Career Guide · Commerce to Analytics
Data Analytics Career for B.Com Students: The Complete Roadmap from Excel to Python and AI
Learn the tools in the order a data job will ask you to use them: spreadsheets first, then databases, then a dashboard tool, then code, with AI assistants layered on top once you can check their work. For a B.Com student this order matters more than for most people, because each step builds on something your commerce papers already taught you, and skipping ahead to "Python and AI" without the earlier steps is the most common reason commerce students give up.
This roadmap is written for students who are still in their B.Com (Pass or Honours) and want to use the three years well. If you have already graduated and are job hunting, the companion guide on becoming a data analyst after B.Com or M.Com focuses on job titles, interviews and projects.
Where your commerce papers already help
| B.Com paper | What it trains | Where it appears in analytics |
|---|---|---|
| Financial accounting | Ledgers, reconciliation, double entry | Matching two data sources and explaining every difference |
| Business statistics | Mean, median, dispersion, correlation, index numbers | Descriptive analysis, the core of most analyst reports |
| Cost and management accounting | Variance analysis, budgets, standard costing | Actual-versus-target reporting, one of the most common reports any finance team asks for |
| Income tax and GST | Rules applied to many records, compliance checks | Data validation: finding rows that break a rule |
| Computer applications (where offered) | Basic Excel, sometimes Tally | A starting point, though usually not deep enough for analyst work |
Step 1: Excel properly (first year)
Go beyond data entry: tables, SUMIFS and COUNTIFS, XLOOKUP, pivot tables, conditional formatting and charts. A good test is whether you can answer a question like "how much does Sharma Traders still owe us?" from an invoice sheet without sorting by hand:
=SUMIFS(D:D, B:B, "Sharma Traders", E:E, 0)
Here column B holds the customer, D the amount and E the paid flag (1 = paid, 0 = unpaid).
This Excel tutorial for beginners covers the basics if you need a refresher.
Step 2: SQL (second year)
Companies keep invoices, orders and customers in databases, not spreadsheets. SQL lets you ask the same question of millions of rows. The query below gives each customer's open invoices, total outstanding and the age of the oldest unpaid bill as on 31 October 2026:
CREATE TABLE invoices (inv_no TEXT, customer TEXT, inv_date TEXT, amount REAL, paid INTEGER);
INSERT INTO invoices VALUES
('INV-101','Sharma Traders','2026-07-02',48000,0),
('INV-102','Gupta Stores','2026-08-15',12500,1),
('INV-103','Sharma Traders','2026-09-20',30500,0),
('INV-104','Mehta & Sons','2026-10-05',22000,0),
('INV-105','Gupta Stores','2026-10-28',9800,0);
SELECT customer,
COUNT(*) AS open_invoices,
SUM(amount) AS outstanding,
MAX(CAST(julianday('2026-10-31') - julianday(inv_date) AS INTEGER)) AS oldest_days
FROM invoices
WHERE paid = 0
GROUP BY customer
ORDER BY outstanding DESC;
Result (SQLite 3.45; julianday is SQLite's date function, other databases use DATEDIFF or date subtraction):
customer | open_invoices | outstanding | oldest_days
Sharma Traders | 2 | 78500.0 | 121
Mehta & Sons | 1 | 22000.0 | 26
Gupta Stores | 1 | 9800.0 | 3If you have studied debtors and the provision for doubtful debts, you already know why a 121-day-old invoice matters more than a 3-day-old one. That understanding is what turns a query into an insight. This beginner's SQL guide for analysts is a good place to start.
Step 3: a dashboard tool (second or third year)
Learn Power BI or Tableau well enough to build a one-page report that a manager can filter by month and customer. Power BI Desktop is free to install and is widely used in Indian companies. Build a sales or receivables dashboard first, because you understand those numbers. See how to start learning Power BI.
Step 4: Python with pandas (third year)
Python is worth learning once you have hit the limits of Excel: files too large to open, the same cleaning steps every week, or analysis that needs many steps in a fixed order. The script below turns the open invoices into a standard ageing report, the same report an accounts team prepares at month end:
import pandas as pd
inv = pd.DataFrame({
"inv_no": ["INV-101", "INV-103", "INV-104", "INV-105"],
"customer": ["Sharma Traders", "Sharma Traders", "Mehta & Sons", "Gupta Stores"],
"inv_date": pd.to_datetime(["2026-07-02", "2026-09-20", "2026-10-05", "2026-10-28"]),
"amount": [48000, 30500, 22000, 9800],
})
as_on = pd.Timestamp("2026-10-31")
inv["days"] = (as_on - inv["inv_date"]).dt.days
inv["bucket"] = pd.cut(inv["days"], bins=[0, 30, 60, 90, 365],
labels=["0-30", "31-60", "61-90", "90+"])
ageing = inv.pivot_table(index="customer", columns="bucket", values="amount",
aggfunc="sum", fill_value=0, observed=False)
print(ageing)
Output (pandas 3.1):
bucket 0-30 31-60 61-90 90+
customer
Gupta Stores 9800 0 0 0
Mehta & Sons 22000 0 0 0
Sharma Traders 0 30500 0 48000observed=False keeps the empty 61–90 column, so the report always has the same four buckets even in a month when one is empty. Small decisions like this are what make a report reliable.
Step 5: AI tools, used with checking
AI assistants such as ChatGPT, Gemini or Copilot are genuinely useful to an analyst: explaining an unfamiliar formula, drafting a SQL query, suggesting chart types or summarising findings. They are also confidently wrong at times. Use them only after steps 1 to 4, so you can check what they produce. A practical habit: when an assistant writes a query for you, run it on a small table where you already know the answer, as with the five invoices above, before trusting it on real data. Never paste confidential company or client data into a public AI tool.
Fitting this around a B.Com timetable
- Semester exams first. Your degree result still matters for many employers' eligibility filters. Plan lighter study weeks around exams.
- If you are also preparing for CA, CMA or CS, do Excel and SQL alongside, and leave Python for later. Both paths value the same accuracy.
- Use holidays for projects. A month-long break is enough for one solid dashboard project.
- Look for a short internship in accounts, MIS or audit support in your second or third year. Even basic reporting work gives you real examples for interviews.
Projects that suit commerce students
- A receivables ageing dashboard like the example above, built on a larger made-up or public dataset.
- Budget-versus-actual variance analysis for a sample company, with commentary on the three largest variances.
- An analysis of a public dataset, such as GST collections or retail sales figures published by a government body, explaining a trend in plain language.
By the end of your third year, aim to have Excel, SQL and one dashboard tool at a confident level, plus a start in Python and two projects you can explain line by line. That is a realistic, honest base for entry analyst roles, and the rest can be learnt on the job.