A data analyst answers business questions with data and makes sure the numbers everyone else quotes are correct. In practice that means about a fifth of the week keeping recurring reports alive, a large chunk of it cleaning and reconciling data that arrives broken, a steady stream of ad hoc questions from other teams, and, if the week goes well, one genuine investigation that changes a decision.
Job descriptions describe the last part and skip the first three. Below is a week written the other way round, at a Bengaluru direct-to-consumer brand selling home and kitchen products, with a four-person data team and an analyst about eighteen months into the role. The details are a composite, but nothing in it is unusual.
Monday
The week starts with the weekly business review pack, which has to be in the founders’ inbox by 11am. It is a Power BI report the analyst already built, so Monday morning is refreshing it, checking that the numbers reconcile against the finance sheet, and writing four lines of commentary about what moved.
The refresh fails. A marketing agency changed a campaign naming convention over the weekend, so ₹6 lakh of spend is now landing in an “unassigned” bucket and the customer acquisition cost by channel is wrong. Forty minutes goes into finding it, another thirty into patching the mapping table, and a Slack message goes to the agency asking them to stop doing that.
The pack ships at 11:20. Afternoon is the growth stand-up, where the head of growth asks whether the Bengaluru and Hyderabad discounting experiment is working. The honest answer is that it has been running for nine days and the sample is not large enough yet, which is not a popular answer but is the right one.
Tuesday
Ad hoc day. Four requests arrive before lunch: repeat purchase rate by city for a board slide, a list of customers who bought a specific product line for a customer-support campaign, why last month’s returns number differs between two dashboards, and a “quick question” from the category team that turns out to require joining three tables and a fortnight of history.
The returns discrepancy is the interesting one. Two dashboards disagree because one counts returns by the date the return was requested and the other by the date it was processed. Neither is wrong. Nobody had written down which definition the company uses. The analyst picks one, documents it, updates both dashboards, and posts the definition where people will find it. That single hour of work will prevent the same question being asked six more times.
Wednesday
Investigation day, protected on the calendar because it will not happen otherwise. Repeat purchase rate has been sliding for three months and nobody knows why.
The work is unglamorous. Pull the order history, define what “repeat” means for a brand where some products are bought once every three years, segment by acquisition cohort and channel, and check whether the drop is real or whether the denominator changed when the company started counting marketplace orders. It turns out to be about a third an artefact of the denominator and two thirds real, and the real part is concentrated in customers acquired through one discount-led channel.
That finding takes most of the day and produces one chart and four sentences. Those four sentences will change how the next quarter’s marketing budget is split, which is what the job is for.
Thursday
Half the day goes to the finding from Wednesday. Not more analysis, but making it defensible: re-running with a different cohort definition to check the pattern holds, writing it up in a page anyone can read, and pre-empting the objection that the discount channel simply had a different product mix.
Then a meeting with the marketing lead to walk through it, where the first fifteen minutes are spent on why the number in this deck differs from the number in her weekly email. Anticipating that question is a skill nobody puts on a job description.
Late afternoon: a data engineer flags that a source table will change schema next week, so the analyst goes through every dashboard and query that touches it and works out what will break.
Friday
Maintenance and margins. Two scheduled reports need small changes, a new joiner in the category team needs access and a walkthrough of what the tables mean, and there is a backlog ticket to make the returns dashboard load in under ten seconds instead of forty, which is mostly a matter of aggregating in the warehouse rather than in the BI tool.
Friday afternoon is when the analyst finally writes down the SQL definitions of the six metrics the business argues about most, because Tuesday made the case for it. That documentation is invisible work that quietly reduces next month’s ad hoc load.
Where the week actually went
The week above, broken down into hours.
| Activity | Hours | Tools |
|---|---|---|
| Ad hoc questions from other teams | 9 | SQL, Excel |
| Recurring reporting and dashboard upkeep | 8 | Power BI, SQL |
| Cleaning, reconciling and fixing broken data | 7 | SQL, Excel, Python |
| The one real investigation | 6 | SQL, Python, Excel |
| Meetings, stand-ups, walkthroughs | 5 | Slides, plain English |
| Writing things up and documenting definitions | 5 | Docs, Slack |
Six hours of the forty were the analysis that the job title implies. That ratio is normal, and it is the single biggest gap between what people expect the job to be and what it is. Analysts who are unhappy in the role are usually unhappy about the other thirty-four hours.
The good news is that the ratio is not fixed. Analysts who automate their recurring reports and document their definitions buy themselves more investigation time, which is exactly the work that gets people promoted.
The three kinds of request an analyst handles
The recurring report. Same numbers, same cadence, and it must be right every time. Low intellectual demand, high reliability demand. The skill is making it robust enough to survive somebody else changing a campaign name on a Saturday.
The ad hoc question. Someone needs a number by 4pm. The trap is answering the question asked instead of the question meant. “Send me repeat purchase rate” almost always means “I am worried about something”, and two minutes of asking what the number is for saves an afternoon of rework.
The investigation. A trend nobody understands, or a decision that needs evidence. This is where the job is actually worth what it pays, and it is also the first thing that gets squeezed out of a busy week, which is why good analysts defend the time on their calendar.
What the job is not
It is not machine learning. Analysts occasionally build a simple model, but predictive modelling belongs to data scientists and most analysts go years without training one.
It is not building data pipelines. That is data engineering. An analyst is a heavy consumer of pipelines and is usually the first person to notice when one breaks, which is not the same as owning it.
It is not making dashboards all day, though a badly run analytics team can turn into exactly that. If every request that arrives becomes a new dashboard, the team has stopped answering questions and started manufacturing surfaces.
And it is not a quiet, solitary job. A large share of the week is spent talking to people who do not work in data, which is why communication shows up in nearly every Indian analyst job posting alongside SQL and Excel.
How the work changes with the company
At a large IT services firm, an analyst usually sits on a client account, and the work is more standardised: defined reporting deliverables, stricter processes, less say over what gets asked. It is a good place to learn discipline and client handling, and the pay bands sit lower.
At a startup, an analyst is often the entire data function. More autonomy, more chaos, more of the week spent on things that are technically somebody else’s job, and far more influence over decisions.
At a global capability centre, the analyst supports a business unit that sits in another country, so the work is closer to a product company’s but the stakeholders are on a call at 8pm.
For what these differences do to pay, AmbitionBox’s median for the role in India was ₹6,84,517 across 1,44,866 reported salaries when we checked on 16 August 2026, and its industry breakdown ranged from ₹5,96,453 in analytics and KPO to ₹10,00,166 in financial services. Same title, very different weeks.
What changes as you get senior
The work moves upstream. A junior analyst is handed questions. A senior analyst is handed a problem and decides which questions are worth asking, which is a different skill and the one that actually gets rewarded.
Two SkilloVilla alumni stories show where that path leads: one into a senior data analyst role at Shadowfax and another into an analyst role at Google. In both, the technical stack was the entry ticket and the progression came from owning a business area rather than from learning another tool.
Frequently asked questions
What does a data analyst do on a daily basis?
Most days combine keeping recurring reports and dashboards running, answering ad hoc questions from business teams, and cleaning or reconciling data that has arrived incomplete or inconsistent. Genuine open-ended analysis takes up a smaller share than people expect, often a day or less in a five-day week. Meetings and explaining results to non-technical colleagues take up more.
Is data analyst a stressful job?
It has deadline pressure rather than crisis pressure. The stressful parts are month-end and board-review cycles, and the moments when two reports disagree an hour before a meeting. Analysts who document their metric definitions and automate their recurring work find the role considerably calmer than those who rebuild everything by hand each month.
Do data analysts need to talk to people, or is it solitary work?
It is a communication-heavy role. A large part of the week goes on clarifying what someone actually wants, walking stakeholders through findings, and explaining why two numbers differ. Nearly every Indian analyst job posting lists stakeholder communication next to SQL and Excel, and it is usually what separates analysts who get promoted from those who stay in reporting.
What is the difference between a data analyst and a data scientist?
An analyst explains what happened and why, using SQL, Excel and a BI tool, and hands the business a recommendation. A data scientist builds models that predict what will happen, using statistics and machine learning. The roles overlap at the edges, and analysts who want to move across usually add statistics and Python modelling on top of the analyst stack.
What tools does a data analyst use every day?
SQL and Excel are used almost every day, a BI tool such as Power BI or Tableau most days, and Python a few times a week for cleaning and automation. Beyond the tools, the daily work involves whatever internal documentation and messaging systems the company runs on, since a good share of the job is asking clarifying questions and writing findings down.
Can a data analyst work from home?
Frequently, yes. The work is laptop-based and much of Indian analytics is delivered in hybrid arrangements, though many companies expect two or three days in the office, particularly for junior analysts who learn faster near their team. Roles handling sensitive financial or customer data sometimes require office presence for access reasons.
If that week sounds like your kind of week
Notice what the week actually demanded: fluent SQL, fast Excel, a BI tool you can build in, enough Python to clean a mess, and the ability to explain a finding to somebody who does not care how you got it.
SkilloVilla’s Data Analytics with Python track teaches that stack over 4 to 5 months of live classes at ₹71,999, currently ₹58,999, with 1:1 mentorship from working analysts, real projects and placement support. If you want to start with the single most-used skill first, SQL: Beginner to Advanced is ₹33,110 on its own.
Fees and ratings last checked August 2026; confirm current numbers with the provider before enrolling.
