223: The Data Analyst Role is Changing. Here’s My Advice To Beginners.

The data analyst role is changing, and it can be a little bit scary, especially if you're thinking about someone breaking into the field. It honestly feels pretty daunting and honestly probably impossible. But it's not impossible, and in today's episode, I wanna break down what people are saying online, what people are saying on Reddit on how the data analyst role is changing and what you should do because of it. So recently on the subreddit DataAnalyticsCareers, someone posted this, and I want to actually react and see if this person's giving sound advice or not. So let's go ahead and get into it So the first thing that they say is to start top-down, not bottom-up. Most people start with tools. They open up a SQL course, memorize syntax, and then wonder what to do with it. That is backwards. Before you learn a single tool, research requirements and responsibilities of data analyst positions in companies you want to work at. List the skills they ask for, list the responsibilities, and then try to understand what the job is actually for. Ask yourself which industries actually industry interest me, what does a day-to-day look like in the companies that I am targeting? What problems does the analyst solve for the business? What is the goal of the role? This will give you a global view and the idea about the job. This is more for the private sector companies, but in the public sector, some things will still be applicable. Okay. I totally agree with the idea of we're not doing data analytics to analyze data, and you shouldn't just learn data skills and data tools to learn data tools. Like, they should always be used to solve business problems in some sort of a way. So we don't wanna just like memorize syntax, we don't wanna just write SQL queries. We're always writing SQL queries to solve business problems. And the quicker you get that in your mind, the better. So I agree with that. I do think it is a little interesting here that they're saying, "Hey, go look at the companies you're actually interested in, and list... look at what skills they're requiring." And I think that's important to like actually look at job descriptions and see what is actually being in demand, 'cause you might think, "Oh, Python's really important to learn." Guys, it's really not. Uh, like I think it's only like 80% of data analyst jobs don't require Python. So if you're gonna spend so much time learning Python for only apply for those 20% of the jobs, it might not be worth it, and that's my, you know, my take on it. I've done a bunch of episodes in the past about what skills you should actually learn. You know, you can go to finddatajob.com and click on our s- uh, skills report to actually see what skills are in demand right now. But I think that's important to do, but I wouldn't necessarily just do it for the companies you're interested in, because that might change over time, and also it's like a small subset. Like let's say you really wanna work at Meta, and their data analysts might use Python. I don't know. If you're really set on working for Meta or a certain company, that makes sense. But if you're open to like working for any company really, like I think instead of just looking at a few companies you're interested in, you should look at the aggregate. And so I think that's why our, um, skills report at finddatajob.com is really useful, 'cause that's like on average what skills should you actually learn, and the answer is Excel, SQL, and one BI tool like Power BI or Tableau. That's basically the, the short answer, but you can go check it out, the long answer, on our website. Also, like what industries actually interest me, I think that's one way to look at it, but also I would say what industries are interested in me. Because it's like a lot of us, I mean not a lot of us, but me for example, and I know I've talked to a lot of you guys, are interested in sports analytics. Like, "Oh, I love sports. It'd be so fun to be, you know, one of these analytics guys for the NBA or for the NFL or something like that." And that would, would be fun, but the truth is, those positions are so rare and they're so in demand that they're impossible to land, and they don't pay particularly well most of the time. So it's like, yes, you would be interested in those jobs, but it's probably better to reverse it and think, "Well, what industries would be interested in me?" And usually it's what industry you're coming from. Like what's your domain and what's your degree, and those types of things. So when I was a chemical lab technician, it's like, oh yeah, I could have tried to land a sports analytics jobs and that would've been great. You know, maybe their job description say you have to learn R, and I would've spent all this time learning R, but the truth is they would probably never be interested in me, and so I just wasted a bunch of time. And so I think it's actually better to reverse this and start with the, you know, what companies and what industries are interested in me, and then try to focus on that, unless you're really set on, like, a specific j- company or industry. But I think most of us aren't really that case. Also, this is more true for the private sectors versus the public sector. I don't know what they're saying there. Like, I don't feel like there's that big of a difference between the two. Uh, like they both post jobs online. Like, I don't get why making the company public or private, that actually changes. But I don't know. So overall, I'd give this b- advice maybe like a B plus. I think for the most part it was right on, but there were some small things I would make some changes on. Let's move on to their next paragraph. "Learn how businesses make money before you learn how to query a database. This is the whole thing nobody teaches, and it is the thing that separates analysts who get promoted from analysts who stay stuck executing requests." We have that classic AI line right there, right, with the double dash. I forget... Em dash is what that's called. Also, I thought this was for people who are pivoting in and not for people who are getting promoted. But okay, I'm gonna... I'm getting distracted here. "Every business runs on the same fundamental loop: attract customers, convert them, deliver value, retain them, grow revenue. The department you work for determines which part of the loop you spend most of your time on. Your analysis will always serve one or many of these stages. Once you understand that, you can think about data as numbers. You can stop thinking about data as numbers and start thinking about as evidence for decisions." I do like that line. That is a great line. "Four questions to ask before any piece of analysis. What is the business actually trying to achieve? The real goal, not the metric. Who will use the analysis? A CEO and product manager need completely different things from the same data. What decision does this analysis need to support? Analysis without a decision attached is just exploration. What action will they take after seeing these numbers? If the answer cannot change anything, the question is not worth asking. Build this thinking into every project from day one, and you'll be ahead of most junior analysts within six months." I think this is pretty solid advice, especially those four questions that they listed there, like what are we actually trying to solve with this analysis? Who is this analysis for? How much level of detail do we need to have in there? I think that makes a lot of sense, and he or she is right that the quicker you understand that, the better analyst you will be. I think trying to study how businesses make money is an interesting concept. I don't necessarily think every business runs on the same fundamental loop of attract customers, convert them, deliver value, retain them, grow revenue, or at least those stages are very convoluted, and I think that's an oversimplification of how businesses work. Like, yeah, like for instance, if you do like a D2C, direct to consumer brand, and like I'm selling, you know, water bottles, like that makes sense. But like I've worked for companies who don't care about growing their revenue. They just want to be acquired someday, and that's how they're going to make money is eventually via an acquisition. And so they're not concerned even about revenue. Like that, that's a real thing, and that's, you know, there's lots of companies like that. Also, like I worked for ExxonMobil, like one of the biggest oil, uh, producers or gasoline producers in the world, and In terms of like, you know, this method that they're talking about here, I don't know. I was so far away from that. Like, I was in research and development. Like, I guess research and development maybe would be like in deliver value stage, but I think this is an oversimplification of how business works. But it is important to remember that we're only doing analytics to help the business usually make money, and I hate to say that. Like, it could be to save lives, it could be to help people improve their lives, but most of the time it is to make money. And so you always need to tie your analysis back to a dollar bill. That's really important, and the quicker you figure that out, the better. So if you're trying to land a data job and, you know, you're unable to tie what you've done historically to dollar bills, that is difficult to perform well. So the quicker you can kind of bridge that gap, the better. Moving on to point number three: learn tools to answer questions, not to memorize syntax. Once you understand the business context and the types of questions you'll be asked, tools become obvious. I don't think there's any way for you guys to be able to predict what, what questions you'll be asked, unless, unless from the job description. Sometimes from a job description, you can actually pull out what type of questions you would be asked, but I would say that every job description and every role is so unique that, like, you can't just b- be like, "Oh, these are the type of questions I'll be asked," you know, different types of roles. Like, it's so different every place you go. That being said, like, there are some analyses that are very common depending on where you go. So like, for example, if you're gonna work for Google or you're gonna work for Amazon or you're gonna work for Meta, like one of the things they're always thinking about is AB testing or hypothesis testing. Like, if we change this on our website, do we get better results? Yes or no? If you work for one of those companies, you will probably face some sort of question like that. So there are some generic questions like that, but every other role I've worked at has been so unique. It's kind of hard to predict just off of a job description. Okay, you learn SQL because you need to query customer data. You learn Python because you need to clean a dataset or automate a report. Uh, okay, this is dumb. Like, no. I could clean a dataset and create an automated report in Excel. I could cr- clean a dataset and do it in SQL. Like, okay, I don't think that's sound advice. Like, you learn Power BI because a stakeholder needs to interact with a trend, not just read a table. Once again, I could do that in Python, so I don't know. "Learning a tool without a question to answer is why most people plateau. The syntax does not stick because there is nothing to attach to it." I don't know, that sounded like an AI-generated sentence. "Use AI and Kaggle to find datasets that match your target industry." I like that advice. It's always good to find unique datasets. "Build projects around real business questions." I like that advice. "What is driving churn in a subscription business is a better starting place than let me practice SQL joins." That is probably true. Like, I think that is sound advice that, like, if you can actually do real-world projects with real-world data, that's, that's better than just doing pointless exercises online. That's one of the reasons we created the accelerator, and we do the different projects that we do with real-world datasets to kind of get you in the habit of doing this. All right, next point here. "Talk to people already doing the job. There is more useful information in one honest conversation with a working analyst than in 10 hours of tutorials." Eh, debatable. "Threads, LinkedIn, Reddit, you will find those willing to help you if you ask a specific question rather than a generic one." I mean, that is true. Like, the more specific you can ask, the better. Like, the more specific advice you're going to get. Like, the more context you give someone, the better. I can't tell you how many, like, "How should I break into data?" questions I get, and it's like, "Okay, well, who are you? What do you do? What do you want to do?" and those types of things. So the more specific you are, the better. "Prepare in advance, though." And, uh, that's the whole post, and that got 500 upvotes on, uh, this subreddit. I mean, I think it's sound advice, but it's not really even that amazing. It's like, okay, you need to make sure that you're tying your analysis to dollar bills. Okay, I agree. And do real-world projects with real-world data. I mean, that's sound advice. Like, you can't go wrong with that advice. But what does that actually mean to you guys who are trying to land a data job? Like, you guys aren't even in the role yet. Like, how is this actually going to change? And one important thing I think this person's really missing out on is why this is actually changing, and the truth is it's 'cause of AI. AI is changing so many different things, the way we analyze data, the way we interact with data. And I still think it's really important to learn the fundamentals first because otherwise I think it's going to haunt you down the road where you can't actually fix things on your own if AI gets really expensive, or you can't actually catch, like, any errors that AI makes, 'cause AI does make errors all the time. So I think it's first important to learn the fundamentals of the business and actually, you know, like, okay, how does this business operate? How does data support business? That's great. Then make sure you know the fundamentals of data, like columns, rows, how we actually analyze data, when to use a line chart versus a bar chart, why we don't really like pie charts in the data visualization community. The basics of all things data fundamentals. And then I think it's really important on top of that to learn how to do things in AI, because I do think the future is very AI-focused, and that's one of the things that I'm trying to talk a lot about on this episode and on this platform. So if you guys want to learn more, I try to talk about AI every single week in my newsletter. You can go to datacareerjumpstart.com/newsletter and subscribe for absolutely free.