fractl and tried to do thought leadership posts, when I try and think of something I want to write, I'm not really ever satisfied unless I feel like it is at the intersection of what's known, and what's completely new. And I think it's really for that reason that I've gravitated so much towards this explosion in new open source code and technologies. Because it just seems like this completely open space where there are so many possibilities and exactly like you said, there's not a lot of like cross pollination between these different siloed areas, you know, the academics, they understand some of the implications within their little sphere but they're not, you know, they're not marketers, they're not content writers, they're not journalists. They're not thinking about how the code that they've written and that they've made publicly available for anybody to use and to play with could be applied to X, Y, Z situation. And a lot of those implications, just by virtue of being someone in the content marketing, SEO industry, you can come up with those ideas relatively readily, and I think that's especially important now, when you know, 10-15 years into content marketing and SEO, a lot of the basics of what there is to be said has been said, a million times and there's not a lot of low hanging fruit, so to speak, in terms of coming up with new and exciting ideas that aren't connected to these-- this new technological revolution that we're seeing in open source code and in machine learning, and in the democratization of code through Python and through tools like Google Colab and Jupyter Notebook. So, it's an amazing time to be a marketer, it's an amazing time to be a content writer, it's an amazing time to just be somebody who has likes to tinker with these new technologies and so, what you're saying really resonates with me.

Hamlet: Yeah. And I have a question for you. So, you know that there is this technophobia in marketers, right? Because I see that, I talked to a lot of them, and you know, how do you overcome that, right? So, you said, "Look, you know, there's all this exciting stuff you can do with combining your skills with technology, and especially because things are getting easier.", right? There's a lot of free material, paid and high-quality information about how do you become this? How do you learn enough so that you, and one thing that I want to mention is that you don't need to become an engineer. You don't need to spend months and a year to go through a degree and learn every little detail of the language or every possible use case, which is the path that engineer will take, right? If you're building professional applications, you only need to learn enough to be a tinker, right? So, how do you know enough so that you can put together a prototype, like, connecting code from different places, and even get it to a place that you can describe it in enough detail that a coder will turn your idea into something that is, you know, useful, right? Because that also requires skill, if you don't, I was just talking to a new client and he was like, I told him, "Oh, this idea that we're going to say.", he says, "Look, no, I went to the IT team and I said, I asked, you know, what you told me, and they were staring at me, like, I was asking for something that they were saying, "What are you talk about?". So, he says, "I need your help to tell me, how do I put together the BRD for that, right? The business requirement document.", because that is also a skill, you know, being able to describe things that you're, you know, what you're imagining, or even know what is possible in a way that your developers or IT people can do, it's also a skill, right? But how do you get past that technophobia of, "Oh, you know, I don't want to learn this.", and you know that exists in, I see it here often. How do I get past that? So that, at least I'm comfortable enough to play a little bit with it and if I don't want to be the one doing the writing of the code, at least I know how to describe it, you know, correctly.

Kristin: Yeah, I think it's hard because it is really intimidating, especially at first. And you make really, really slow progress when you're first learning a lot of things that are just kind of inherent knowledge that you get as you build up some experience you don't have when you're first starting so, it just it feels very confusing and frustrating. I think doing the basics of tutorials is really important to just get that base level understanding of how the pieces fit together and like what the syntax of the language is, once you have that, then I think you need to find a project that you feel really excited about, where accomplishing it has like a really concrete benefit. Either you find it super interesting, like I did trying to get some of these transformer language models to work or you have an idea for the thought leadership piece, you know, using code in some way that would make the process much easier. And that for me has taken me through like the painful parts of trying to figure it out, piecing things together, troubleshooting, you know, learning aspects of Python that were confusing, getting to slightly more advanced stages with it. But I'm by no means, you know, I would qualify myself still, as a beginner Python user, I'm constantly googling everything, you know, I don't have like a great memory for, you know, not having to look things up numerous times. 

Hamlet: Well, I'll tell you, I've been doing programming for 20 years, I still Google stuff up, Stack Overflow is my friend. I mean, you're not expected to be memorizing all this stuff, right?

Kristin: Sorry, that's the other amazing thing really is that, just by googling whatever the error is that you're getting, more often than not, you can figure out what the problem is.

Hamlet: Exactly you get multiple options to try and then, it's a learning process. Would you say that, thinking about it, like when you were learning to ride a bike is a similar analogy? It's a good analogy to explain, right? So, it's painful at first, you know, it's very, you know, it takes a lot of practice to master it but, you know, once you start, once you manage the basics, right? You can do a lot, you can find places, you know, go to places you couldn't go, you know, walking or even riding in a car, right? Because there are these corners that you can find. So, I started using that and I use it to explain it and people say, "Wow, that makes sense.", right? So, it's going to be tough, you're going to fall a few times, you're going to have you know, it's going to hurt initially, but once-- it builds muscle, once you build a muscle to know, you know, how to go through the basics, it's incredible what you can do, right?

Kristin: Yeah, that's definitely true. And I think one thing that people should keep in mind as they're doing these things is that, there will be times where you're trying to get something done and you realize it's just beyond what you're capable of, and it's been maybe two days of working at it, and you know, you just have to give up on it or get somebody else's help that knows a lot more than you, and that's fine. Like, I think previously, maybe I felt like that was a little bit of a waste of time but over time, I've realized that all of those failures and frustrating places, they've allowed me to do more and more advanced things, and to be less frustrated and to do things more quickly, later on. So, it's not wasted time, even if you don't accomplish what you were setting out to accomplish.

Hamlet: Yeah, no, and I tell you, you know what? You only complain a couple days, I have stuff that I tried for weeks, some, I even tried you know, over a period of a few months, so I gave up and I said, "Okay, I'm going to try it, you know, in a few days or weeks.", and I come up with a nice new approach. So, I wrote an article for the Bing Webmaster Blogs recently, you know, which is quite an accomplishment for me, but the initial idea that I was trying for that it was about learning 10 classification from within Excel. And that was so difficult, it was using the Transformers, using JavaScript, you know, I was very, it was so ambitious that I was trying to do that, I went through a lot of different, probably like 10 different libraries, I bugged out first when I got stuck because a lot of stuff I was trying to do appears it was not going to be possible. So, I was embarrassed because that was three, four months, they say, "Where is your article?", I said, "Look, I'm going to give up and I'm going to try something completely different.". The article that I published, it took me about a week to put it together and I said, "Look, I'm going to do a different one, because this other the one, I couldn't, you know, I got stuck with a lot of different things.". And then, like a month later, you know, I figured it out and I got it to work, and I say, "Oh, you know, I learned.", right? And part of the problems were some things that were not documented about, you know, common API from within Excel; it was very, very tough. And as you said, I said, "Wow, I feel like I wasted all the time.", I said, "No, I didn't waste a lot of time because I learned things that I can use.", you know, TensorFlow, j. s, right? You know, some of the other things that I didn't, and I had to get really granular in understanding the, even encoding of the words the BERT model was using so, it was insightful, right? But it's what you're saying, it takes-- a lot of the articles that I publish, you know, some I didn't get to complete, you know, didn't get to what I wanted. And in every one of them, I learned something, right? I learned something. And the other thing of what you mentioned is, you get better over time. So, I found that it takes me less time, it's easier and faster to produce new stuff because you're building that skill set that is valuable.

Kristin: Also, I feel like in terms of the errors that you see, it's sort of like everything else, there's an 80-20 rule, like you're going to see a lot of the same errors over and over and over again, and once you've encountered them, and you know how to solve them, it just makes the whole process a million times easier, because you're not, you know, you don't have to stop and take a half an hour to figure out what's going on, you just know. And I think that compounds and builds on itself a lot more quickly than people realize it will, and that's what that learning curve really is all about. It's like, it's really tough in the beginning, but things compound quickly and then you get to a spot where you have an idea and maybe if you don't know exactly how to do it, you know roughly how to do it and you have some idea of what problems you might encounter in trying to implement it, and it gets more fun during that time as it gets less frustrating.

Hamlet: Yeah, no, I agree. And let me ask you a little bit about your ideation process by your creative process, because I've been really impressed. You're taking some of my articles or come up with them, you know, it's very clever the ideas that you come up with for this content and even we haven't even, got some early feedback on some of the articles that you have published ahead of time. So, how do you come up with those such powerful ideas? That, yeah, they stand out, right? If you think about, you know, you're coming up with new information and combining, you know, this research that you're doing, for example, the one that we're reviewing about the YouTube videos, finding, you know, their optimal thumbnails, you know, so, do you start with the research or you start with the idea, and then try to plug into the research?

Kristin: Well, I think I have, I think part of it comes from having a lot of experience in this industry and writing in this industry for quite a long time, so I'm aware of what's been done and what hasn't been done. But then, the idea, they come from, yeah, from the code and from the papers. So, like I said earlier that Papers with Code site, I check it religiously, and also some subreddits. So, media synthesis subreddit is a really interesting one, language technology subreddit, machine learning subreddit, a few others, but it'll often you know, when there is a breakthrough, when there's a new paper, when there's something really interesting that came out, I'll usually become aware of it pretty quickly and then, I'll try and dive into it and think about it critically, "How can this technology, you know, if there is open source code associated with it, what applications might it have in any different area that I'm aware of within content marketing, or SEO?".

Amanda: This is Amanda jumping back in again. So, like I said, that was part one of the conversation between Kristin and Hamlet. If you want to hear more, check out next week's episode and subscribe to the podcast so that you don't miss it. Next week's episode is going to zoom out a little bit and look at the overall impact that developments like GTP-3 and other AI can have on search and content online in general. It's really fascinating, and I highly recommend you check it out. Till then.

Join the Fractl community and get marketing tips directly to your inbox!