My Tech SEO Story with Elias Dabbas

Episode Summary.

Elias started his career selling Procter and Gamble products, but later moved to an internet company where he became the marketing manager. It was there that he was tasked with handling SEO for the company’s 45 products.

He currently specializes in online marketing and advertising and has been running advertising accounts for brands of all sizes and in all industries for more than ten years. He believes data science is critical for online marketing success.

Guest Profile.

elias Dabbas

✉️Name: Elias Dabbas

✉️What Elias does: Elias Dabbas is an online marketing and data science practitioner.

✉️Company: Advertools.

✉️ Noteworthy: He is focusing on building tools for improving the productivity of online marketers through his Python package, advertools. He also makes data dashboards and apps.



Key Insights.

Transition to data science.

Elias became interested in data science after hearing about it repeatedly and realizing that it was a powerful tool for analyzing and handling data, which is an essential part of his work in SEO.

Lack of third-party packages for marketing.

Elias was surprised by the lack of third-party packages for marketing and advertising, especially those that were geared toward marketers rather than developers. So he created advert tools to solve this.

To answer questions that involve data from different sources, it is essential to have the ability to combine tables from different sources.

For example, you might want to analyze how social media affects brand awareness and brand searches. You need to be able to combine Facebook and Twitter reports and analyze the data to see if there is a correlation.

Episode Highlights

💡Using website data to gain insights.

The conversation between Elias Dabbas and the students highlights the importance of using data to understand a website’s content and gain insights into its performance. By analyzing data from sitemaps and competitor websites, marketers and SEO professionals can gain a better understanding of what’s working and what’s not, and make data-driven decisions to improve their website’s content and search rankings.

💡Ethics in data collection and analysis.

Allison Kingsley’s question about accessing a blocked sitemap raises an important ethical issue in data collection and analysis. As data becomes increasingly important in business and marketing, it’s important to consider the ethical implications of how we collect and use data. Transparency and permission are key considerations when collecting data from websites, and businesses should strive to follow best practices to ensure ethical data collection and analysis.

💡The role of visualization in data analysis.

The conversation also touched on the importance of visualization in data analysis. As Elias Dabbas explained, creating custom visualizations like treemaps can help marketers and SEO professionals better understand the content of a website and its performance relative to competitors. Visualizations can make complex data more accessible and easier to understand, allowing businesses to make more informed decisions based on data insights.

💡Is it necessary to learn for a career in SEO?

It is argued that while Python is not necessary for SEO, it can be useful for data analysis and answering specific questions that cannot be answered through the default reports of Google Analytics. Python is not a requirement for SEO, and programming skills are not the sole focus of SEO. While programming knowledge can be helpful in some aspects of SEO, it is not the main focus.

💡Use API for a thorough analysis of web data.

Analyzing data from web platforms requires accessing the API to retrieve more than the limited rows available through tools like Google Analytics and Search Console. This allows for a more thorough analysis of data. To effectively analyze data, you need to connect to the API and extract all the necessary data for better analysis.

Connect with Elias;




Episode Transcriptions

Chima Mmeje 0:03
Alright, we’re good to go.

Elias Dabbas 0:05
Cool. Hi, everyone. Thank you for coming. We’re completed right now.

Chima Mmeje 0:10
Yeah. Why did you say?

Elias Dabbas 0:11
It’s everyone’s attending, right?

Chima Mmeje 0:14
Yes, yes. Yes, sir.

Elias Dabbas 0:16
So when we started with with the main questions, you raise the I guess we start with how I started with SEO, I basically started using I was, I, my background is I went to business school. And I was working in something that has nothing to do with the internet, I used to sell Procter and Gamble products, physical detergent, soap, that kind of stuff.


Then I changed fields, whatever the story is not important, but I ended up working in an internet company. It was one of the largest in, in the region. It was modelled after Yahoo. And I became the marketing manager. And we had similar to Yahoo, we had like 45 products, a huge variety of products, like sports, and jokes, and business and forums and all these things. And then SEO was not that established as it is right now.


And the CEO decided one day that we need to focus on SEO, and then he sent an email saying, you’re going to handle this in your department. So this, this is how it started. And there was this one lady who was handling the stuff, the SEO stuff, and we did like 45 projects on every website that we had in the group.


It was a lot of fun, many challenges, many things to learn from, and this is where I started getting to know this, this field. Yeah, so we did a lot of work there. I was I was handling the marketing in general. And part of that was was was the SEO at the time. So and I continued working in the company, and then I started, I moved countries and I started working on my own. As a freelancer, I think many of you are doing this or want to do this.


And it was everything was was was going well, I was doing more bit of SEO and a bit of sem campaigns. And a few years ago, I heard the word data science, a zillion times data science, data science, data science, machine learning machine learning, what the hell is this? So I was really I was very intrigued and interested because most of our work is based on data and the science of data should be interesting. And I got in and I never got out.


So I got addicted very quickly, because because this is this is I realised quickly that this is exactly what we do. We, we analyse data or not, that’s not the only thing that we do. But analysing data is an essential part of what we do, we need to know, we need to understand something about the business about the website, how it’s going, what’s working with it, what’s not working, and we handle a lot of data all the time. And this was like amazing, because it was much more powerful than doing it with Excel, or manually.


And I can handle much more types of data, quantities of data and so on. And very quickly, I was able to use the stuff I learned in three, four months, with real life data, like it was, I was messing around a lot. And I made many mistakes, of course. But it was It wasn’t like you have to wait five years and then go to the field. It was like, I don’t know, learning a musical instrument where in lesson one, you you hold a guitar and you start strumming.


Okay, so that was you know, we start playing in day one, lesson one, even though it sounds terrible. Doesn’t sound nice, but at least you’re playing music from from lesson one. So it was something like this with with data science. And then there was this object called the data frame. You probably know it, which is basically a table.


And I also quickly realised that my life consists of a bunch of data frames. We will search console you go you see a table of URLs, impressions, clicks, whatever, you go to Google Analytics, bunch of tables, anything social media. So being able to work with this object called the data frame was was is definitely important.


And data science has many tools for that. And so it was great. It wasn’t like going away from what work, even though at the beginning, it was, it was, but then, because it’s a practical field. And because our work is based on data. It was great. So you’re learning at the same time and you’re implementing something, messing around making big mistakes, of course, but you’re learning a lot.


Another thing that was surprising for me was, I was looking for third party packages, for marketing for SEO for advertising. And I couldn’t find many, like core any. And I didn’t believe it, because most other fields had packages for finance, economics, medical stuff, scientific stuff, physics. But there was nothing. I don’t want to say nothing. It’s unfair, because there are, but there isn’t like this, this big category of packages for the marketing people, there were many marketing and advertising packages, but most of them were not for the marketing person.


Okay, they were for a developer doing some advertising on the page and how to put it on a website or something to change title tags, but it’s more for the developer, rather than the marketing person or the SEO. So I wanted to create a package such a package, and I started working on it.


And, you know, it was, again, messy and frustrating and not a clear path. And I’m still working on it. Maybe, you know, it’s called advert tools. It has a crawler, it has a log file analyzer and a few other tools that it Yeah, and so on. I kept on working on these things and developing them. And it’s it’s very nice. It’s fun.


It’s it’s cool. Yeah, so this is this is basically the background. I don’t want to lecture you much. But I’d like to see if you have specific questions, or if any of you, I’d like to know where you guys are. What are you doing now? How can I help? What do you want to know? Or did I miss something Chima?

Chima Mmeje 7:24

Yeah, okay, just give me one moment. So you talked about trying to organise my thoughts here in the way that you talked about hearing a lot about data science a couple of years ago. And that led you to take up an interest in it, in what contexts when you hear me in about data science.

Elias Dabbas 7:48

It was data scientists was going to be the sexiest job in the 21st century, there was this famous saying, there was, there’s a lot of research saying that there’s going to be a lot of need for data scientists. It was just general, like in the business and marketing. Publications, I would I would read about this. And that was just a treat. At the time. I didn’t know that it’s this is something that I could start working with you No.


But what I did, there was like immediately clear, and I think this is this is becoming way more important than it’s going to become much more important in the future. Because in SEO, like if you if you look five, or 10, or 15 years ago, the amount of data that we worked with was was if it was if now we work with this much data we used to work with with much less much less data. Before think about the world before mobile apps were before social media.


And before, it was like Search Console, Google Analytics, a few things. Now, the smallest website has like 15 dashboards, of data about the traffic, the apps, the social media, SEO, you have third party tools, you have business data for the company itself, like maybe they have a CRM system or sales or customer service.


So the ability to to understand what’s going on. It’s always been crucial. But now you have so much data, right? So you need to be able to to combine the stuff that you want to analyse it the way you want to visualise it, and so on. And we will only have more data. Moving forward more sources, like you have like three sources, you’re going to work with 10 and 20, and 30. And more data is in megabytes. So if we used to work with 20 megabytes now we’re working with gigabytes, and so on. Does that make sense?

Chima Mmeje 9:48

I like the part where you said that it’s only going to there’s only going to be more data going forward in the future. That’s interesting, because it brings me to my next question. to know, and if you could give us some examples of all the different use cases of data analysts of data science in SEO.

Elias Dabbas 10:10

Sure. The first one is, is analysing the data that you have in your web platforms like Google Analytics and search console, you have a limited amount of rows that you can export. Like, it’s 1000. Here, it’s 5000. There like it’s, that’s it, but actually, you have 300,000 rows of data. You need to be able to connect to the API, and get the stuff that you want and and run a much better analysis than that. This is this is this is an easy, clear thing that you that you need.


The other one is the ability to combine different tables from different sources. Okay. So a very simple, very simple question could be, is my social media affecting my brand awareness to the degree that people are searching for me more? Am I getting more brand searches? Because of my social media? How do you do that? Well, you get your Facebook and your Twitter reports. And you see how many impressions you have here.


And you see how many impressions you have on your, on your brand keywords, and you see, you analyse if there’s a correlation between them. Now, it could be very simple, it could be very complicated, because your brand’s keywords might be very easy to pick out of your keyword report, if you have if your brand name is very distinctive.


But if you have a brand name that’s quite generic, if you have something called, and people are searching for cars, how do you know that it’s, it’s, they’re looking for this website, or they’re looking to buy cars in general? So you need this flexibility to be able to deal with data and filter your keywords properly to say, these are the brand keywords, these are the non brand keywords. And this is what we’re doing with Facebook, this is how many impressions we have here.


Now, if if this question is going to take you a week to answer if you’re going to spend three days getting data from Facebook, and aligning it, and three more days getting connected to the Search Console API, and you’re gonna say it’s not worth it. But if this is something it’s going to take you half an hour to do. It could be very interesting.


You know, another very simple thing. Let’s say you want to fix you want to do something with your title tags, okay? You want to analyse them first, you want to understand what’s going on? And do you want to modify them on a large scale and send those recommendations for them to be changed?


Okay, now, maybe you have 123,000 title times? Do you go one by one? From one to three up to 123,000? Or do you have some techniques to say, Okay, show me the title tags that end with a brand name. Or give me the title tags that have the brand name, duplicate it. So maybe I can make them shorter.


You want to analyse the length of those title tags, yes, the average might be 72.3 characters, but you might have half of them 25 characters and the other half, like 150 characters, the average might be somewhere there. But half of your title tags are terrible. They’re very short, they don’t say anything. And the other half are extremely long.


And they say way too much, right? So you need this flexibility to be able to have a conversation with your data, you want to be able to ask exactly the questions that you want to get a good overview, and be able to visualise summarise count words. And so on these and these are very basic things, you know, but but they’re very important.


And they become more important when you have large amounts of data. Like if you have 500, title tags, it’s not a big deal. But if you have 100,000 it becomes more of an approach that requires some programmatic thinking. Does that make sense?

Chima Mmeje 14:25

Yes, definitely. Definitely kind of like it speeds up the process of collecting that data, analysing it, and then your job is now kind of like making strategic decisions and all the other stuff that we’re doing manually before now becomes stuff that data analysis can help you automate.

So absolutely, yeah, that’s, that’s, that’s brilliant. That’s brilliant. That’s that’s a very good use case. Right. So some of the students recovered axon. How, what is the entry points? How did you How do they start? If they want to learn data science won’t Some resources to start with.

Elias Dabbas 15:01

So once once you have the keyword data science, you’re good. As long as you don’t start with how to do Python for SEO, you’re good. One, one, I started with data camp data It’s the great thing about this is that they teach you in an interactive way. And you start coding. And lesson one, just like I said, with the guitar example, you start playing music and day one, you start coding in lesson one, okay.


And this, because it’s data science focused, they have, they have a path for data science, they have a path for data visualisation, they have a path for machine learning, and all these things set up for you. So you can just go there and follow this and, and learn and you learn at your own pace. In in the afternoons, or whenever you have time, you finish one course, every two, three days, one or two courses a week that like these are short courses.


And it doesn’t have to be this, there are many other platforms that that you can use. But keep in mind that once you know how to just open an Excel file with Python or with R or something like that, you’re starting to analyse data using a programming language and not excel. And, you know, you’re you’re starting your work there. So at the beginning, it might be a little bit frustrating if you’re not using us to programming or but but the good news is that you’re probably not learning this to become a software developer.


Okay. So an important distinction to keep in mind is something that is called a programming user, or a programmer, and software developer. Now, the difference here is that, as a programmer, you’re using programming a programming language to do work, okay, you want to scrape 5000 pages and pick up some prices and create a table, analyse the data, tell your clients something, you’re you’re you’re doing stuff for your own analysis or for your colleagues, a software developer is someone who builds software so that other people can use it.


Okay, and this is very different, and requires so much more knowledge and information, you probably don’t want to do this. Okay? Because your your work is SEO, you want to learn a bit of programming bit of data science, a bit of statistics, math, machine learning, to get productive at what you do best, which is SEO, you probably don’t want to build SEO tools, maybe you do later on, that’s great. But if even if you want to become a software developer, you have to start as a programming user as a programmer.


And this is within, I don’t know, six months, seven months, you can start doing stuff that is useful. And when I say some of the stuff that’s useful, I’m talking about the simple fact of connect to Google Analytics API, and get 20,000 rows of keywords or URLs and start doing some analysis on them.


Right? You’re not getting into software engineering concepts. You’re just getting work done. And once you do this, once you start that this way, it becomes easier for you to navigate and and select which which path you want to go you want to focus on. visualisation, you want to focus on crawling, you want to focus on development, machine learning, whatever you want to focus on.

Chima Mmeje 18:49
Okay, that’s that’s a really good answer. The first website that you mentioned was data., right.

Elias Dabbas 18:57
Yes. Let me type it for you. If you want. Yes. They can all see

Chima Mmeje 19:00

that. Yes, brilliant. All right. Does anyone have any questions for Elias? If you do, just raise your hands and we’ll draw the mic to you. No questions. Yes.

Elias Dabbas 19:20
Alison has a question.


Chima Mmeje 19:22
Which was raised. Alison. Oh, Alison. All right. Brilliant. Obviously, you can turn on your mic. Okay.


Speaker 3 19:31
Thank you for coming. Okay, I have a question. And my question was, what are the pros and cons of using data science in as an SEO what are the pros and cons if as a beginner I want to start using


Elias Dabbas 19:52
I can’t think of cons. Not because yes, I am biassed of course. But the the main idea Here is that an essential part of our work is using data. Right? you visualise it, you filter it, you handle it. And based on data, you make your diagnosis and and you say, Okay, here’s an insight, we are lacking content over here or the, the competitors are doing this. So we want to do that.


So I think it’s an essential part of, of our work. It’s not, I don’t see it as a, as a plus is as a nice to have thing anymore, because as I mentioned, at the beginning, you’re working with 1015 data sources. And each one of them has like 50,000 rows here and 60,000 rows there and you want to combine them.


It’s very difficult to do them with Excel, or other manual tools. It’s not that you cannot, you can, but I mean, try opening 100,000 rows, Excel sheet with with with Excel. It’s gonna be like, you know, so the end of the day like this is this is not, SEO is not like managing a restaurant where you go physically, and it’s everything, everything is data, like, Okay, you see the website, you, you click around, you go to the process, you read the content, that’s very important.


Of course, it’s not just you don’t want to just blindly do it. But you really want to be flexible and say, talk to the database people and tell them give me pleased a list of all the products and descriptions that you have. They’ll say, hey, here you go 200,000 rows of data. And you have a theory in your mind. And you say okay, I want to see how my crawl dataset compares to the products in the database. You want to see what’s going on and you need to be able to analyse, I’m thinking of cons.


And I can’t I mean, even even though even if you don’t do Python or R or stuff, you need to know stuff about data. Okay, like data visualisation concepts, you need to know them. Regardless, you need to know what a histogram is, what a bar chart is, right? You need to know how to read these things. In order to see them on on some dashboards and you need to create them, because you want to understand the data that you have. And diagnose something and shared with the client. Can Can you think of any cons of of learning data science?


Chima Mmeje 22:43
Honestly, Allison, that question is for you.


allison Kingsley 22:47

Okay, I can think of any accident because I’ve been searching for some maybe people giving negative reviews on why maybe they’re saying it’s not really necessary to to learn Python, data science as a check su so I wanted to find out if there are some things that are negative that


Elias Dabbas 23:11

actually do not so about this, this Python thing. And it’s I think it’s unfortunate that we have this Python for SEO topic. I think it’s it’s it misses the opportunity.


It’s it’s really making people frustrated the beginners and the experienced people. Let me let me tell you a bit about that. Because I think that that’s a very important distinction to make. So Python is the tool. Okay. Data science is the thing that you and I are doing. We care about analysing data. I think this is clear, we don’t need to discuss this.


You know, you go to your Google Analytics, you’re analysing data, you have an Excel sheet, you have Google Data Studio, analysing data is an essential part of what we do. Now, you can do it with paper and pen, I don’t care. You can do it with Google Data Studio, you can do it on the back of an envelope. I really don’t care about which tool you use. Okay. The important thing is to know the concepts behind it. Okay. Let me give you a specific example.


Let’s take the concept of standard deviation. Okay. How do you calculate the standard deviation of a bunch of numbers with Excel? That’s a silly question. That’s a very easy question. Because the answer to this question is, you go to a cell and you say equals STD V or STD, I don’t know what it is. And you select these numbers and you hit enter, you get the standard deviation. How do you do it with Python? Same simple answer.


There’s a function called standard deviation or STD or STD Dev, whatever. That’s very easy. The important question is What the hell is the standard deviation Why would we care about it? When do we use it? When we get the standard deviation? What do we act? How do we act based on that number? How does it relate to the average? Right? And these these questions are not straightforward questions. They don’t have a straightforward answer.


And that’s why they’re important questions. Now, if you learn Python, you’re not going to find the discussion about standard deviation. But if you learn statistics, or you learn data science, you will, and this is the thing that we are interested in. Let me give you another example of for copywriting. Okay. How do you create an article on WordPress? That’s a silly question. You know, you go you log in, you click Create article, enter some text here, enter some text here. Hit publish, you’re done.


So if you want to learn WordPress, you will learn how to create an article, which is a very easy thing to do. click click click fill in some text. But that’s not the important question. The important question is, what do I write about? It’s the content, right? When do I write? How long should the article be? What about how do I do research?


How do I know that I have the right topics? There’s something called link building, there’s something called social media, there’s something called SEO, there’s something called outreach. And these are the things that you care about. When you create content, right? How do you do it on WordPress, easy, three, four clicks. And, and you and you create it, and it doesn’t matter.


It doesn’t have to be WordPress, you know, there’s, there’s a million content management systems and blogging platforms, and they all work pretty much the same way, you know, click, click, click publish. Done. That’s easy. So this is the difference between Python, like WordPress, in this case, and data science, which courts which would correspond to blogging, let’s say and or content creation, this is the soft skill, this is the more the thing, that’s it’s not straightforward.


That’s important for understanding our data. And this is the tool that we use for it. Now, again, you can use Excel, you can use our Python, and so on. So many. And there’s a lot of stuff with Python for SEO, that is focused on something called automation. Okay, how to do click link to Google Big Query and send notifications to slack. Okay, that’s very interesting. But that has nothing to do with SEO. Now, if you can do this, that’s great.


It’s wonderful. But you don’t need that to do SEO. And most of the stuff about Python for SEO is something like this. How do you automate getting the data from here and putting it there? Okay, it’s that’s cool. But it’s it’s not? It’s not an SEO task. Okay. You have engineers, you have web developers that can do this much better than you and me? And, therefore, yes, you don’t need Python to do SEO? Yeah, of course you don’t. But that’s, that’s not the thing here.


It’s it’s about data science. It’s about knowing the colour scales. In data visualisation. It’s about knowing which colour scale to use, how to use size of bubbles, and make a bubble chart to express more stuff about the data. How do you count values? And a bunch of these are the questions that that we care about. Because we care about analysing data and understanding it. Automating pipelines is definitely great and cool. But that’s not part of SEO. Does that make sense?

Chima Mmeje 28:57

Okay, this is interesting, because, well, I think we’ve had a lot of discussion on machine learning, or we’ve had one discussion on machine learning. And part of that was on Python for SEO and how it can help you automate tasks. Yes, and I think we’re even going to have a training session on that.


So are you saying that? Yes, I agree that you don’t need Python for SEO. But are you saying that it’s a skill that’s, you don’t need to learn at all? To become a good cashier?

Elias Dabbas 29:31
You don’t need to learn Python?

Chima Mmeje 29:32

Yes. It’s,

Elias Dabbas 29:34

I mean, if you want to go into data science, part of it is learning a programming language. Okay. But it’s just like, Can I share my screen? I have, I have a few. Because because you can kind of share it’s going on. I think you need to enable. I have I just did that. And I want to show you why I think I mean this is is not just semantics about oh, oh, this is the right name. Not that I don’t care about the names. It’s just trying to give the right keyword for beginners.


So they don’t get Miss misled. So this is maybe some of you read this. I don’t know, this is about data science for SEO and digital marketing. And I, you know, I go through the recommendations of what you could do and what my recommendations are. So that let’s see what happens if you we have two options. Okay. You can you can go and google learn WordPress, the tool, Python, or you can say, I want to learn blogging.


Okay. Now, if you can you see, maybe it’s better. Yes, yes, yes. Okay. So if you learn WordPress, you will end up learning these things, these topics over here. Now, some of them are going to be intersecting with with the blogging course, or the blogging book, which is how to create a post, how to embed images, how to change a blog theme. And these are important things to learn.


And you will also learn stuff that you really don’t need, you know, if you want to blog, you don’t need much PHP, WordPress, REST API and Ajax and plugin development and blocks and, and connecting to payment gateways. And you don’t need that to blog, which is you can, you can repeat the same thing and say you don’t need Python for SEO.


At the same time, if you if you learn WordPress, you end up learning these topics. Obviously, these are not all the topics, this is just a simple, simplified version, you will miss the very important stuff over here. Okay, because if you want to blog properly, you need to know about keyword research and social media outreach and engagement and finding a niche and an SEO and all these things. So if you learn Python, you will end up learning stuff for software engineers. And it’s going to be frustrating for you, because you won’t see why this is useful for you.


And many of it will not be useful for you, especially at the beginning. And at the same time, you’re going to miss out on very important things with data science, that are going to be really interesting for your work in SEO. What do you think? Does this answer your question? Does it make the distinction clear?

Unknown Speaker 32:35

allison Kingsley 32:39
Oh, yes, yes, yes, I get it, I finally get it. Thank you.


Elias Dabbas 32:43

Is it clear? Why? Why Python for SEO is misleading. And it’s not just again, I don’t It’s not like just oh, this is the right name. Who cares? It’s because because if you if that’s the keyword that you use, how to learn Python, it means you want to become a software developer, right? And you’re gonna go through a course that teaches you how to build websites? Or how to make games or how to build some systems.


And you’re gonna be like, why am I doing this? And this is what many people are criticising correctly, that you don’t need Python to do SEO. And that’s, that’s, that’s correct. You don’t need to automate data pipelines from Google to this to that it’s great if you can do it. It’s absolutely wonderful. But what you need for SEO is data analysis. Right? You need to know how to handle data in a flexible way.


And the more powerful you are in analysing data, the better you can answer questions, because because the Google Analytics interface answers, like 96% of the question is, okay, you have, you just log in, you have, I don’t know, 200 or 250 reports ready for you. By date, you can have filters and you can, you know, it’s a great interface. But there’s these four or 5% of questions that you cannot answer them there.


And these are the important ones. You know, that’s, that’s why you’re the expert that we rely on, because many people can log in and see those default reports. But with your data skills, you’re going to be able to answer more specific questions or uncover more insights. Okay. And this is, so if you learn Python, programming, software engineering, the natural thing to go towards is automation. Which again, is cool and clear and great, but not essential for for SEO.


Our keyword is insights. Okay, you’re my consultant. Here’s my website. Tell me something that I don’t know. About my website. This is where I started listening. Right? Give me opportunities for content for linking for tech SEO for, give me show me some opportunities that I am missing as a business owner. And I’m going to listen to you. And I’m going to love you for that. I have engineers, I have web developers, they can do automation stuff, they can do slack notifications, and they can connect to this API and put the data from here to there.


But tell me something that I don’t know. You’re my doctor. Okay. We go to the doctor and we say, Doctor, my stomach hurts me, the doctor does some analysis. Part of it is data analysis with with, you know, tests and reports. And doctor comes back and says it’s time your stomach, this is your liver.


Oh, really feels like my stomach? No, it’s your liver, whatever. And this is, this is our job. We are the doctors here. And we are we should be able to give those insights and come up with stuff that the client doesn’t know. And hopefully create great results. Insights is the key word coming up with insights. Automation is is cool, but not essential for SEO.


Let me give you just one example of what I mean. Okay, so what I do is I have this this. I always like to get SERP data. Okay. And I like to analyse it. Where did that go? One second. Lots of charts. This is for I just want to show you the one for insurance. Okay. Nope, this one? Yes. This one.


Okay. So you have a client, and the client says we are in the insurance business. Okay. And the first question you have in mind is who’s ranking for insurance? Keywords? Very basic question. So question number one, it’s just like question number zero. And one way to do it is is this I, these types of insurance, I got them from I just copied and pasted them, car insurance, Van insurance, motorbike insurance, there’s like 29 types.


And I created my three variations, cheap car insurance, best car insurance, and car insurance quotes. Okay, so we have 29 times three is like 50 Something, keywords, and I run each one through the API, and I get the results as a table. Okay, now I want to, I want to get some insights about those, those, those ranks. And as you can see here, this is a summary.


So In those 86 or 87 times that we ran the queries was in position number 134 times was on position to 19 times 15 here, and so on. Okay, so this is kind of like a heat map for SERPs. So this webs and you know, each domain you can see where, where they rank most. So this is automation.


Okay. I’ve created a function that takes a table and this table has to have three columns, only the keyword, the rank and the domain. And you just feed the table in this function and it gives you this chart, yes, this is automation. But this the idea behind this kind of automation is that whenever I start on a new project, I can very quickly get this overview summary of how things are going.


So you can see that MoneySupermarket is dominant, especially number one, and two and three, compare, compare the market is, is here and you can immediately visually see where things are concentrated. Now, this is the UK, the same chart for the US, it’s not as concentrated as the UK. Right? And here, I’m starting to give you stuff that you probably didn’t know about your business. That in the UK, it’s very concentrated here. But in the US, it’s much more spread, spread out and so on. Is that does that make sense? Yes, yes, yes. Yes. Yes. Thank you.


Chima Mmeje 39:48
All right. We have a question. Sadiq wants to ask a question. Please go ahead.

Sodiq Ajala 39:54

Um, thank you very much. Hello is a wonderful session and you already touched points on some of the problems I have. And all of them is that. So take for instance for a, an SEO like me, who wants to delve into that, you know, data science part of SEO, and then I need to, you know, practice all of these things.


So that I can like have an urge so that I can have stories to tell whenever I find myself in an interview room. And I do not have a knowledge of these contexts, like the way it works. So take, for instance, in an interview, somebody asked me a question about, you know, data science, you know, in, you know, in the SEO space, like, what do I say, so, for me to like, be able to do that, I believe that I also have, like practised before getting myself into those rooms.


So, I’ve been problems to solve in that specific area that I can solve on my own before getting myself in an interview. How do you think or what would you advise that, you know, I do, we, I do think I can go look for problems I can solve as an SEO, on combining data into rates where I can have stories to tell or something to create a portfolio or pleasure.


Elias Dabbas 41:15

Sure. Great, that’s the these are very important points. Basically, just to confirm your your you, okay, you want to take this to the next level and use your data science knowledge and skills to get clients and to answer questions in an interview. So one, the way I think about it is, and you mentioned a very important keyword, which is create a portfolio.


The way I think about it is you can do marketing for yourself by creating great content, this content shows that you’re an expert that you know, what you’re doing, and especially with with data science, content, the difference than traditional articles is that you, you, you include the code, okay? So if when you read articles in The Economist or New York Times, or wherever there’s going to be text, there’s going to be a chart that says this is the unemployment rates in this country.


That’s it. But with when you create a Jupyter notebook with with code as well as the text, you say, this is the code that produced this chart, and I can see your skill. I can verify the answers, I can see that you’re confident. This it’s transparent. You know, you’re not just giving me a picture of a chart. You’re saying, here’s the raw data. Here’s the code that produced this. visualisation. Okay. Now, there’s there’s a lot of stuff to do in in public data.


So you can you can do a crawl, you can you can crawl websites, and and analyse those websites and publish them on something like Kaggle or GitHub. Okay? You can you have the complete freedom to do, I don’t know, the crawl dataset of the week, every week, you can crawl a website, and run an analysis of this website and come up with recommendations. It doesn’t, it doesn’t have to be for a client doesn’t have to do anything, pick up a website and say, look, the the the h1 tags have this issue.


There are too many, there are many pages with seven title tags, or the title tags have this problem. The performance of the website is this. And you can keep creating these things as your portfolio. Now, the more you do this, you’ll start getting some you’ll start seeing some patterns.


Okay, you’ll realise that wait a minute, every time I crawl the websites, I ended up doing 12345. So you can have your own templates. And I have a bunch of these that I can I can share with you on my Kaggle account on my GitHub. You can you can definitely take these as a starting point and build on top of them.

Sodiq Ajala 44:08
Sorry, Kaggle.

Elias Dabbas 44:10

One second. Thank you. So here’s here’s my, my account, you’ll see two portfolios, one for datasets, which is just data files that I uploaded, in one four notebooks, which is analysis that I ran and many times I analysed the data that I upload, okay. And this is this is your portfolio and in this case, you are unstoppable in the sense that you don’t have to wait for someone you create.


You crawl a website three 4000 pages, but the data, analyse the URLs analyse the title tags and so on. And two weeks later, you might think, Wait a minute, why didn’t I do this, you go back again. And you publish version two of the same notebook. And you share it with the world and say, Look, I improved the process over here, and so on.


If you do this once a week, by the end of the year, you’ll have 20 3040 notebooks, a portfolio of crawl data sets and their analysis. And you will become the data science SEO crawler that is very well known and people seek does this make sense this answer your question?

Sodiq Ajala 45:38
Yeah, absolutely. Thank you very much.

Elias Dabbas 45:42
Great. Milena.

Chima Mmeje 45:44
Yeah, and then I have a question. So we’re going to go to Montana.

Speaker 5 45:49

Right. Um, thank you very much for being here earlier. So I’m really enjoying the conversation today. So mine is probably to add on to what somebody just asked. I’m sitting in I’m wondering of, I’m interested in data science, I am an SEO.


But you already mentioned you don’t want to get too much into the tech, the technicalities of maybe learning programming or data science, you want to solve actual SEO problems, right. And I’m wondering what, from your experience right now, what SEO problems are the hottest things to solve right now that if I were to show I can solve this, I can get a couple of words and as you know, walk into a room and say I can do this. It’s not

Elias Dabbas 46:41

about using ours. It’s about this. Sure, okay. Let me sorry, let me again, share with you some some things that I that I do. So we can we can make this clear. So first of all, I have I’ve created a library in Python called advert tools, which you can use. So you don’t need to create your own SEO crawler. You don’t need to do that stuff.


But you can. You can. You can use it. Okay, so this is XML sitemap. Do you analyse the XML sitemap usually? Sorry? Yes, yes. Okay. So what if I gave you a sitemap that contains 218,207 126 URLs with a last month? How would you analyse them? What would you do with it? Here’s my XML sitemap. Do you have an 18,000 URLs?


Munene 47:57
What do you do? I wouldn’t know what to do is the site.

Elias Dabbas 48:03

Okay, so so this is this is the actual site map of a website called It’s probably the biggest UK website for use cards. Okay. So there’s, there’s a function called sitemap to the F sitemap to data frame, all you have to do is give it a URL of a sitemap, and it converts it to this table data frame format. So now you have an Excel sheet basically, you know, XML is messy and difficult to handle. This is a table.


Now the next step would be to take these URLs and split them into their components. So this is the URL, you have a scheme HTTP, HTTPS, you have a network location, which is the domain subdomain, and so on, you have the path you have the free query fragment. And what this function does this function is called url to DF convert a URL to a data frame. You have directory one directory 2345, and so on. Okay, so one thing I would like to know is, how many how many cars do they have?


Which brands do they have? And so on. So what I did is filter this URL data frame, where directory one is equal to cards. Okay? So there’s cars, there’s plant, there’s dealers, I don’t want dealers, I don’t want content I want this table just like you do with Excel, filter it where directory one is equal to cards, okay. And then remove the city names and count the values.


Okay, so directory two is directory one is cards and count the ones you see here. It’s it’s a, it’s a brand name of a car. So what we could get is that the top cards listed on this website are these. Now I know that they have 1787 Mercedes Benz cars listed on this website. This is how many BMW is they have Audi, Volkswagen and so on. So directory one says cars, directory two says BMW directory three tells me which which model of BMW this car is. Okay.


So we create a subset of only three columns, directory one, two and three. Director one is only cars. Directory two contains only the top 16 And three is the type the specific type of car, convertible, coupe, estate, hatchback, SUV, and so on. And now we can we can do a visualisation of of the cars that are listed on autotrader right now, July 6. And you can see visually how what kind of inventory they have. And this is interactive.


So if you want to know more about BMW, you can click through and see what they have over here. So three, Series C, this is cars BMW three series. So we know that they have 146 cards of three series, which is 9% of BMW, and 1% of cars. Okay, so now you have an overview of of the actual inventory of cars, makes and models and you have the relative stuff, and it’s interactive. And you know, all this about them running by running this notebook and takes you like, a few minutes to run. And this is 200,000 URLs. So we just extracted the stuff that we we are interested in. What do you think?

Speaker 5 52:11
Um, I think it’s pretty cool. But I think a follow up question I would have is then what does the client do with this information?

Elias Dabbas 52:21

Well, this is this is good question. Great question. This is as a starting point of understanding what’s going on. Okay, so if the market is very much interested in Kia, and they don’t have much focus on Kia, you, you look here and you see, you can compare this with your keyword research. Okay, let’s say, you know, they just have 645 kids versus 1700 Mercedes Benz.


But your keyword research says that Kia is much more popular in that country. Or you can you can do the same analysis for different websites. Your sitemap competitor, a competitor, B competitor? See. Now if you have 23 BMW cars on your website, and your competitor has 2000 It’s going to be very difficult to rank for this because and this is these are used cars.


Okay. So this BMW is listed now in seven days, it’s sold in two days, it’s sold maybe a week. I don’t know what what the thing is. But you now have an overview of what’s going on. In terms of content split. And this doesn’t have to be cars, it could be sports, it could be news, it could be anything.


So we’re building an understanding of, of the of the content of the website, we’re creating our own custom, simple dashboard, visualisation. And, and this is this is automated, you can take this notebook, replace the the XML Sitemap URL at the top, make a few changes, and create your own tree map of some other website. This is a very simple example, where at the beginning, you told me, I don’t know what do I do with 200,000 rows? Does that. Does that make sense?

Munene 54:19

Yeah, that breaks it down really well. That I think that train of thought from fine, I can call I can programme to what is what decisions can I make with this data is

Elias Dabbas 54:30

absolutely we don’t care about I can code I can programme. We care about insights, understanding what’s going on. Now maybe the site owners know this. They probably do. They should. But not maybe not the person who’s who’s handling the marketing there or the SEO, or maybe they don’t know this for their competitors.


And you can do the same thing for like five, six websites and put them next to each other and say, this is the distribution for competitor a competitor B and we can see what’s going on in the market. Just like the other one I showed you for the SERPs. It’s about understanding.


It’s about insights. Okay? In the UK, it’s very condensed into these websites. But in the US, it’s it’s spread apart. I’m telling you something here that you didn’t know before. And that thing you cannot get from looking at a search result. You can keep googling so many times, but you won’t see it in that way. Allison, you had? Have you raised his hand? Yes.

Allison Kingsley 55:41

Okay. My question is, okay, supposing I want to do a case study on his side, and I used Screaming Frog to see how I can access a sitemap or to do a crawl on your site. And it’s blocked by robots. txt and new access on there to access the URLs and abusing or to add advanced tools to to see I can, of course, and there is a there is something that says okay, use robot CAC and, and put force like put.


So you can access the IRS is I want to actually see it. Like, because after doing these have to post on my blog, like, this is actually what have done. I’m asking if it’s illegal to use to use it now to particular? Then if you get my question.

Elias Dabbas 56:36
So you’re asking how, how can you publish stuff on your blog?

Allison Kingsley 56:42
Is Is that Well, I don’t have access to set maps and I had to use? Okay, well,

Elias Dabbas 56:53

well, one of the most important things in data science is transparency. So you have to mention that this is how you got the data, maybe you have permission from the website, maybe you decided, you just have to make it clear that this is how you got the data. And the code has to be there.


If you’re not comfortable with that, or the website might, they might object or if you don’t think it’s the right thing to do maybe, maybe get another website that they don’t have pages blocked by robots. txt, so you’re not doing something that you’re not comfortable with. And I agree with you that’s that’s important. We don’t want to not follow the rules of, of the website. So maybe if, if that’s the case, and you know, there’s millions of websites to to crawl. Thank you.

Chima Mmeje 57:54

i This has this has this very interest. I love the fact that everyone here has learned. If nobody has any questions, I think we’re going to bring this to a close. But if everyone can just turn on their camera real quick. So last can see your spaces. Yeah, so it doesn’t doesn’t look like he’s talking to ghost or something. Yeah. Hello. This is a class last year, so they need to see their faces.

Elias Dabbas 58:23

Before we Great to meet you. Nice to see you. So can you I’m really curious like what’s what’s your view? Like? Where did this data science thing come up? come from? What are you planning to do? Based on what we talked? Are you changing your mind? Is it more interesting? Like I’m, I’m interested to know more from you.

Chima Mmeje 58:44

Anyway, I think I’ll start by saying that you have a special request from Alison. So yeah, she was the one that wanted. Yeah. And I’m still processing everything you said today, because it’s kind of conflicting with what I’ve been hearing generally in the industry about Python. I personally believe that both can exist together.


I don’t think you need to go the full way. When any Python I think you can learn Python enough for automations. And to kind of speed up your workflow, because there was also something that Mike King said about when you can code then a lot of the stuff that he was doing, that will take like maybe two months he could do them in like us. So so you can you can learn that to the extent that he helps you make your workflow faster, but I think that’s about it. That’s exactly the point guested completely something I’m not sure?


Elias Dabbas 59:42

No I’m not I’m not against it. I’m just saying that the keyword Python for SEO is misleading for beginners, because it will get you through the process of learning Python, which is a programming language which gets you focused on software engineering products, and away from this In the Dousset, exactly right now, you want to learn just enough programming to be able to do work at a large scale. Yes, right? You have a you have an Excel sheet and 50,000 rows.


What do I do with it? Well, here’s here’s a notebook that I created, that you can take a copy and just modify some stuff, and work with it. That’s exactly what I’m trying to say. Maybe clear. But it’s just when somebody comes in, says, Python first you okay, how do I learn Python? Well, you take a Python book, you take a Python book, you go into software engineering, away from what you want to do. Right? Yeah, I think


Chima Mmeje 1:00:41

I understand what you’re saying. You’re saying most of the advice out there kind of needs you away from the focus on SEO to focus on developer skills, which is not trying to learn it. Now.

Elias Dabbas 1:00:51

This is where you don’t need Python to do so if this is Python, it’s about automating stuff and doing stuff. Yeah, that you don’t need that for SEO. That’s true. I fully agree. Is that clear?

Chima Mmeje 1:01:06

Like do you? Yeah, that’s, that’s, that’s, that’s good. That’s, that’s, and once you know, once you

Elias Dabbas 1:01:11

know, just enough Python, or any other programming language to do these things, then it will become much easier for you to make a decision and say, Okay, I want to focus more on programming, or I know enough programming to do to get things done, I want to focus on data visualisation, because this is what I think I can excel in, or automation or machine learning or, you know, whatever you like, you know, but at the beginning, just the data science course, or the data science book, gives you just enough Python, as you mentioned, to start working, within a few weeks, you start playing with data frames, and then slowly you grow your programming knowledge.


It’s important, it’s very good. The more you learn, the better. But don’t focus on software engineering, per se, focus on data science. And by the way, part of the the science is if you if you if you look at any article, you’ll see like three circles of Venn diagram, one of them is programming, and one of them is math and statistics. And the other is domain knowledge. Okay. And domain knowledge in this case is SEO.


So the best data scientist in the world is not going to analyse SEO data better than you, because they don’t know about SEO. And this is this is something that’s that you’re building, and I think it’s it’s, it’s great. Okay, and I let me share with you this this article, because if anyone’s interested, you can you can read the details and see the distinctions. And and if if something is still not clear, maybe tomorrow or later, please ask me or let me know or because I think this is an important distinction to keep in mind. Does that make sense to you, man?

Chima Mmeje 1:03:06

Yeah, does. It does? It does? Yes. I’m glad we had this session. I’m really glad we had this session, at least make the distinction clear. Great, right. Yes, I think this is the end of our session today. Let me just end the recording.

Transcribed by