Editor’s note: Alexander Ferguson is CEO and founder of Raleigh and Charlotte-based YourLocalStudio.com, which is partnering with WRAL TechWire to present a series of in-depth interviews with thought leaders in the field of Artificial Intelligence as part of the firm’s new series UpTech. Today’s package includes a video of an interview with Brandon Anderson of Ceralytics and a full transcrpit of that discussion.
RESEARCH TRIANGLE PARK – Artificial intelligence and machine learning – these emerging technologies are changing the way we live, work and do business in the world, for the better. How is AI actually being applied to business today, though?
Let’s look at content marketing.
This week on UpTech Report, I share my interview with Brandon Anderson, from Ceralytics, whose intelligent platform basically tells a marketer exactly what content they should be creating to generate more traffic and sales. I ask him, in the interview, why can’t someone just do it manually, and also how the technology behind it works.
Let’s jump in.
- With me here I have Brandon Anderson with Ceralytics. Man, thank you so much for joining. Why don’t you give me the quick 30 second spiel. What is Ceralytics, what is your product?
You bet. So, Ceralytics is a content intelligence platform and what it really does is takes the guessing out of content marketing.
So, what we’re doing is we’re identifying what topics actually work on your site, what topics your clients and your prospects actually care about, and what topics your competitors are having success with that you need to use on your own website, to get the success that they are, and even surpass them.
- Couldn’t someone say, then, “Well, I can kind of do that with my team already, “and we can just look at our competition “and what they’re doing, we can look at our content, “do the research, so why would I need your product?”
Absolutely, and you can, you absolutely can, and this is something that’s been done by agencies for basically as long as websites have been around, agencies have been trying to get to this information.
And the way that they do it is they usually take some poor summer intern, and they say, “Okay, your job is to read every single page “of content here, and what you’re going to do “is keep an Excel spreadsheet, put the page, “put what it’s about, and then you’re going to make “this whole spreadsheet of every single page that we have.”
Sometimes it’ll be like, you know, a few thousand pages, takes the poor intern the entire summer to do. By the end of the summer, they now have a categorized audit of all their site content, and they start putting metrics to it and say, “Okay, well based off of this.” And they do a bunch of pivot tables, and they do that.
All right, cool, but what they don’t realize then is this audit started three months ago so everything’s already out of date. So, by the time they have actual data, it’s out of date. So, that’s the thing that we’re filling in. We’re saying you could do it, you absolutely can, great. And if you’re small, you’re a tiny company, do it by hand, that’s fine, do it.
Take a couple of weeks, and do it. But if you’re a mid-size company, you’re a larger company and you need to do this at scale, and you need to do it timely, this is the way to do it.
- In today’s age, you have to be able to quickly modify and change because your competition is doing the same thing. So then, question is, obviously, this is automated, that’s how it’s able to happen so fast. You’re able to do all the research and collate it, and provide, then, this feedback, this insight to your clients. How is AI, artificial intelligence, machine learning, playing a role in your product?
Okay, there are two major components where AI comes in.
First, there’s just identifying what topics the page is about. What it basically is it uses natural language processing, and natural language processing basically goes through a page of content, whatever you give it, and it’s going to identify what different things are within the page. So, sometimes it’s just looking for things like nouns and verbs, and all that.
Well, what natural language processing that we use does is it actually goes in and identifies what’s the thing that this it actually about. So, it’s keeping track as it reads through and says, “Okay, I recognize that there a few things “that we’re talking about here, “here are some ideas that are very similar.” And it pulls all that together and says… “Okay, at the end, I think that this was really “about these three or four topics. “Like, that’s what I’m going to put in front of you “and say this is what this piece “of content was actually about.” And, so that’s what we use the natural language processing for.
Now, interestingly, natural language processing has been around for a very, very long time, and it’s really become commoditized. And one of the places where you can get it from, I mean, not to throw any stones here, but IBM… IBM Watson, and Watson is spectacular, and it won Jeopardy, and it made a big impact. And so, they actually have it available where you can just feed Watson information and it will give you information back. And so, for us, we said, “Wow, that makes a lot of sense.”
So, we went with Watson, and that became our calling card for a while. “Hey, we use Watson, we used Watson, we used Watson.” What was great about it is we were getting results back really quickly, Watson’s super fast, and actually very affordable. The bad thing was the results were not great, and for us, that proved a challenge. So, even though you have something like IBM, and IBM Watson, doing this information, they are made to basically handle generic information, it’s generic to them, they don’t care if you’re giving them an encyclopedia page or a website or a bunch of comments from Yelp, they treat it all the same.
And what we realized is that websites are very different and we actually ended up taking the data that Watson brings back to us, and we have to do a lot of massaging, in terms of the algorithm itself, to say, “Okay, we know what web pages are, “and we know how people read web pages, “so before we even give Watson anything, “we need to actually give Watson some rules “on every single page.” And you can imagine how hard that is, because every site is different.
So, we use that process first, and then we get an actual language and then we bring it back. So, there’s like a whole step to do that. So, that’s actually the first place that we use AI. The second step is for your competition. So, for your competitors, we don’t have access to their Google analytics. Wish we did, we don’t. But we still want to give you actual insights about what your competitors are doing, and where they’re having success. Because we don’t have cut and dry place just to pull information from and say, “Oh yeah, this works and this doesn’t.”, we had a big process what we were going through at the beginning of the company of identifying what things actually relate to traffic.
So we were looking at things like, okay, well, what about a key word that has high volume, that they’re ranking really well for? What kind of traffic does that drive? So we built a model for that.
What about social shares? You know, if you have social shares that are brand new, yeah, that’s currently driving traffic, but if you have 2000 social shares on a post that’s 12 years old, or, you know, probably not 12, but anyway, five years old, that’s probably not going to be doing much for you right now. That’s not a post that’s going to be driving traffic right now from social, so that social signal is kind of antiquated. We don’t want to pull from that.
So, we basically have to teach this thing and say hey, depending on when and where things are shared and where they rank for certain things, and all this information, we then say, “All right, based off of that model, “here’s how much traffic we think this particular page, “or in our case topic, is then getting.”
So, those are the two places. Natural language processing and then looking at the competitor data.
- Thank you again so much for sharing this great insight. For more information, where can people go to and who’s the best type of clientele for you?
You can go to ceralytics.com, you’ll find us there. And for our clientele, we range everywhere from agencies work with us to help with their process. We have a lot of success with B2B companies. Anybody who has content that they are creating to become thought leaders, we are the tool that you need to identify what you should actually be creating, when, where, and why.
That was just one example of applied AI. Have you seen a company utilizing AI or machine learning to transform the way we all live, work, and do business? Send us a message, and let us know!