I got a message on LinkedIn last week from someone I mentored two years ago. "Dattatray, I just rejected an offer from a unicorn because the salary felt low. They asked if I'd negotiate."
When I dug into what happened, it turned out the company made a *wrong offer* — their data stack was a mess, and they needed someone who could fix it. Not someone who could just run a SQL query or build a dashboard.
That's the thing about data skills in 2025. The game has changed completely.
It's not about memorizing Python libraries anymore. It's about understanding *why* a business is bleeding money, and having the toolkit to prove it. And here's what surprised me while researching this: most data analysts in India are still learning 2019 skills.
Let me break down what actually gets you hired in 2025 — and what you can ignore.
The Skills Everyone Thinks Matter (But Are Just Baseline Now)
Okay, so I need to get this out of the way first. SQL, Python, Tableau, Excel — these are table stakes. Like knowing how to drive is for getting a job as a delivery guy.
Nobody's going to hire you *because* you know SQL. But they'll reject you if you don't.
Here's what's shifted:
1. SQL isn't enough anymore — you need to understand data architecture
You know what I see in 90% of job descriptions? "Must know SQL." But when I talk to hiring managers at fintech companies (Zerodha, CRED, PhonePe's internal teams), they actually want something else.
They want you to look at their data warehouse and understand *why* it's structured that way. Whether they're using Snowflake, BigQuery, or a Postgres database — they want you to think about partitioning, indexing, and query optimization. Not just write correct queries, but *efficient* ones.
I spent three weeks last year trying to optimize a query that was taking 45 seconds. Sounds stupid? It was killing their real-time dashboard. Once I understood the table joins and rewrote it, it ran in 2 seconds. That's the difference between a script-writer and an actual analyst.
2. Excel is still crucial, but for different reasons
And honestly? Most analysts are terrible at Excel.
They can build a pivot table, sure. But can they write a formula that dynamically adjusts based on 500 rows of changing data? Can they use XLOOKUP instead of VLOOKUP? Can they structure data in a way that makes sense for a non-technical founder who's looking at your spreadsheet at 11 PM?
I've seen analysts lose contracts because their Excel models were sloppy. One woman I know lost a ₹15 lakh project because her dashboard broke when someone added a new row.
The Skills That Actually Separate You From the Crowd
1. Business Acumen — Understanding Why, Not Just How
This is the one that surprised me when I started hiring for my own projects.
The best analyst I've worked with wasn't the person with the fanciest Python skills. It was someone who asked stupid questions.
"Why are we tracking *this* metric instead of *that* one?" "If we increase conversion by 5%, what happens to our unit economics?" "Does this data point actually matter for the decision we're trying to make?"
And here's the real kicker: most founders and CEOs don't actually know the answer. They think they do. So when you come in with a clear head and start questioning assumptions, you become invaluable.
How do you build this? Read. Spend 30 minutes a week on fintech news sites, startup breakdowns, product reviews. Follow people like Vedavati Khare or Girish Mathrubootham. When you analyze data, ask yourself: "If I were running this company, what would I do with this insight?" That changes everything.
2. Data Storytelling and Communication
Let me tell you something nobody wants to hear: your analysis doesn't matter if you can't explain it.
I once spent three days building a model that showed exactly where a company was losing money. Brilliant work. Airtight numbers. And when I presented it? The CEO spent the whole meeting looking at my graphs instead of listening to what I was saying.
The next week, someone else did a worse analysis, but presented it like a story. "Here's what's happening. Here's why it matters. Here's what we do about it." They got the promotion.
In 2025, if you're a data analyst and you can't communicate, you're just an engineer doing manual work. Learn to:
- Write clear insights (not "CTR increased 2.3%" but "Every ₹1 we spend on retargeting returns ₹4.20 in revenue")
- Design dashboards that tell a story, not just display numbers
- Know when to use a chart vs. a table vs. just talking
- Anticipate questions before they're asked
3. Product Analytics and User Behavior
This one's getting hot. And I mean *hot*.
Because startups have finally figured out that tracking page views is useless. What they need is someone who understands cohort analysis, retention funnels, and what makes users stick around or leave.
Can you look at a product and think like a user? Can you spot why a feature launch flopped? Can you design an experiment that tests whether adding a new button will actually help or just clutter the interface?
Tools like Mixpanel, Amplitude, and even the basic analytics in apps like PhonePe or Razorpay — these are gold right now. And most analysts don't know how to use them beyond the obvious.
What gets you hired: understanding the *why* behind user behavior. Why did DAU drop 12% last Tuesday? Is it a bug? A bad update? Seasonal? Or did a competitor launch something? That analytical thinking is what startups are desperate for.
4. Statistical Thinking and Experimentation
And here's where a lot of analysts get uncomfortable.
I'm talking about A/B testing, statistical significance, and the guts to say "No, this result is probably just noise."
You'd be shocked how many "successful" campaigns are just luck. Someone runs an experiment, sees a 3% uplift, celebrates, and then it doesn't repeat. Why? They didn't account for sample size, didn't run it long enough, didn't control for confounding variables.
In 2025, companies are finally getting serious about this. They want analysts who can design experiments, calculate required sample sizes, and know the difference between correlation and causation.
This doesn't mean you need a PhD in statistics. But understanding p-values, confidence intervals, and when you have enough data? That's table stakes now.
The Tools and Tech Stack You Actually Need
Here's where I'm going to be real with you: the tools don't matter as much as understanding the *principles*.
But obviously, you need to know some tools. So here's my breakdown:
| Tool Category | What You Need | Why It Matters |
|---|---|---|
| Database | SQL + one modern warehouse (Snowflake or BigQuery) | Most startups are moving here. Learn the paradigm shift. |
| Visualization | Tableau, Looker, OR Metabase (not all three) | Know one deeply. It's the last mile of your work. |
| Programming | Python (Pandas, Matplotlib) OR R | For deeper analysis. Python is more popular in India. |
| Product Analytics | Mixpanel, Amplitude, or Segment | This is growing faster than traditional analytics. Learn it. |
| Experimentation | Understanding of A/B testing + one tool (Optimizely) | The future is experiments, not reports. |
But here's what I've learned: if you understand the *principles*, you can learn any tool in two weeks. I know analysts who switched from Tableau to Looker and owned it in 10 days because they understood what they were trying to do.
Don't waste six months learning Tool X when you could spend that time understanding statistical concepts, business models, and how to think.
How to Position Yourself Right Now
Okay, so you're reading this and thinking, "Dattatray, this is nice advice but what do I *do* tomorrow?"
Here's my playbook:
Month 1: Shore up your fundamentals
Go back to SQL. Write queries that would make an engineer proud. Learn about indexing, execution plans, and why your query is slow. If you already know SQL, move to understanding one data warehouse tool (Snowflake is the sexiest right now).
Month 2-3: Build a portfolio project
Pick a real problem. Not a Kaggle competition. A real problem. Here's an idea: analyze the pricing strategy of any fintech app you use (Zerodha, CRED, PhonePe, whatever). How do they make money? What's the unit economics? Build a dashboard that tells that story.
When you interview, this becomes gold. Not because it's perfect, but because it shows you think like a business person, not just a technician.
Month 4: Learn to communicate
Start a blog (or a LinkedIn where you post weekly). Explain one analytical insight every week in language a 16-year-old could understand. This does two things: it forces you to actually understand what you're talking about, and it builds your credibility.
Month 5-6: Get into experimentation
Take a course on A/B testing. Understand hypothesis testing. This is where the future is, and most analysts are still building dashboards.
Final Thoughts
Look, I've been analyzing data for long enough to see the patterns. And the pattern is clear: the best analysts are the ones who stop thinking like analysts and start thinking like business people who use data as a tool.
You don't need to be the smartest person in the room. You don't need to know every tool. You don't even need a fancy degree (though it helps).
What you need is curiosity, the ability to explain things clearly, and the willingness to ask "But why?" until you find the truth.
The jobs in 2025 aren't for people who are good at tools. They're for people who are good at solving problems.
Start building that skill today. In six months, you'll look back and wonder why you waited.
And hey, if you're stuck or want to chat about this, find me on Twitter or LinkedIn. I answer most messages.
Written by Dattatray Dagale • 07 May 2026
.png)
.png)
0 Comments