If you are coming up as a new analyst right now, the picture you are getting from outside the field has to be confusing. You read every other day that AI is about to flatten finance, that the analyst role is on its way out, and that anyone planning a career around spreadsheets has not been paying attention. Then you open a job board and Excel is on more or less every posting, sometimes in the headline. So which is it?
I want to be honest that even the people who make a living teaching this material are not entirely sure how to answer it. Tools are changing under our feet faster than we can update the course outline, the role of the person at the keyboard is being redrawn in real time, and the trainers who used to have a clean story to tell have mostly stopped pretending we do. We are improvising, the way you are.
So I want to write down what I would actually say to a new analyst who asked me the question, knowing the picture is unclear and that I am working it out alongside everyone else.
Be less long-winded than the robot
The first thing I would say is the one that sounds the least technical:
Do not sound like a robot. Really, actively, do not.
The reason this matters more than it used to is that everything coming out of AI is long. Emails that should be three sentences run twelve. Memos hedge for two paragraphs before they argue anything. That style is now the default sound of professional writing for anyone who learned to write with ChatGPT open in another tab, which is to say basically everyone entering the workforce.
The opportunity here is enormous, and I do not think it is overstated to say that clear, direct, human writing has never been more valuable than it is right now. If you can write three sentences that land where eight would have wandered, you are immediately doing something the AI is not doing for the person sitting next to you. A senior partner reading two memos can tell within a paragraph which one came out of a model and which one came out of a person who actually thought about the problem.
This is the part of my advice I have most changed my mind on. I used to treat communication as a finishing skill. But these days, it is closer to the work itself, in that being readable is now one of the few things a human can offer that a confident, verbose model cannot.
Be fearless. Drivers wanted.
The second thing I would say is to throw yourself headfirst at whatever topic is in front of you, without waiting for permission.
The old Volkswagen tagline was “drivers wanted,” which I have always loved and find increasingly more relevant. The analysts who come out of this stretch in good shape are going to be the ones who got into the seat, pushed the pedals, and saw what the car did, rather than the ones who waited for someone to hand them a syllabus.
The reason this works now in a way it did not before is that the experts are learning right next to you. There is no longer a quiet room where the senior people have already figured out Power Query, Power Automate, Python in Excel, and Copilot, and are gatekeeping the answers until you have earned them. Those tools are new, the integrations between them are newer, and the AI layer on top of all of it is newer still. The person two rungs above you is reading the same blog posts you are, often the same week, and is probably as unsure of what is coming next as you are.
That changes what fearlessness costs. Trying a new tool used to mean stepping into a room where everyone else already knew what you were trying to learn, which is a hard thing to do when you are new. But these days, you can wander into Power Query on a Tuesday because it came up in a meeting, ask Claude to walk you through what just happened, and be reasonably useful with it by Friday. The trainer who would have told you ten years ago that you needed to learn lookups first is not waiting for you to finish lookups. They are off learning whatever came out last week.
If the experts are improvising, the new analyst gets to improvise too, and that is permission worth using.
Build a foundation anyway
All that said, you still need a foundation.
AI or not, the only thing between you and a wrong answer in front of a senior partner is your ability to read what is in the workbook. The list of things you need to actually know to do this effectively is often not long, things like:
- Cell references and how they propagate when you copy a formula
- What an Excel Table is and why it matters that your data lives inside one
- How a PivotTable summarizes data, and what assumptions it is making while it does so
- The way Excel handles dates underneath the surface
None of this usually shows up on a LinkedIn skills page or gets its own podcast. But it is the difference between trusting your own work and crossing your fingers, and that difference matters more now than it did before, because the work you are checking is increasingly work an AI handed you with confidence.
These are the moves a new analyst should be able to make without thinking, the way a musician thinks about scales. If you cannot read a formula someone else wrote and tell me what it does, you have not built a foundation. You’ve just stacked up tutorials, and AI will beat you there every time.
The path is not a staircase
The last thing I would say is that the shape of the learning path has changed, and the new shape is friendlier to a curious newcomer than the old one was.
When I started teaching this material, the path looked like a staircase. Functions first, then PivotTables, then Power Query, then maybe Power BI, then Python if you wanted to keep going. Every step is sitting in my LinkedIn Learning catalog and in a chapter of one of my books, and it made sense at the time, because the tools were gated behind each other in a way that more or less forced the sequence.
That gating is gone, and the path now looks more like a small core with a series of side quests around it.
You hold the core close, because that is what keeps your work trustworthy, and you wander into whatever side quest the work in front of you happens to need. Regular expressions one weekend because you saw someone use them on LinkedIn. A Power Query merge on a Tuesday because a colleague mentioned it at lunch. A Power Automate flow because you got tired of saving attachments by hand. Nothing has to come in order, and nothing has to be finished before the next thing starts. You are filling out a map rather than climbing a ladder, and the map is yours to fill in based on what your actual work needs.
This is friendlier to the newcomer because it does not punish you for being curious about the wrong thing first. There is no wrong thing first anymore.
Conclusion
So if you are looking at the job postings, seeing Excel on every one of them, and trying to square that with the headlines telling you it is all about to be automated away, here is the version I would offer.
Excel is on the job postings because the work is still there to do, and the people doing it are still going to need someone in the seat who knows what the cell is doing. Trainers like me are not entirely sure what the next two years look like, and we are improvising, and that is the honest answer rather than a problem to hide. If you want the short version of what to do about it:
- Write like a person: Everyone else is letting the robot write for them. Three clean sentences will get you further than three padded paragraphs.
- Get in the seat: The experts are improvising too. There is no longer a quiet room of people who have it figured out.
- Hold the core close: A confident wrong answer is harder to catch than a tentative one, and the foundation is how you catch it.
- Treat the rest as side quests: The map is yours to fill in based on what the work in front of you actually needs.
The car is sitting there… Drivers wanted.
If that is the kind of analyst you want to become, that is most of what I work on. You can read more about how I work here: