Use this guide to prepare for Data Analyst interviews, with a focus on learning and growth, insight, sql. Explain your reasoning and connect it to experience you can substantiate.
These preparation themes come from the questions in this role’s bank. They help you organise your examples; individual employers may assess different things.
Learning and growth
Insight
SQL
Stakeholders
Metrics
VLOOKUP basics
A useful preparation sequence
Choose your experience level and the round you expect.
Answer one question in your own words before opening its guide.
Compare your reasoning, evidence and trade-offs; adapt the answer to your experience.
Practise the follow-up, then revisit one answer you want to improve.
Representative questions and answer guidance
Open any question to read its answer. The complete guidance is included on this page.
HR · Fresher
1. Why do you want to work as a data analyst, and what kind of problems interest you?
Answer guide
I would explain what draws me to the work in a genuine way, for instance enjoying finding a pattern in messy information and turning it into something people can act on. I link that to a real experience, such as a project, course or personal dataset where I saw a useful result. I then say what kind of problems interest me, perhaps understanding customer behaviour or improving a process, while staying open to different areas. I add that I want to build strong fundamentals in querying, statistics and communication in my first years.
What this question explores
Whether your interest in analysis is genuine, backed by real experience, and realistic about the job.
Common mistakes
Saying data is the future and every company needs it, with no personal reason given.
Focusing only on salary or job security and showing no curiosity about the work.
Practise a follow-up
What have you done on your own to practise working with data?
What do you think is the hardest part of an analyst's job?
2. Tell me about an analysis that changed a business decision.
Answer guide
Pick a real example with a clear business result. Say what question you were asked and what you thought might be true. Explain the data and method in simple terms, and any checks you did. Then describe your recommendation and how you presented it to decision makers. Give the effect, like cost saved or higher conversion, ideally measured after the change. End with what you learned about communicating findings to people who are not analysts.
Key points
Explain hypothesis, method, recommendation and measured effect.
3. How would you calculate monthly active customers without double counting repeat transactions?
Answer guide
First, agree what active means, for example at least one completed purchase in the month. Then count each customer once using a distinct customer ID, and not the number of transactions. Set clear date boundaries, such as the first day of the month up to before the first day of the next, and watch for time zones. Filter out cancelled or test orders if your definition says so. Check the result against a small sample and a known total.
Key points
Define active, distinct key, date boundary and validation.
4. How do you prevent dashboards from multiplying with conflicting KPI definitions?
Answer guide
Conflicting KPIs cause confusion and lost trust. Start by creating one shared list of KPI definitions, with the formula, data source and filters. Give each KPI an owner who can approve changes. Use change control, so any change is reviewed, dated and shared. Track data lineage, meaning where each number comes from. Use certified datasets for dashboards. Retire old dashboards, and train teams to use the approved ones.
Key points
Create definitions, owners, change control and lineage.
5. A leader asks whether a campaign worked but the success metric was never agreed. What next?
Answer guide
Do not just produce a number. First, talk to the leader about what the campaign was meant to achieve, such as sales, sign-ups or awareness. Agree a baseline, for example last period or a similar group without the campaign. Explain the limits, since other things may have affected results and attribution may be imperfect. Agree what result would count as success. Then analyse honestly, share the findings with caveats, and set the metric up front next time.
Key points
Clarify objective, baseline, attribution limits and decision threshold.
6. What does VLOOKUP do, and why does its fourth argument matter so much?
Answer guide
VLOOKUP searches for a value in the first column of a range and returns a value from a column I specify to the right. The fourth argument controls match type. FALSE means exact match, which is what I want for IDs, names and codes. TRUE, or leaving it out, means approximate match, which assumes the first column is sorted ascending and can quietly return a wrong row if it is not. I always type FALSE explicitly unless I am deliberately looking up a band, such as a tax slab or grade boundary, in a sorted table.
What this question explores
Whether you know the basic lookup syntax and understand the risk of silent wrong answers from approximate matching.
Common mistakes
They leave the fourth argument blank and do not realise approximate match is the default.
They try to return a column to the left of the lookup column, which VLOOKUP cannot do.
Practise a follow-up
What would you use if the return column is to the left of the lookup column?
How would you stop a column index from breaking when someone inserts a column?
7. A dashboard total differs from finance's report. How do you trace the discrepancy?
Answer guide
Do not pick a side yet. First, agree what each number means and the level of detail, or grain, of both reports. Compare filters, such as date ranges, regions, and included or excluded products. Check timing, for example cut-off dates and late-posted entries. Look for join mistakes that duplicate or drop rows. Confirm both use the same version of the source data. Reconcile step by step, document the cause and agree one definition.
Key points
Reconcile grain, filters, timing, joins and source versions.
8. You inherit an unfamiliar financial model that feeds board decisions. How would you audit it before relying on it?
Answer guide
I would first copy it, then map the structure: inputs, calculations and outputs, with sheet flow. I would use Trace Precedents, Go To Special for formulas, constants and errors, and Inquire or similar tools if available, to find hard-coded numbers inside formulas, inconsistent formulas across a row, and broken links. I would check totals against source data, test with extreme inputs, reconcile balance sheet and cash flow, and rebuild key outputs independently. I would document findings by severity, agree fixes with the owner, and avoid changing the original until the review is signed off.
What this question explores
Whether you have a structured, risk-based approach to reviewing a model and can separate checking logic from fixing it.
Common mistakes
They start editing formulas immediately instead of mapping and documenting the model first.
They check only the final outputs and miss hard-coded plugs hidden inside formulas.
Practise a follow-up
Which checks would you automate permanently in the model?
How would you report findings to a non-technical sponsor?
9. How would you reduce the risk from business-critical spreadsheets across a finance and operations organisation without banning Excel?
Answer guide
I would start with an inventory and risk-rank spreadsheets by financial impact, regulatory exposure and complexity. For the highest tier, I would require ownership, documentation, independent review, access control, version history and standard checks. Lower tiers get lighter guidance. I would publish modelling standards and templates, train power users, and offer better alternatives where Excel is a poor fit, such as shared data platforms. I would measure progress through the number of reviewed models and incidents. Control must be proportionate, otherwise people will work around it, and leadership should treat it as operational risk.
What this question explores
Whether you can take a proportionate, risk-based approach to end-user computing that balances control with productivity.
Common mistakes
They propose a blanket ban, which drives spreadsheet use into unmanaged shadow files.
They apply the same heavy controls to every file regardless of risk.
Practise a follow-up
How would you measure whether the programme is working?
10. What is a vector in R, and what happens when you mix different types in one vector?
Answer guide
A vector is R's basic data structure, an ordered collection of values that all share one type, such as logical, integer, double or character. When I combine different types with c(), R silently coerces everything to the most flexible type, in the order logical, then integer, then double, then character. So combining the number 1, the value TRUE and the text 'a' gives a character vector, with TRUE becoming the text 'TRUE'. A single number is just a vector of length one. I check types with class() or typeof(), because a stray text entry can quietly turn a numeric column into character.
What this question explores
Whether you understand R's core data structure and know that type coercion happens silently and can cause surprises.
Common mistakes
They think a vector can hold mixed types like a list, without noticing the silent coercion.
They forget that single values are also vectors, so they misunderstand how vectorised operations work.
Practise a follow-up
How is a list different from an atomic vector?
How would you find out why a numeric column was read as character?
11. What is the difference between an inner join and a left join, and when would you use each?
Answer guide
An inner join returns only rows that have a match in both tables, while a left join keeps every row from the left table and fills the right-hand columns with nulls when there is no match. I use an inner join when I only care about records that exist on both sides, such as orders that have a matching customer. I use a left join when the left table is the population I am counting, for example all customers including those who never ordered. A common surprise is that filtering on a right-table column in the where clause can quietly turn a left join into an inner join.
What this question explores
Whether you understand how join type changes the rows returned and can pick the right one for a business question.
Common mistakes
Describing joins only as overlapping circles without saying what happens to unmatched rows.
Using a left join everywhere out of habit, so unexpected nulls and extra rows slip into results.
Practise a follow-up
What happens to the row count if the right table has several matches for one left row?
How would you find customers who have never placed an order?
12. How do you choose between a table, bar chart and trend line for an executive review?
Answer guide
Start with what the executives need to see. Use a table when they need exact numbers or many values. Use a bar chart to compare categories, like sales by region. Use a line chart to show change over time. Keep it simple, with a clear title that states the point, few colours and no clutter. Do not use fancy charts that are hard to read. Test it by asking whether someone could get the message in ten seconds.
Key points
Match visual to comparison, trend and precision needs.
Choose one answer containing an example or practical sequence. Explain what you would actually do, what you would check and when you would ask for help. Keep claims about your experience honest.
For technical or regulated work, check current documentation and applicable local requirements alongside this practice material.