# How Much SQL, Python, Power BI You ACTUALLY Need in 2026 (AI changed the answer)

https://www.youtube.com/watch?v=2q7nQnfl_y0

[00:00] How much sequel is enough to get hired in 2026?
[00:02] How much Python?
[00:02] How much Power BI?
[00:04] I get asked this in the comments all the time.
[00:07] And the answer has changed, you guessed it, because of AI.
[00:09] Postings for entry-level jobs in the US [music] overall have declined about 35% since January of 2023.
[00:16] [music]
[00:18] Beginner level is not quite enough anymore.
[00:20] There's also the fact that someone has to be able to check AI's work.
[00:23] So in this video, I'll show you where enough is for each of the tools that I mentioned, as well as which skills are worth a premium.
[00:29] Back when I was a teacher with zero tech background, I would have loved to know this.
[00:33] Now I work for a Fortune 300 company and here's what I see changing from the inside.
[00:40] And quick thanks to TWI for sponsoring this video.
[00:42] More on them later.
[00:44] All right, so first, let's talk about why beginners aren't getting hired.
[00:47] Before, if you learned the basics of SQL, Excel, and a BI tool, that was enough.
[00:52] To be honest, the bar was relatively low.
[00:54] It's always been competitive, but the way to get in was not complicated.
[00:58] Now, simply knowing
[01:01] Those skills is not enough.
[01:03] But again, I'll say this, it has always been hard.
[01:05] The focus has just shifted now.
[01:08] AI now does many of the simple tasks that juniors used to do.
[01:12] So companies are hiring fewer juniors.
[01:14] And what is it that juniors did?
[01:17] Well, simple querying, if you just had like a quick request, like a lot of ad hoc stuff, uh really simple report building.
[01:23] And this is why just knowing the tools alone are not enough anymore.
[01:27] Again, now that the work is going away, honestly, you should be skeptical of a lot of the headlines you see.
[01:34] The reality is that the net count of data analyst jobs overall it's going up.
[01:36] So that's a positive sign.
[01:38] Again, this doesn't mean that fewer analysts exist because there are more open roles than ever before.
[01:44] It just means that the bar has gotten a little bit higher.
[01:46] And you've probably heard this mantra, but it bears repeating, the analysts that will be replaced are going to be replaced by analysts who use AI, not by AI itself.
[01:54] And the part that the people miss is that checking AI's work is harder in a way than doing the actual.
[02:01] Work.
[02:03] Or at the very least, it carries more risk.
[02:05] Checking AI's work is not necessarily harder in the traditional sense, but what makes it challenging is that you could be moving further and faster in the wrong direction without knowing it.
[02:14] Or you can ship something completely wrong if you don't do the proper validations.
[02:18] And validation is the name of the game when it comes to AI.
[02:24] Hiding AI's work is crucial, and if you don't do it every step of the way, you could end up down a deep rabbit hole that is mostly wrong and no one even knows it.
[02:33] So, the bar has moved up.
[02:36] And the slowest way to clear that bar is by doing what most people do.
[02:38] Grind through another 40 hours of courses, hit course purgatory, and still have no idea if you're actually job ready.
[02:46] But the fastest way I know is learning straight from analysts who are already on the other side of that bar.
[02:50] People doing this work right now who can tell you what actually matters in 2026 and what's a waste of time.
[02:57] And that's exactly what today's sponsor, TWI, is all about.
[03:03] They're hosting a massive data and AI conference from September 20th to the 25th in Anaheim, California.
[03:07] It's built for the stuff that we're talking about in today's video.
[03:12] It has over 50 classes taught by people doing this work every day.
[03:17] Not theory people or influencers who haven't actually worked in this space for years.
[03:21] There's a class on data analysis LLMs, a machine learning class with Python made for beginners, and one on building dashboards in the age of AI.
[03:32] On top of that, you've got data leaders from companies like Apple and General Motors showing how they actually run this stuff internally.
[03:38] And what's great is that most companies have a training budget for this.
[03:41] Most people don't realize it.
[03:43] So, ask your manager.
[03:46] Check out my link in the description for $100 off registration.
[03:48] You can save that money and spend it on massage instead because I know your back hurts from sitting all day long doing lots of fun code stuff.
[03:57] So, in order to check if you've learned enough, there's a simple test.
[04:01] If you pass it, you'll know exactly where you stand and that will.
[04:04] Make it easier to spot the gap.
[04:07] Can you catch AI's mistake?
[04:09] So, there's a time where I asked AI for total revenue by customer.
[04:12] I gave it a clean prompt, gave me a clean-looking result.
[04:17] The number came back a little too big, just enough to feel off.
[04:22] The cause of this was an improper join that duplicated orders across every product.
[04:24] I won't get into the details, but the SQL ran perfectly.
[04:28] It just answered the wrong question.
[04:30] The reason I caught it is because the total didn't match what I know the business did that month.
[04:35] I almost looked really stupid, but what I had to know was the join logic plus the real business numbers.
[04:41] That's the gut check.
[04:43] That's where business context is so important.
[04:45] If AI's answers always look right to you or you never question it or validate it, then you're probably not ready yet.
[04:52] And it's not about how smart or clever you are.
[04:54] It's that looks right and is right are two different things.
[04:59] The only way to tell them apart is knowing the work well enough to be suspicious of the results that you get from AI.
[05:03] Analysts.
[05:05] Getting hired, don't just trust the output, they validate it.
[05:08] They see a clean result and still ask if the numbers make sense.
[05:10] Question everything.
[05:13] And if you want to work efficiently with AI, you need a clear baseline.
[05:15] So, let's go tool by tool.
[05:18] Let's examine where enough is.
[05:21] So, let's start with Excel.
[05:23] I see the base minimum things that you need to know in Excel for it to be enough are things like XLOOKUP, pivot tables, how to handle and clean a messy data set, basic visuals, and formatting.
[05:35] That is the floor.
[05:37] What still matters is knowing which functions answer which questions.
[05:41] And of course, the almighty pivot table is something you should know how to throw together quickly.
[05:45] What AI handles now are the really complicated nested formulas.
[05:48] You could at least get a start.
[05:49] Let it write them for you and tweak accordingly if needed.
[05:54] Your job is judging the output and planning the analysis itself.
[05:59] But it is true that all roads lead back to Excel, so in my opinion, you better know it.
[06:02] Let's talk about SQL next.
[06:05] So you should still be able to read SQL pretty fluently.
[06:07] Need to be able to write well enough to fix what AI hands you.
[06:12] But you don't need to be able to write 50 lines from a blank page because after all we do have AI.
[06:17] But one thing you should be pretty comfortable to is how the different types of joins work because this is where things can break pretty often.
[06:24] But I'd say a lot of the same stuff applies here.
[06:26] You should of course know the big six, select from, where, order by, group by, having.
[06:30] You should still know window functions.
[06:32] This is going to help you to understand what AI is giving you.
[06:37] Just because AI is helping you write a query doesn't mean you shouldn't know how to interpret it.
[06:41] Okay, let's talk about Python.
[06:42] Do we analysts even need it?
[06:43] How much?
[06:44] [clears throat] My honest take is that most analyst roles don't need deep Python.
[06:48] You will see roles that need it, depends on the type of role.
[06:52] I'd say if you're going to learn it, you should know enough Pandas to read and tweak AI's cleaning scripts and enough comfort to not panic when you see it.
[07:00] Heavy Python postings tend to be analytics engineer or data scientist roles wearing an analyst title.
[07:05] It's often a different job.
[07:07] When it comes to Power BI and
[07:08] Tableau, you should definitely learn one of them.
[07:11] My recommendation would be Power BI because Salesforce is not doing great things with Tableau lately.
[07:17] I would pick one and go deep.
[07:19] Shallow in two tools is not as well as really deep in one.
[07:22] I will say Power BI also tends to have more postings nowadays.
[07:27] It plays very nice with Excel and knowing DAX is a real differentiator.
[07:32] But I also think it's good to know Power BI because of how it fits into Microsoft ecosystem.
[07:36] I plan to talk more about how Power BI integrates with GitHub Copilot, for example.
[07:40] Stay tuned for that.
[07:42] But in terms of what you should know, should definitely have a resume ready depth, which means knowing how to put together a clean star schema model, knowing basic DAX measures, measures versus columns, what to use what where.
[07:54] You should know your way around Power Query pretty well, and you should have some dashboards in your portfolio that aren't just default or guided project.
[08:02] Which tool are you learning right now, or which one are you stuck on right now?
[08:05] I'd love to know in the comments and have some discussion about it.
[08:07] But, so those are the tools.
[08:09] So, let's turn that into a job.
[08:11] Let's talk about how to actually get the job.
[08:13] So, what goes in a 2026 portfolio that wasn't in a 2023 one?
[08:18] So, I'd say back in 2023, a handful of dashboards and a clean SQL project was fine.
[08:24] In 2026, I'd say that you need to be able to show more judgment.
[08:25] You should have an end-to-end project, maybe even one that incorporates AI in some kind of way, and show a complete workflow with plenty of back-end.
[08:33] The same projects as before, kind of, but narrate the thinking more than you would have in the past, not just the result.
[08:42] But, essentially, I still think that two to three projects is enough as long as you're showing outcomes and judgment rather than just a bunch of charts.
[08:49] When it comes to LinkedIn, you want an optimized profile for the role that you're going for.
[08:54] I always recommend posting about what you're learning, building connections, reaching out to recruiters and hiring managers.
[09:02] All of this just makes it easier to find you.
[09:04] It also builds your network, which is an investment towards the future.
[09:06] But, I'd say applying with focus is really important as well.
[09:08] I'm a
[09:10] I'm a big fan of picking a type of role that you want, like let's say a sales analyst type role, and making your resume all about that type of role.
[09:18] That's a strategy I've employed multiple times that has worked for me.
[09:20] So, rather than just having a super general data analyst resume, you're picking a sub-niche or niche in data analytics and going all in on that.
[09:30] Hey, real quick, I make these videos so that people without tech degrees get a fair shot at the data field.
[09:35] And if that's your goal, too, whether it's for yourself or for others, please subscribe to keep the channel going and help out the cause.
[09:41] So, again, let's recap how much for each tool.
[09:43] For Excel, XLOOKUP, pivot tables, cleaning and formatting, visuals.
[09:48] SQL, you still want to be able to read it fluently.
[09:51] You want to be able to interpret what AI is giving you.
[09:54] Python, I'd say light pandas and that's going to do a lot for you for most analyst roles.
[09:58] And then Power BI, just pick the one tool, know how a data model works, know how Power Query works, know some DAX, how to clean a data set, and be able to make killer visuals.
[10:07] So, now you know where enough is.
[10:09] You have.
[10:11] Two options.
[10:13] Option one is to stay at tutorial level and compete with AI for the same tasks.
[10:18] And option two is learn just past the line I showed you and become the person who checks AI and can do those things really, really well.
[10:25] Again, my link to the TWRI conference is in the description.
[10:27] You'll get a $100 off for taking advantage of that link.
[10:31] And if you're wondering which AI tools are worth your time, I tested a whole bunch of them and only five made the cut.
[10:37] Check out this video next.
[10:39] I think you might like it.
[10:40] Thanks again for watching and I'll see you in the next video.