I used to think being a data analyst meant being good at Excel and knowing SQL. That was 2019. I was wrong.
Three years into working at Morningstar, commuting from Kalyan to Mumbai five days a week, I've learned that the job has shifted. Completely. The analysts getting hired now—the ones with offers from ICICI Bank, Goldman Sachs, and startups paying ₹18–25 lakhs right out of college—aren't the ones with the fanciest certifications. They're the ones who've figured out which skills actually move the needle in 2025.
Let me walk you through exactly what's working right now. Not theory. Real stuff I use, with real numbers.
1. Product Thinking Over Pure Technical Skills
This one surprised me. My first data job was all about dashboards and reports. Build the dashboard, hand it over, move to the next one. Three months in, I realized something: nobody was actually using half of what I built.
Here's the thing—companies in 2025 don't need more dashboards. They need analysts who understand why someone is asking for data in the first place.
Understanding the Business Question First
At Morningstar, I work with fund managers and advisors. A manager came to me last month asking for "daily performance metrics." Sounds straightforward, right? Wrong. If I'd just built the dashboard, they'd have gotten the wrong numbers.
I spent 20 minutes asking questions instead:
- What decision are you making with this data?
- When do you need it? (9 AM? End of day?)
- Who else looks at it?
- What happened the last time you made a decision based on similar data?
Turns out, they needed something completely different. Not a daily dashboard. A weekly summary that showed performance against three specific benchmarks. The technical work was 15% of the actual value I provided.
This is what gets you hired in 2025. Not "I know Tableau" but "I ask the right questions before building anything."
The Skill You Actually Need to Practice
Product thinking is learnable. And honestly? It's much harder than SQL.
Start here: Next time someone asks you for data, pause. Write down three questions before you touch your laptop. Do this for a month. You'll start seeing patterns in what people actually need versus what they think they need.
2. Python and Automation (But Strategic, Not Showy)
Most junior analysts I've mentored try to automate everything. I watch them spend 40 hours building a Python script to save someone 2 hours a week. That's not a win. That's ego.
The skill that matters is knowing what's worth automating and what isn't.
My Framework for Automation Decisions
I use a simple rule: I only automate if the task meets two conditions:
Condition 1: It happens more than twice a month.
Condition 2: It takes more than 30 minutes when done manually.
That's it. Two conditions. If both are true, I automate.
Real example from my work: Every month, I was pulling data from three different sources—Morningstar's internal database, a third-party API, and an Excel file someone sent via email. Merging them took 90 minutes. This happened every single month.
Automation math: 90 minutes × 12 months = 18 hours a year. Add setup time (maybe 15 hours). Net gain: 3 hours a year. Still worth it? Actually, yes. But here's why—those 3 hours aren't the real win. The real win is that I can't mess it up anymore. Manual work = risk of errors. Automated work = consistent, repeatable, reliable.
I built a Python script using pandas. Took me about 15 hours. Now it runs every month automatically, pulls data, cleans it, merges it, flags inconsistencies, and sends me a summary email. I built it once. I maintain it maybe 30 minutes a month.
What Python Skills Actually Matter
You don't need to be a software engineer. You need to be comfortable with:
- pandas — manipulating and cleaning data (this is 70% of actual data analysis)
- API handling — pulling data from APIs instead of waiting for CSV exports
- scheduling — using cron jobs or task schedulers to run scripts automatically
- basic error handling — if something breaks, it logs the error instead of silently failing
Companies love this because it scales. One analyst with Python skills can do the work of two or three analysts without Python. Your salary reflects that immediately.
3. SQL That Doesn't Feel Like SQL
Let me be real—everyone says "learn SQL" and 60% of people stop after basic SELECT queries. That's not enough anymore.
The analysts getting hired are the ones who think in SQL. Who can look at a business problem and instantly know: I need a left join with a window function and a CTE.
The SQL Skills Hierarchy
Here's what I see from junior analysts applying to work at Morningstar:
- Level 1 (Basic) — SELECT, WHERE, basic joins. Gets you past the resume filter.
- Level 2 (Competent) — GROUP BY, aggregate functions, subqueries. Gets you an interview.
- Level 3 (Hired) — Window functions, CTEs, performance optimization. Gets you the job.
Most candidates are Level 1. Some reach Level 2. The ones who actually get offers? They're Level 3.
Let me give you an example. I needed to analyze fund performance over rolling 3-month periods for every fund we track. That's thousands of funds. Simple approach would be nested queries and multiple passes through the data. Slow. Inefficient.
The right approach uses a window function:
SELECT fund_id, date, returns, AVG(returns) OVER (PARTITION BY fund_id ORDER BY date ROWS BETWEEN 89 PRECEDING AND CURRENT ROW) as rolling_3m_return FROM fund_performance
One query. Calculates rolling returns for every fund in seconds. This is the difference between "I know SQL" and "I solve problems with SQL."
How to Actually Get to Level 3
Stop doing tutorials. Start with real problems.
Pick a dataset you actually care about. For me, it was fund data. For you, it might be Zerodha trading data, or CRED transaction patterns, or PhonePe merchant data. Download it. Then ask real questions:
- What are the top performing funds by different time periods?
- How do returns correlate with fund size?
- Which funds had the most consistent performance?
Build queries to answer those. Google when you're stuck. Build increasingly complex queries. That's how you actually learn, not by doing practice problems on some tutorial website.
| SQL Concept | Why It Matters | Time to Learn |
|---|---|---|
| Window Functions | Allows comparison within groups (ranking, running totals, rolling averages) | 2–3 weeks |
| CTEs (Common Table Expressions) | Makes complex queries readable and maintainable | 1–2 weeks |
| Query Optimization | Difference between 5-second queries and 5-minute queries | 4–6 weeks |
| Index Behavior | Why some queries are slow and how to fix them | 3–4 weeks |
4. Communication and Storytelling With Data
This is where most technical analysts fail. And I almost failed too.
I can write excellent queries. I can build dashboards that technically answer every question someone might ask. But I spent my first 18 months at Morningstar not really being heard.
I'd present analysis. People would nod politely. Then nothing would change. No decisions would be made. No action taken. I'd think, "Why aren't they listening? The data is right there!"
Then one day, a senior analyst presented something simpler than my analysis. But different. She didn't start with the data. She started with what mattered.
The Structure That Actually Works
I rewrote my entire approach. Now every presentation follows this structure:
1. The Question (30 seconds) — What are we trying to figure out? Make it personal. Not "fund performance comparison" but "We're considering increasing our emerging market allocation. Will this move improve returns without increasing volatility beyond our 8% target?"
2. The Finding (60 seconds) — What did the data show? Just the headline. Nothing else. "Emerging market funds outperformed domestic funds by 2.3% over the last 5 years, but volatility was 23% higher. That exceeds our 8% target."
3. The Implications (90 seconds) — What does this mean? What should we do about it? "We'd need to either increase volatility tolerance, or hedge the emerging market exposure to bring volatility back to acceptable levels. Hedging would cost approximately 0.4% annually."
4. The Data (as needed) — Support all of the above. Only show the charts/tables that directly support your finding.
That's it. Doesn't matter if you're talking to a manager or a C-suite executive. Same structure.
Making Data Stick
One thing I've learned: people forget numbers. They don't forget comparisons.
Instead of "Morningstar's average fund has a 6.2% annualized return," I say "Morningstar funds returned 1.8x what you'd get from a savings account over the last decade."
Instead of "This fund underperformed by 340 basis points," I say "This fund's underperformance cost you ₹3.4 lakhs on a ₹10 lakh investment."
Numbers stick when they're relative and relevant to the listener's life.
5. Curiosity and the Ability to Learn Systems
This might sound soft, but it's the most important skill. And it's the hardest to assess in interviews, which is why companies care about it so much.
Every company has systems. Legacy databases. Weird processes. Quirks. The analysts who thrive are the ones who don't just complain about the systems—they learn them and work within them until they can improve them.
What This Actually Looks Like
When I started at Morningstar, we had data scattered across five different systems. Our internal database, two vendor APIs, one Excel file that someone updated manually, and a Google Sheet that definitely shouldn't have been a data source but was anyway.
My first instinct was to propose a massive data warehouse overhaul. Modernize everything. Seemed logical.
But I talked to the team first. Turns out, that Google Sheet was there because the Excel file had limitations that didn't get addressed for two years. If I'd just built a data warehouse without understanding the human and process context, I'd have built something that wouldn't solve the actual problems.
So instead, I spent my first month asking questions, understanding why systems evolved the way they did, and finding the pain points that actually mattered. Then I fixed those specific things, one by one. Took longer overall, but the solutions stuck because they solved real problems.
Building This Skill
Practically? It means:
- When something is confusing, don't assume it's bad. Ask why it exists that way.
- Learn the context before proposing solutions. "Why do you do it this way?" is more valuable than "You should do it this way."
- Acknowledge constraints. Budget, technical debt, team size, skill gaps. Solutions that ignore constraints don't happen.
- Prove value with small wins first. Fix one thing, show the impact, then propose bigger changes.
This is also the hardest skill to teach yourself, which is why it's so valuable. Junior analysts can learn SQL from YouTube. But learning how to navigate organizational complexity? That takes time and someone has to show you how.
My Perspective
My best friend Rohit works in consulting. A few weeks ago, he mentioned that his firm hired an analyst from IIT who knew advanced statistics, machine learning, the whole package. Six months later, that analyst was miserable. Not because the work was hard, but because nobody cared about the advanced models. The client just needed clean data and clear answers.
That conversation stuck with me. Because I realized I'd wasted a lot of time early in my career learning things that didn't matter. I spent three months learning advanced data visualization techniques when what I really needed was to get better at asking clarifying questions.
I used to think the most valuable analyst was the one with the longest list of technical skills. I was wrong. The most valuable analyst is the one who understands what problem they're actually solving and has the skills to solve it. Sometimes that's advanced SQL. Sometimes it's just clear communication.
What surprised me most? The skill that got me my best opportunities wasn't technical at all. It was curiosity. Genuinely wanting to understand how things worked, why people made decisions the way they did, what was really holding them back. That opened every door.
Final Thoughts
If you're 22–28 and thinking about data analysis as a career, or you're already in your first analyst role trying to figure out what to actually learn, here's my honest take:
Don't chase the shiny certifications. Don't try to learn every tool. Don't build fancy models nobody will use.
Instead, pick one real problem you care about. A company you follow. A dataset that interests you. Then actually solve that problem. Use SQL to get the data. Python to clean it. Visualization to show it. Communication to explain why it matters.
Do that once. Do it well. Then do it again.
That's how you become unhireable-able. Not because you have all the certifications. But because you can actually solve problems and explain why they matter.
The commute from Kalyan to Mumbai is brutal, but it's taught me something: the job you want isn't determined by where you studied or what degree you have. It's determined by whether you can actually deliver value. Everything else is noise.
You've got this. Now go build something.
Dattatray Dagale
Data Analyst • Blogger • Mumbai
I'm a data analyst from Kalyan, Maharashtra, working at Morningstar. I write about personal finance, career growth, and everyday life for Indian millennials — the stuff I wish someone had told me earlier.
Written by Dattatray Dagale • 29 September 2026
.png)
.png)
0 Comments