#326 Is the Data Analyst Role Dying Out? with Mo Chen, Data & Analytics Manager at NatWest Group

Generative AI is transforming industries at an unprecedented pace. But as AI changes how you or your team work, one thing is clear. Your skills also need to evolve. At datacamp, we offer everything you or your team need to adapt and thrive with AI. Whether it's business users looking to get the Most out of ChatGPT and Copilot, or developers and data scientists looking to fine tune models, you can learn the entire AI skills spectrum on Datacamp. Power your AI transformation today. Start learning@datacamp.com.

I always say this, and I may sound like a broken record, but you need to be able to deliver any type of insight in a way that a 6 year old could understand and in a way that would satisfy me or even someone even more technical as well. So if you really know your stuff, you can really dumb it down and you can make it so complicated that honestly, only the people who are really, really up there in terms of technical expertise can understand. When you go into a business, the ability to solve business problems and identify them is so valuable. Common sense and intelligence, I feel like, is so underrated nowadays in the sense that everybody's talking about implementing, implementing, implementing. But then imagine if you don't even know what questions to ask. Like you just go to work and you literally have no clue what to look for. AI will not solve that problem for you because every business will have its own unique problems.

Welcome to Data Framed. This is Richie. Every now and then, someone on my social media feeds will declare the imminent death of the data analyst role. Today we're going to find out whether that's likely to happen. At least for now, the job does exist. So I want to know how you get hired. I want to know what skills and qualifications matter and how you make a great data portfolio. Our guest is Mo Chen, a data and analytics manager at Natwest Bank. Mo's also the creator of Ultimate Data Portfolio and a tool to make Mo the ultimate data portfolio. He also runs one of the most popular YouTube analytics channels. So let's hear Mo's career advice. Hi, Mo, welcome to the show.

Thank you so much for having me.

Richie, great to be chatting with you. But to begin with, every now and then I sort of hear that the data analyst role is dying out. So tell me, is that really happening?

My answer will be no, because I clearly still have a job. So ChatGPT, I think it became super popular in 2022, November, I believe. And we're speaking at the time of this recording in September 2025. I still have my job. I actually progressed in my job as well. So I'm making more money, I am more senior than I used to be. So I would say no. I think AI is changing the skills needed in the industry right now. So rather than remembering everything, I think it's about learning the foundations well and then using AI to produce even better work, even more complex work. And there are many things that at the beginning of my career, when I started in 2018, so I'm getting relatively seasoned now, there were many things that took me quite a while to do, maybe even a day or five days. Now AI can do it in probably 30 minutes, maybe an hour. But then again, you have to know how to prompt, you have to know what to look for, you have to know how to combine various AI workflows and tools so that you can actually do the thing you want to do. The example I usually bring up is that imagine if I'm just a small coffee shop owner in the sense that there's nothing wrong with coffee shops, but I have nothing to do with data. So if I'm a coffee shop owner and I need to build a data engineering pipeline, even with all the AI that's right at my fingertips, I won't build it as fast as a data engineer. So that's it. It's just one of those things that when you tell me that AI will take away my job, I don't think so. I think you go to work and it's different skill sets compared to 50 years ago. Let's say you used to go to a hotel, you had the man who used to press the button in the lift and then you know, went to whichever floor non existent anymore, let's say automatic cashiers. Right. The self checkout, it eliminated that specific position. But I'm pretty sure those people were able to do something else within those grocery stores, add more value by doing other activities. I'm not a big grocery store expert, so I'll, I'll leave that to you. Okay.

Well that's very reassuring that the data analyst role is sort of still persisting. But you mentioned quite a few ways it's changing. I'd love to get into that in a bit more depth. Can you jump into which areas of data analysis is generative AI or AM or general like really good for? And which bits are still more human oriented?

Debugging, code, fixing code. Super good, super fast. Better than me for sure. I'm not an amazing Python coder or any type of coder in the sense that I'm pretty much self taught. And the way I see Data analysis is that I add business value because I have domain knowledge and I know how to analyze data. It's not that I'm really good at Excel, I'm really good at Excel and I'm really good at data analysis. If I'm just good at Excel, I'll be good at product demos. But the thing is when you go to work, and that's the differences between social media data analysis and like when I go to work, the real data analysis, because when I go to work I have to solve business problems. So just using Excel and being amazing at it is not enough. So the bit I think AI right now is really good at is sequential debugging for code or fixing code or at least what I use it for when it comes to technical things. And you didn't ask for any non technical use cases, so I'll leave that for now. Maybe you'll ask me or not. So technically use cases, SQL, Python, and those are really, really good use cases because it actually works from my experience. And then we have the vistools, Power bi, dax, measures tableau, calculated fields, quicksight, right now, for example, or even in Excel functions and formulas, because you have more flexibility and it's not that sequential in the sense that it's hard for the AI tool to see everything you've done. I think it doesn't work that good. So this is what I use it for mostly, but for coding, honestly, I never code from scratch anymore because I'll dive down to a super simple example and super exact example. Let's say you're using the Seaborn or the matplotlib Lib library in Python. By the time you annotate the chart with labels and titles and all of that. I mean, remembering that syntax takes up so much memory, whereas nowadays you can just prompt whatever chatgpt or whatever model you want to plug into and it'll do it in no time for you. So that's just so, so, so quick. That's one example that works extremely well.

Absolutely. I mean, certainly writing code for generating plots, even if the AI doesn't get it quite right, it's just going to generate a silly plot, but you'll be able to see that it's done something wrong. So very easy to correct that. You mentioned since writing code is like the strong point, it's been a bit of a trend for data analysis and for business intelligence to move like from sort of coding from R and Python to using tools like Power bi, Tableau, Qlik, all the rest of these kind of point click interfaces. Now those also have AI helpers in them. But do you think the fact that AI is good at writing code, is that going to reverse the trend or do you think it's not going to.

Make a difference with the VIS tools? I don't think AI is there yet and I don't see it's going to be there just because there's so much flexibility. And the way I see it is that a good data pipeline for me is something that is within the ETL process before it actually gets into the BI tool. The code is so clean and neat that basically you only need to do simple drag and drop and then it's just about answering the business question. So I actually think in those VIS tools, anybody who does really, really advanced things, it goes against my data engineering ETL best practices. But then you can, you know it, it's up for debate. So whoever's watching this or I don't know how the interaction is going to be, you can let me know if you don't agree with me. But the way I see it is that I think by the time the code gets into a schema database and the tables, those tables need to be really, really simple in the sense that if I visualize something in tableau, it's mostly drag and drop and that's about it. And I think this is again biggest difference between social media data analysis and at work. At work I produce bar charts and line charts. Hands down, that's it. Because it's simple to read, easy to understand. People are after insights, people are not after the shiniest thing. Obviously make it look nice, but if your dashboard doesn't even tell a story, it doesn't matter how nice it looks. And this is where I think AI is not going to replace anyone because you need that human touch. I really think you need that human touch. AI cannot solve the complicated business problems I know for this for a fact, because say, for example, I've been working on so many data projects recently, so I help other people create their personalized data projects in various domains. And I tried prompting ChatGPT and perplexity. I have Perplexity Pro using all of the models. I tried to give myself a huge shortcut, let's say analyzing customer churn rates or looking for good investments using stock data, or looking at healthcare data within the US to see which hospital is actually efficient in terms of how much money they spend per patient. And I'm telling you this right now. ChatGPT, Microsoft, Copilot, whatever, Claude, Google, Gemini, they could not do the Analysis even in the sense that they couldn't even get me to a point where they could give me the right questions to ask.

That's interesting. And it's worth knowing, like, what's not really possible yet as well as what is possible. So let's jump back there. Okay, so first of all, you talk about how there are actually only two plots you ever draw. And this actually reconciles with my experience as well. It's like, for business, like, plots, whatever. You're basically drawing line plots, which is like, does this metric go up or down? And you're drawing bar plots. Like, what is the most popular thing out of these categories? Which is amazing because, I mean, we teach all kind of different data visualization, but actually, you get a really long way with two different plots there. So, actually, yeah, do you want to talk a bit more through your potting strategy, and then maybe we'll go back to some of these AI topics.

My plotting strategy, I have to say this is where I'm gonna. I am gonna give a shout out to a company, the Economist, because I love their visualizations. There's something so simple and neat about the way they do things, and they mostly use, guess what, bar charts and line charts. So my plotting strategy is make it interactive. So if it's a dashboard, make it as interactive as possible. And then there are gonna be times, many, many times when you work for a large organization that people want you to build a dashboard that they can extract in a PowerPoint format or a PDF format. It's just the way it is. If that is the case, obviously you cannot make it interactive. So then you have to fit in everything onto one page. But my advice here is that try and go for only one big story point per visualization or per dashboard. Because if you have too many things and if you cater for too many asks, if someone comes to use that and says to you that, can I put this in the XCO pack? Can I put this in the committee meeting? And you say yes to everything, you'll end up with, no joke, 300 different metrics on a single page. And then nothing will be meaningful because everything will be lost. So number one is focus on fewer things. I would say one per visual. Even on a dashboard, I would go for, let's say, five, six visuals. Then there's five, six things you want to focus on. Use the headline or the subtitle to tell the story, give some context. So instead of saying, customer, churn dashboard, I'll just come up with an example. You could say that the highest Churn rate is in the H segment, 25 to 35, because of whatever the reason is. So you give a bit of the story, and then you guide the audience's attention. Color contrast would be another really, really good one. Don't use too many colors. Use every single color with a purpose. So don't just chuck on a color because you feel like it's. You know, it's good if there's no purpose to the color. Keep everything blue, and then if you keep everything blue and you make one thing pop, that thing will surely pop. Don't go against traditional conventions. Let's say we associate blue with cold and red with hot, so don't go against that. Or let's say red, amber, green, right? Red bad. Amber better. Green good. If you start color coding the good things with red and the bad things with green, I'm already confused just trying to say it. So imagine how confusing it could be for your audience when they're looking at it. I would say these would be my top tips. So just to recap and make your job easier as well, focus on fewer things, ideally one or two things per visual. Focus on context and storytelling, using your headline or subtitles. Use colors for contrast, and always use colors with a purpose. And don't go against traditional conventions. Let that be naming colors or anything along those lines. Another good example for the traditional conventions would be the way you plot the time right. You always put the time on the x axis because you want the line chart to kind of flow through from left to right because it would just look weird the other way around.

That's brilliant. I feel like we just had, like, a whole plotting masterclass in five minutes because there's probably gonna be, like, one physicist in the audience going, actually, blue should be hot and red should be cold, because it is in astronomy.

Oh, is it?

Hey. Yeah.

Well, clearly. Clearly, I'm no astronomer or physicist.

Yeah, they do things back to front. But again, you're right. There's a tradition, and you need to follow stuff. One thing you mentioned before, you're talking about how once you're in your BI tool, you don't want to be doing all the fancy transformations or whatever that should happen in a data pipeline beforehand. Which is interesting because all the power BI enthusiasts, they're like, oh, look at all these cool transformations I can do. So this feels like an important lesson on how you approach learning how to do good BI and, like, how you set up the infrastructure. So talk me through. Like, basically, if you think you shouldn't be doing, like, Fancy transformations in your Power BI or whatever BI tool, what should you be learning about in order to do good bi? What skills do you need?

First of all, let me just put it out there. This is just my opinion. There is no right or wrong. There's advantages to both sides. So let me just put it out there in terms of what you should be learning, let's say when it comes to Power bi. So I think you need to learn the basics well and first of all you need to feel confident enough with the tools. So rather than focusing on how do I do this? You focus on what do I do. I think that is the first part and whatever that means to you. Because if you are working, let's say in a data analyst role where you do tons and tons of visualizations, it'll be a little bit different from a data analyst role where you do some business analyst work as well. The level of technical expertise will be different. So it is really hard to pinpoint what exactly you should learn. But in terms of how comfortable you should get, I always give people the advice that instead of focusing on how to do things, focus on what to do and then if you get to that point, it's enough. That's all you need. And in terms of just going back to which engineering is better, where you do everything in the BI tool and you have complete control or you just have a pretty neat data pipeline in the sense that in the BI tool you do drag and drop. There is no right or wrong. I think it depends on the size of the organization because imagine a startup, 10 people, 20 people, right? The Power BI guru is going to be the data guru. It's probably going to be the data engineer, the data scientist as well, maybe a bit of business analyst as well. But then if you work for a super large org, I think you want good processes in place because you want reusable things. And the only way you can have reusable data, reusable code, reusable visualizations, reusable insights, is if everything is clear along the way and documented and simple. Rather than living in someone's head or living in one massive Power BI file or one massive Power BI data model that nobody except the person who created it understands.

That's really interesting. So basically as organizations get larger, you're going to want to focus more on lesson like solving individual problems, more on how can I solve similar problems over and over again. So reusability becomes more and more important, I guess as your analysis needs grow.

Yes. Yeah. I've only worked for a super Large organization. And again, this is just my opinion and experience, but the way I think things should be done at a large scale is you need to make them reusable because you need to be efficient. You cannot create keyman dependencies. When you have a startup of 10 people, you probably all know each other quite well and it's just the way things are. Like startups, they just move so, so fast. You do a little bit of everything and then from idea to product, you could do it in, I mean whatever, two, three weeks if you are super, super fast, depending on what the product is. But at large organizations you have sign offs, you have committee meetings, you debate things. And I think the way it should just happen is that you have clear roles and responsibilities and you produce neat and clean and trackable code. So you can trace it all the way back to source, wherever the data is coming from at source and you track it through all the way, wherever it goes in the organization. So you know where exactly your data ends up and how you're using your data. It's just good data management practice.

Absolutely. And that's a very important point about code is that you can just store it in GitHub or wherever. Like it is very traceable, you can track the changes. And so you do have like that audit sort of built in which you don't get with a, with a point and click interface. Before you were talking about tech use cases of AI, I suspect you wanted to talk about non technical use cases as well. Like how are things changing for data analysis in the non technical sense?

In a non technical sense. I think it makes writing super, super easy. Email writing I think is one of them. Easy to check for typos, easy to rewrite your emails in a certain tone, easy to rewrite the same message to cater for different platforms. So I'll give you an example. I send out an email to many, many people. I think an email has a different tone compared to we use. Actually I'm not going to disclose what we use. Anyway, it doesn't really matter in the sense that it's Viva Engage, it's not a big deal. Actually Viva Engage is like Facebook for work, whatever. It's a social media platform that you use at work and obviously when you make a post it needs to be a little bit different. Use some emojis, you know, make it more social media. Like whereas if I start using those emojis in my email that I send out to people, it's going to look a little bit weird because I think there is a specific Comms platform for a specific purpose. So the same message, I can just ask AI to rewrite it for me. One catered to a social media post, one catered to just a traditional, more formal email. And then I can just type in whatever I want to in three, four sentences, and then AI will quickly generate two versions. And that immediately saved me, let's say, 20 minutes. Writing. That could be one of them. Audio transcription, recording meetings. Nowadays, who takes minutes anymore? Nobody. I remember when I started my career, I used to sit in the meeting room taking minutes, obviously, because I was quite new and I was quite junior and I was sitting there with my notepad, taking the notes, taking the minutes. Whereas now, I mean, there's no need. Meetings are timestamped. You have audio transcription, you have live transcription. Even better. And then after that, you can just ask whatever tool you use, it can be Microsoft Teams or it can be Zoom. You can ask the tool itself to summarize the meeting for you, give you the action point. So I think this is one of those things where it just helps so much, eliminate the really, let's be honest, boring tasks. Who likes writing minutes? Probably nobody. If you do, I don't know, comment below. How does it work? I don't know, but let me know if you do. So these are the things I use AI for a lot. Just writing, summarizing, meetings. Copilot is really good. So I use Microsoft Copilot a lot at work and it's very good at finding information across my mail, across my team's chats. For example, let's say if we've been chatting for a while and you sent me a hyperlink to whatever, something a long time ago, I can just ask Copilot, hey, can you find me that hype the last 10 hyperlinks that Richie sent me and they're all just listed?

Absolutely. Yeah, I'm right there with you. That I absolutely hate taking notes from meetings. So, yeah, having that done automatically, real. A real blessing there. And yeah, I'm sure there are one or two people who really, really love taking notes. But yeah, I guess you still can take notes if you want to. You just. Can I get the AI to do it for you as well as a backup? Okay, so I do have the idea of like, anytime you're dealing with text, you know, changing, like styles or whatever, communication styles, then that worked really well. So I guess on that note, do you see communication aspect of data analysis changing? Like, do people need different communication skills? You need to worry about communicating differently? What do you need? What Communication skills do you need?

As a data analyst, I always say this, and I may sound like a broken record if you watch my content, but you need to be able to deliver any type of insight in a way that a 6 year old could understand and in a way that would satisfy me or even someone even more technical as well. So if you really know your stuff, you can really dumb it down and you can make it so complicated that honestly only the people who are really, really up there in terms of technical expertise can understand. And I think that's it. If you are not an expert in your field, then you're just going to be talking left, right and center. You know, oh yeah, I did this and then you did this and then this. Whereas if you know what you're doing, it doesn't matter how complicated the analysis is, you can break it down super, super simple and you can make it really difficult as well in the sense that you can really elaborate on the details. So to me it's the ability to provide information to both technical and non technical audiences that's wonderful.

It's kind of like the holy grail of communication, being able to scale that depending on the level of technicality. Do you have any tips then on how you might go about converting a technical explanation into a non technical one or vice versa?

Let's go with customer Churn because I think it's simple to understand and everybody does it, or probably most people care about Customer Churn enough. At an executive level, I think people only care about the top level figures. So you probably want to tell them which segments churn the most, why they churn the most, and then you give your execs options. So option A will do this, option B will tackle this and option C will tackle this. That's all they want to know. They don't care about how you got to the numbers, what calculation you used, what, what's the formula for a customer churn, for example. And then if you go level down, people will be interested in the details. It's like, how did you create this tableau dashboard? Where did you get the data from? Is that like, did you validate the data? What calculations did you use to calculate customer churn? How did you segment your customers? What was your technique? And then this is the way you dive a level down and then explain specific things. And a level further down would be you take one segmentation and what you did with that specifically, like let's say for example, you segmented the customers by age and something else. Whatever that be, let's Say for banks, it can be a product, for example, so credit cards, loans, car loans, whatever, mortgages for example. And then you dive level down and then you take that analysis apart, you show the specific code used, you show exactly which system you pulled the data from, how you pull the data from. Did you use SQL or did you use some kind of third party system? Did you use reference data to validate? Did you do data quality checks, you know, accuracy, timeliness, all of those things. And that would be taking it really, really low level because people who do the technical bit, they'll be interested in that. And then people at the very high level, all they care about is who churns the most. How can I make them not churn? That's it.

I love that example and definitely something I've fallen over early in my career. Like you do not spend 10 minutes talking to a CEO about data cleaning. They do not care, they will not appreciate it. But yeah, the lower down the organization, the closer people are for the data, the more they're going to know about the technical details and the how rather than just the outcome. Beyond that, are there any other soft skills that you think are important beyond communication? If you're working in data analysis, I'll.

Say this soft skill that would apply, I think, to anything you do in life. Treat others the way you would like to be treated. I think that's it. I speak to everyone at work with the same respect, the same patience. It doesn't matter what level they are, can be the CEO, can be someone else, it doesn't matter. Obviously, if the CEO is asking me to do something, there is clearly more weight to it. But just because the person is the CEO doesn't demand weight, more or less respect than, let's say when I go to work and someone is, you know, cleaning the desks, I treat them with the exact same respect as my CEO because it's a job. We're humans and I think at the end of the day, as long as you're a good human being, that matters probably the most. So if you're likable at work and you're genuine and to achieve that, I think just simply treat people the way you'd like to be treated. I think that's a pretty good. I don't know if it's a soft skill to have or not, but I think it's a good characteristic to have and it'll take you far in your career if you're genuinely just a nice person.

Yeah, certainly everyone spends a lot of hours at work. You're going to be interacting with your colleagues a lot whether you want to or not. So yeah, better to be a good person. I think treating people respectively is good life skill rather than even just a career skill. So, yeah, I do like that idea. I'd like to talk a bit about getting a job as a data analysis analyst. So I guess the first thing is qualifications. What sort of qualifications are necessary to become a data analyst?

Now I'll talk about my qualifications. What I have right now, I actually have to think about it because once you have experience qualifications, they really go into the background. So I am currently AWS certified. AWS Cloud Practitioner certification. I believe that's what it's called. That's the official name. I also have a Udacity Data Engineering Nano degree. I'm also a Neo4J certified professional. And for those of you who don't know what Neo4j is, Richie, do you know Neo4j? That's the graph database, of course you know. Yeah, okay. No way I can catch you out. But yes, Neo4G is the graph database. So we use that at work as well. So I thought what's going to make me more credible than get certified, right? Nobody cared out of work because I already have a job. So honestly, nobody cared. So These are the three certifications I have. I also have the CFA Level 1 Chartered Financial Analyst from 2019. I think it was a while ago because I was working in investment banking and I thought that was the way I was going to go. So I think these are the four qualifications I have. I believe I have a master's degree in finance and economics, if that counts as a qualification, how important they are when getting a job. I don't want to lie to people and say that they are the most important because I don't think they are. I think experience beats anything and if anybody says otherwise, they can find me. I'll put my hands up. And I think experience does beat anything and everything. But if you don't have relevant experience, certifications are fantastic in a way that recruiters know what certifications to search for for specific roles. Like let's say you have the Power BI PL300 certification, or that's why you have the tableau certification, or that's why you have numerous AWS certifications. Because recruiters know that if you have that certification, you'll be at least be able to do the technical bit. Maybe you won't have the business knowledge, maybe you won't be able to come to work and solve problems immediately. Business problems. But if there is a business problem that people tell you about, they can have the confidence that you'll be able to use your technical expertise to solve that problem. So this is where I think certifications come in, where if you don't have relevant experience, certifications are the next best thing you can do and achieve and get.

Okay, so really it sounds like the certifications, qualifications, these matter a lot more at the start of your career when you don't have the experience, they'll get you that first role and then after that the experience is much more important. Talking of experience, I guess the best way to demonstrate this is perhaps through a portfolio. So talk me through. What do you need to have in terms of a portfolio to become a data analyst?

I would say go for anywhere between two to five projects. Choose tools that the company you want to apply for or the industry you want to work in use. So for finance it'll be Excel, for healthcare it'll be, I think, a mixture of Excel and Power bi, I would say. And SQL is always handy. So SQL is one of those that if you have a database and you cannot get the data out of the database, then you have no data. So SQL is super, super important. And then this is where I think depending on the industry, the VIS tool will widely differ. So I find in marketing, you know, it's more SQL plus tableau skills are really, really good. You can use Looker Studio, but that's just because of, you know, Google as an engine and then it really depends on the industry. So it's, it's so hard to say which projects you should build exactly. But in terms of from a high level, two to five projects, pick the tools your industry actually use. I would say for a data analyst role, if you know Excel, SQL, Power bi, tableau, you're covered. Because Python, in my opinion, is quite specialized in the sense that most data analyst roles don't require Python. They say it's a nice to have and genuinely it's just a nice to have at work. Python makes my life super easy because there are many things I can automate and just make it quicker. And if you use code, it's reproducible. So it just helps me do pretty repetitive tasks. I see it way more as a data science, data engineering skill set. Way more, not a data analyst skill set. When presenting your portfolio, build something that is super simple on the front end. So don't use a GitHub page with 500 lines of code. That's one thing you should definitely not do. For example, I've Seen so many bad portfolios. Honestly, I don't want to go into it. I'll go into some good ones. So let's say a clean Google Doc at its very simplest with clear titles, subtitles, and then you add your analysis into it. I myself use Notion and my portfolio is in Notion. So I like using Notion a lot. And actually my best product and my most popular product that has the best, best feedback and I've been offering for multiple years now. It's the ultimate data portfolio. It is built in Notion and it just, it's got such a nice front end to it and the back end you can store the code, you can tell a story and it caters for both technical and non technical people as well. So even if you know nothing about data analysis, if you look at my portfolio you'll know exactly what's going on because on the front end it's basically just like a Medium article. Super simple, easy to understand, no jargon, no heavy data analysis concepts. It's just this is it, this is what happened. I actually have a Stack Overflow project that I wrote on Medium where I analyzed, I think the Stack Overflow survey responses. And it's so simple that I mean, even if you knew nothing about Stack overflow, data science or data analysis, you would still be able to understand it.

Okay, so I like the idea you've got some, you want to have some demonstration that you can do SQL, you want to be able to demonstrate that you can use a BI tool and then you just want to have it somewhere where it's easy to read, I guess, regardless of the audience and probably not too heavy on the technical jargon because you want it to be read by like hiring managers who may not be that technical. Is that about right?

Yeah, yeah.

Are there any other tips you have for making your profile stand out when you are applying for a data analysis job?

Resume and LinkedIn resume is a must. A good LinkedIn profile is super, super nice to have. The way I look at the resumes is that it's a round where you can only fail because you never get the job after passing the resume round. But you can easily get eliminated for such silly things. You don't need to have a great resume to pass the resume round because mostly, I mean, many companies just use the ATS system and many companies have people who spend. I actually checked this so you can quote me on this. I think it's 7 seconds or 8 seconds that a recruiter usually spends on a resume. And if I think about it, I've Looked at resumes as well at work, and it's probably about right. Because if you know the skill set you're hiring for, all you're going to be looking at is experience, quickly looking at the keywords to see if you can find what you're looking for, and then if you don't find it, boom, you're on to the next one. That's how people look at resumes. Nobody actually reads the whole thing, I believe, or people at least who are, you know, into hiring because you have a limited amount of time. So the best advice I can give on the resumes is that only put in relevant information. A lot of people just think more is better. I don't think so. Less is better if it's relevant. So start with the relevant experience that you have on the top. Underneath that, I would put relevant education or relevant projects. If you don't have relevant experience certifications, those are really, really good. And then the skills you have. So let's say Power Bi, Tableau, Excel, what else is there? Let's say cloud skills, Google Cloud, or it can be Microsoft Azure or aws. Make sure you throw in as many keywords as possible. Because the people who glance through your resume, if you just throw in one or two keywords here and there, it's easy to miss them. But if your resume is loaded with those right keywords, then it's so much easier to spot. Right. It's almost like I cannot even miss it because if I read the top or the middle or the bottom, I can see that this person, let's say this person is applying for a Power BI developer role. I can see in the projects, in the relevant experience, in the certifications, in the education, in every single section that this person knows Power Bi and did something with Power Bi. The second thing is quantify things. So instead of saying, I did this, say I did this, and it led to whatever this. You identify the business problem. You did something, you did something. Not your team, your wider team, not we supported someone, or I or we led this work stream. I did this, I supported this, I had an impact. Highlight your impact. The recruiter and the hiring manager, they're not interested in what your team, what your organization did. They're interested in what you did. So focus on yourself. I think this is pretty sensible advice. Hopefully you agree. Yeah.

I mean, I have to say, it's wild that recruiters only look at your resume for seven seconds. Like, I feel like I've spent longer to, like, reading the back of like a food carton to Decide whether to buy, like for a trivial purchase. I spend longer than like that seven seconds. So yeah, that, that's completely crazy. But it does, I guess highlight the importance of making sure you just have the important information in there. And I do like the idea of talking about what you've done. Just a little simple like language change to say, I did this. That seems very important. So I guess suppose you do get that first round interview. Like what does the interview process for a data analysis role typically involve?

Depends on which company. So I'll pretend that this is a generic company. Usually you would get some kind of first round HR interview. That's usually the case where someone steps in from the HR team and all they're gauging is are you roughly the right candidate. So this person from hr, they're probably not going to be technical, but they work for the company and they've been hiring for tech for many, many years, so they know what to look out for. So let's say if I know nothing about data analysis, they'll be able to tell, like if I have no idea what I'm talking about. Then if you pass that, you usually get a second round interview. This is where things get a little bit more technical. So this is where someone, usually from the hiring team will come in and sit in with one or two HR reps. Again, hypothetical scenario. Let's just say this is a typical process. They'll ask you probably more technical questions. Now some companies make you do technical things, as in, I've even done some through zoom and whatnot. They give you a task just to test whether or not you can actually use the technical tools. And with those technical exercises, I do just want to say here two things. First of all, they want to put you under pressure. So don't expect to finish everything because you won't. Because the whole idea for them is to give you so many things that you will never finish it. But they just want to see you under pressure. Now the second thing is that I like the companies who let you use whatever you Want to Google AI ChatGPT like use the resources that are available to you because otherwise it's not really a real life scenario when you go to an interview and you have to memorize everything. So those are the two things about that. If you pass the second round interview, sometimes they give you a third round where you speak to the head of the team or someone who's quite senior at this time. Now this is where the interview can go anyway because sometimes they just want to judge your character. If you, if it's a smaller company, they want to see if you're the right culture fit. Sometimes you just get completely roasted. It happened to me before. I just sit on the interview and I'm like, what is this person asking me? And, you know, you sit there and you're like, oh, my God, what is going on? But those, to be fair, I have to say I like the interviews where they test intelligence rather than interviews where they ask me, tell me a time when you've overcome a challenge. You know, those cliche HR template questions. So I like it when people throw me curveball questions that are difficult questions that test truly just my intelligence. Because I feel like nowadays any skill can be taught on the job really quickly. If you're smart, the learning curve is steep, but with AI and everything available now, it's quick. So that's, I would say, a typical interview process. First round, hr, second round, probably more technical. Maybe they'll make you do an exercise that's my experience. And then I can give you a different one for how I started my career. Because in the US as well, you have these early career opportunities. And in the uk, where I live and where you're from, Richie, so we have these grad opportunities, and those work a little bit differently, and those are specifically for career starters. So I can expand on that if that is something you're interested in.

Sure, yeah. Maybe we'll come back to the graduate opportunities in a moment, but I feel like maybe you're a bit of a glutton for punishment, enjoying the hard questions. But I do think it's important for people to realize that if you do feel under pressure during a job interview, then that's kind of the point. And it's okay not to just freak out. They just want to see, can you work under pressure? All right, go on, talk me through. How do graduate opportunities differ then?

So grad opportunities in the uk, Speaking from experience, I know pretty much how it works. Super competitive. First of all, you fill out probably hundreds of these and then you complete some online assessments. There's some logical math and reasoning tests once you do that. In my time, I had to answer some culture fit questions and some case study questions, but it was very weird because you speak to yourself. So you have this online interview, but they give you the question. They give you five minutes to answer. You look into the camera like this and then you speak to yourself. You answer the case study and then you move on from that. You usually move on to an in person interview, and then you have an Assessment center where they put the best against each other. And let's say an assessment center would be a whole day event where they test you on so many things. Ten people go there and let's say three people get the job offer. So this is a typical grad experience from my time. Bear in mind that that was in 2018, so it was quite a while ago. But I believe speaking to other career starters, it's roughly along these lines nowadays as well.

Okay. Yeah, I imagine like the, the whole interview format hasn't changed that much over, over time. Like you're still going to want a similar sort of setup where you have go from sort of low level just HR interview through to some certain kind of assessments and then end up with maybe a more senior interview at the end. Okay, so I guess we talked a bit about maybe getting your first job in data analysis. What about for career switches? So suppose you've had a career in a area and then, okay, I really want to get into working with data. How does that transition work?

I always recommend people lean on their domain knowledge. So I've had so many people who had way more experience than me, which is crazy. I had actually someone specific who had a PhD, 25 years of experience in healthcare or something like that. They had more years of experience than almost I've been on this earth for, which is crazy to think. And they had insane business knowledge and I just asked them to build on that business knowledge, really lean on it, learn some technical skills to supplement it, and then go into a role. Because when you go into a business, the ability to solve business problems and identify them is so valuable. Common sense and intelligence, I feel like is so underrated nowadays in the sense that everybody's talking about implementing, implementing, implementing. But then imagine if you don't even know what questions to ask. Like you just go to work and you literally have no clue what to look for. AI will not solve that problem for you because every business will have its own unique problems. So let's say if you work in finance and you're looking for good investments, ChatGPT will not tell you what's a good investment because every bank is going to have a different portfolio. It may point you to the right direction in terms of what to look for, but it won't be like when you go to work, it'll just be completely different. And obviously large organizations don't feed their data into OpenAI's GPT models because it's just a black hole. They develop their own models. So then again, the GPT Models are not smart enough to actually, I think, deal with real life hard business problems. They're fantastic at so, so many things. Which is why I think for these people, the career switchers, that knowledge that they've built up over five years, 10 years, 15 years, that is invaluable. That is something that only other people who also want to transition into data analytics and have also worked for 5, 10, 15 years in that specific industry have, which when you think about it, is not many people. Yeah.

And I suppose even if the AI can answer the question, domain knowledge is still going to help you ask the question to the AI to help refine your knowledge as well. So, yeah, that does seem incredibly important. Are there any challenges you think there are associated with getting into data later in life? Like what do you need to be worried about?

Yes, absolutely. So I won't lie, it's not easy. But data jobs are out there, they are available. And I think when people say that it's hard to get a job, yes, it is, because data jobs are high paying roles. So if you think about Formula one drivers, Premier League footballers, NFL players, NBA players, it's hard because it's highly paid. So I'm not surprised that to get a six figure job you have to compete. If it was easy, we would all do it. So I think the difficulties of transitioning into data or getting a data job, it's the same level of difficulty as getting any other job that is also highly competitive. That's the way I think about it now. Covid was a different time, obviously, because during COVID I think a lot of companies thought that because we had 100x the data, we needed 100x the data resource, which in reality is not true. And then people got laid off and now with AI coming in, it is hard to get a junior job because companies think they can just automate everything, which is probably not the case. But it is really hard for career starters to get into data. But I think for career switchers, I think the difficulty is lower, like I said, because they have invaluable experience. Once you have that invaluable experience, you go to work and the people who can solve the problem think about the problem, ask the right questions and build a solution. Those people are so, so valuable to the organization because you'll find that many people are really good at managing, many people are really good at doing, but the people who are really good at managing and doing, which means you can think you can find problems, take the initiative and build a solution yourself. Imagine how valuable that is I can only speak for my wife, for example, who's a senior business analyst at a, at a medical technology. I believe that is a sector. That's what it's called. It's an American company. So she's a senior business analyst. What makes her senior? It's the fact that she can identify the business problems now and she can work with people to solve those business problems. And on top of that, she's a power BI expert. So she runs the whole team's power bi model, which makes her invaluable.

Okay, yeah, that does seem very important. You've got to have that balance between I can do technical things, I can do management things, and also, yeah, I've got the domain knowledge not to be able to ask the right questions. So you mentioned how things are tricky for junior analysts. So I've heard different opinions on if AI can do very junior things. There's no bond hiring junior people. You only want senior people in your organization. And then there's the sort of counterpoint that actually junior people don't have that sort of preconceived notion of how workflow should be. They're much keener to adopt AI because they don't know the old way of doing things. So do you have a sense of like, if you're a hiring manager, should you be focusing your efforts on recruiting junior people or more senior people?

It'll depend on your team. The thing is, I'm not a hiring manager and I was never in a position to lead big teams ever. So it's really hard for me to comment on something like this. I think if you're hiring someone who has relatively less experience, I think you hire on potential because nowadays, like I said, I feel like intelligence is so underrated. Everybody just thinks, okay, you've been doing this for 20 years, so you'll be great. But the way I think about it is that you've been doing this for 20 years, which means you're not up to date, you're set in your ways. Not everybody, not the majority, but this will be the case. Right? So having lots and lots of experience in the same field, in the same team is an advantage in the sense that you have so much business knowledge. But sometimes these people don't bring in fresh perspectives. So the way I see it is that the way I do things is non conventional at work because I haven't worked for long enough to do the conventional thing. I've only been in the corporate world for seven years and I say only in the corporate world, I See some dinosaurs who can literally. I see people at my company, you know, who had their 40th anniversary with the company. I'm not even 40. I mean, you know, so it's. It's one of those things that I think when you hire junior people, you must hire on potential, because people can really amaze you with what they can do if you just give the right person the right opportunity. And when it comes to a little bit more senior roles, I think it's very different for smaller organizations where you're looking for someone who's super hands on and can do everything, the role will be different compared to larger organizations where you're looking for a people leader. When you're looking for a people leader, I think experience matters so much because dealing with people at work is probably the hardest thing. Surprise, surprise, right? Like, you go to work, you have all these amazing technical skills, but then the hardest bits about working is actually attending meetings, getting the buy in, convincing other people that this is the right thing to do. So it really depends, Richie. I'm sorry. This is the most honest answer I can give you.

No, that's fair. That under different circumstances, there's gonna be different needs there. But yeah, I do like the idea. Well, you know, you feel like you have hard technology problems, but actually your colleagues are even more of a problem.

Not. Not a problem. No, no, no. I wouldn't say a problem. The way I approach this is that I like working with everybody at work, but there are certain people I like working with more than other people. So I like everyone at work.

It feels like there's an undertone of gossip there. All right, to Rafa, just to leave. What are you most excited about in the world of data at the moment?

I'll say I'm excited for the use of genai in a way that you don't need to use English to prompt anymore. So I'm excited for a genai solution that can sort of just look at what's on my screen for this is just an example. Or a Gen AI solution that can sort of look at my data sets or my work pattern, my workflow, and what I want to achieve as a business goal and just give me the solution. So rather than me prompting something along the lines of give me the top five best performing investments over the last six months, how about genai just figures out that I'm getting this data set. I'm trying to, like, look for bar charts and companies, and then just from that, it can suggest something to me. I'm excited for something like that.

So not even having to write the prompt, the prompt just appears automatically. It's like, okay, would you like to do this? I guess it's like the old clippy style thing from Microsoft Office where you start typing and then it pops up and says, oh, would you like to write a letter?

Yes, something along those lines. So that's what I'm excited about. Unfortunately, there's so many things I'm worried about with AI and I think it outweighs what I'm excited about by a mile.

Okay, go on, what do you worry about?

Oh, so I mean, I think, and there's already proof of this is that AI can choose to lie, choose to hide, choose to scam, which is a problem because I know currently we're using AI to test whether or not it can penetrate systems, right? Create fraud. But there have been case studies where the AI model penetrated a system and then when asked, the AI model said it didn't do it. And instead of trying to reveal the security flaws, it hid the penetration. So if you think about it like this, this is my biggest worry, is that if we make Genai truly to be like humans, humans on average, I don't think we're good in the sense that that's why I live in the, I wouldn't say the middle of nowhere, but you'll know this, that I live in a national park in the UK and I love animals and the more people I meet, the more I love animals. So this is why I have a concern around AI. And just on this note, I think this is why there are going to be different fields that are going to be emerging. And I already see that you guys provide training around this as well. So AI ethics, AI governance, super important, because if you don't understand that the models are biased, which most of them will be, because it needs to feed on some kind of a data. So all models, in my opinion, will be biased to some extent because it just depends on what data you feed into it. But it's how you refine the model over time to close the bias. That's what's important. Because if you do it wrong, it'll just expand and expand and expand and then eventually you end up with terrible models that give you terrible results that will make you do terrible things. Oh man.

I just realized we're ending on a really dark note there. But yeah, certainly, I think having a sense of responsible AI, of AI ethics, this is going to be incredibly important skills to deal with, particularly as generative AI becomes more and more widespread and in use, it's going to be a core skill for everyone.

Okay.

Actually, do you have a happy thing to end on maybe?

Oh, yeah, we need to end on a happy thing. So I think. Let me just end on this note because I think this is close to people who use datacamp as well. So your audience and my audience as well. I know it's really difficult to get a data job, but I personally know of successes and I know it's one of those things that if you just put in the work and you persevere and you're consistent, it is completely achievable. It is one of those things that when it gets difficult, don't just do endless tutorials, for example, go to data camp and you can, you can do projects, do real life, hands on learning, do something proactive. If you just keep on watching things forever, I don't think learning will stick. But if you actually do one or two or three projects where you have to think outside the box and do something a little bit differently, then I think getting not just data analyst jobs because the role of the data analyst is changing and will shift over time. It's the ability to work with data using AI. Nowadays, I think the opportunity is there. So anybody who tells me otherwise, like the people, hundreds of them, who said in 2022 November that I will lose my job, I'm still here standing and I'm proving them wrong in a way that I didn't think my job was at risk in 2022 November, because I knew that at the end of the day, it's just a different skill set that I'm gonna have to learn and adapt to. And the way I look at it, actually I welcome using AI because like I said, who likes writing minutes, I no longer have to do it, who likes debugging code, hundreds of lines of Python code and I have to find the problem, whereas now AI just does it for me. So AI in general is good. I think the only thing I wanted to highlight here is that the ethics around it and the governance is super, super important as well. If we can do that right, I think we can create amazing things and we have already done so, so many amazing things. So I know I said penetrating systems, but also fraud detection is so much better with AI. Easier, better, more efficient, quicker. So that's just one example. And there's so many other use cases, not just within finance, but within healthcare. I see companies already use AI to proactively try and forecast who might be at risk of something. So you can do something about it years ago before the actual thing happens. Right. Like, once you hit stage four lung cancer, there's nothing you can do. There's no AI model that will help you. But if you can actually utilize technology, use the data that you have and build a model that will predict it accurately who is more prone to having lung cancer, then you can do something about it five, ten years before that, and then, boom, you just gave someone else another 30, 50 years of life. And how can you put a value on that? So I think this is a pretty good, pretty happy note to end on, Richie.

Absolutely. Yeah. I mean, lots of great things there. Yeah. It is still definitely possible to get a job in data analysis. There's lots of cool stuff and frankly, lots of interesting new things to learn as the. As the job changes. And, yeah, I do idea of, like, AI for healthcare, like, improving people's lives. It's. Yeah, just an unambiguously good thing. All right, wonderful. Thank you so much for your time, Mo.

Thank you so, so much, Richie. And I'm sure we'll speak.

It.