Bank operations to data analyst: branch MIS as your first project

A branch operations day, re-read as analyst work

Think through a typical day and notice where you are already working with data.

9:30, pending list review. You open the list of accounts awaiting KYC completion or document verification and decide which to chase first. That is prioritisation from a queue, which is exactly what an operations analyst builds a report for.

11:00, account opening. Each application moves through stages: form received, documents verified, CBS entry, account activated, welcome kit dispatched. You know which stage usually holds things up. An analyst would call that a funnel with stage-wise turnaround time.

14:00, clearing returns. A cheque bounces and you record a return reason. Over a month, return reasons form a categorical dataset that could show whether a particular reason is rising.

16:30, day-end MIS. You fill numbers into a template for the regional office: new accounts, CASA figures, pending items. Sometimes your count and the system count disagree and you work out why. That reconciliation instinct is one of the most valuable things you bring.

Month-end, dormant and inoperative accounts. You pull lists of accounts with no customer-induced transaction for a long period. That is a date-difference filter on transaction history.

The analyst version of each task replaces manual counting with a query, keeps the logic written down, and adds a "so what" for the reader.

Build the account-opening turnaround project

Choose account opening because you understand every stage. Here is a minimal version on synthetic data in pandas:

import pandas as pd

# Synthetic account-opening log. All values invented for practice.
apps = pd.DataFrame({
    "app_id":     [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
    "channel":    ["Branch", "Branch", "Tab", "Tab", "Branch",
                   "Tab", "Branch", "Tab", "Branch", "Tab"],
    "received":   ["2026-09-01", "2026-09-01", "2026-09-02", "2026-09-02", "2026-09-03",
                   "2026-09-03", "2026-09-04", "2026-09-04", "2026-09-05", "2026-09-05"],
    "activated":  ["2026-09-03", "2026-09-08", "2026-09-03", "2026-09-03", None,
                   "2026-09-04", "2026-09-09", None, "2026-09-07", "2026-09-05"],
    "hold_reason":[None, "Address proof mismatch", None, None, "Signature mismatch",
                   None, "Address proof mismatch", "Photo unclear", None, None],
})
apps["received"] = pd.to_datetime(apps["received"])
apps["activated"] = pd.to_datetime(apps["activated"])
apps["tat_days"] = (apps["activated"] - apps["received"]).dt.days

summary = apps.groupby("channel").agg(
    applications=("app_id", "count"),
    activated=("activated", "count"),
    median_tat=("tat_days", "median"),
    over_3_days=("tat_days", lambda s: int((s > 3).sum())),
)
print(summary)
print()
print(apps["hold_reason"].value_counts())

On this invented data, tablet-sourced applications activate with a median of 1 day and none take more than 3, while branch forms take a median of 3.5 days with two cases above 3. "Address proof mismatch" is the most common hold reason. Two applications are still open, and they are excluded from the median, which you should say explicitly because it flatters the result.

Write the finding the way an operations head reads it

A table on its own is not the deliverable. End the project with three or four sentences:

• What you measured: calendar days from form received to account activation, by sourcing channel.
• What you found: branch forms take longer, and address-proof holds explain most of the delay.
• What you would check next: whether the delayed branch cases cluster on particular days or with particular staff, and whether open applications are older than the closed ones.
• What you would suggest: a checklist at the counter for address proof, then re-measure after a month.

That last point matters. Suggesting a change and a way to measure it shows you think like an analyst, not a report preparer.

What branch work does not prepare you for

Be honest about the gaps so you can close them deliberately:

Also expect to learn calendar logic properly: business days versus calendar days, holidays and cut-off times all change turnaround numbers.

Branch MIS and operations analyst roles

Operations-side roles are the natural first target: operations analyst, MIS analyst, business analyst (banking operations) or process excellence analyst. They value domain knowledge enough that your process understanding counts, while still testing SQL and Excel. Credit risk and fraud analytics roles are a reasonable second step once your statistics are stronger.

A sample answer to "Why move out of branch banking?"

"I have spent three years in branch operations, handling account opening, KYC follow-ups and the daily MIS. The part I was best at was working out why numbers did not match and which pending cases to chase first. I realised those are analysis problems, so I learned SQL and pandas and rebuilt our kind of account-opening turnaround report on synthetic data. It showed which hold reasons caused most of the delay and suggested a counter checklist with a way to measure its effect. I want to do that work full-time for an operations team, where I already understand the processes behind the data."

Where to start this week

Upload your current resume to the free skill check and see which analyst skills it shows. The Data Analyst programme teaches SQL, Power BI and verified use of AI tools on business data, and can be studied live online alongside a job, with optional in-person labs at our Noida centre. If you come from a different finance background, the CA Inter to financial data analyst guide takes a different route.