How to Hire a Data Analyst: SQL, Excel and Case-Study Interviews

Before writing a job description, list five questions you want answered regularly. For a retail chain it might be "Which stores are losing margin?" For a lending startup, "Which customer segments default more?" For a logistics company, "Where are deliveries getting delayed?"

These questions tell you what data sources the analyst will use, how technical the role is and who they will report to. They also make your interview much sharper, because you can test candidates on problems that look like your own.

These titles overlap, and mismatches are a common reason hires disappoint. Be clear about which one you need now.

• Data analyst: answers business questions with SQL, Excel or Sheets, and dashboards; explains findings to managers.
• BI or reporting analyst: builds and maintains dashboards in Power BI, Tableau or Looker Studio; focuses on reliable recurring reports.
• Data engineer: builds pipelines and data warehouses so others can analyse data.
• Data scientist: builds predictive models and experiments; usually needs clean data infrastructure already in place.
• If your data lives in Tally exports, Excel files and an app database, a strong analyst is usually the right first hire, not a data scientist.

Describe your data sources, tools and the decisions the analyst will support. Analysts are attracted by interesting problems and access to data, so mention both.

You can post a data analyst job on BetterJobs from a one-line description; the AI-written job description gives you a solid draft to customise, and the post appears on BetterJobs and Google Jobs.

• Sample line: "Analyse order, delivery and returns data from our PostgreSQL database to help operations cut late deliveries."
• Sample line: "Build and maintain weekly Power BI dashboards for the sales and finance teams."
• Sample line: "Strong SQL (joins, group by, window functions) and Excel (pivots, lookups) required. Python is a plus."

Give a short test with a small, realistic dataset. Online or take-home both work; 60–90 minutes is enough. Use a fictional dataset that resembles yours, such as orders, customers and stores.

Mark not only correct answers but also how they handle messy data. Include a few duplicates, missing values and inconsistent city names like "Bengaluru" and "Bangalore". Good analysts notice and mention them.

1. 1Write a query for monthly revenue by city for the last six months.
2. 2Find the top ten customers by revenue and their number of orders.
3. 3Calculate the repeat purchase rate: customers with more than one order divided by all customers.
4. 4Use a window function to find each store's month-over-month revenue change.
5. 5In Excel, build a pivot table from the same data and a simple chart of revenue trend.
6. 6Write three bullet points summarising what a manager should know from the data.

The case study tests thinking, not syntax. Present a business problem verbally and ask the candidate to talk through how they would approach it. There is no single right answer; you are listening for structure, sensible assumptions and the right questions.

Push gently with follow-ups. If they say "I would check the data", ask which tables, which metrics and what they expect to see.

• "Online orders from Jaipur dropped sharply last week. How would you find out why?"
• "Marketing says a new campaign doubled sign-ups, but revenue is flat. What would you analyse?"
• "We want to open a new warehouse. What data would you use to recommend a city?"
• "A manager wants a dashboard with 40 metrics. How do you respond?"

An analysis that nobody understands changes nothing. Ask candidates to present one past project in five minutes to a non-technical interviewer, such as the sales head or the founder.

Strong analysts lead with the finding and the recommended action, then show supporting numbers. Weak ones walk through every step of their process before getting to the point. Also check whether they are honest about uncertainty. "The data suggests this, but we only have three months" is a sign of maturity, not weakness.

Data analyst pay varies by city, industry and depth of skills. As a rough guide, entry-level analysts are often offered something in the region of ₹25,000–₹45,000 a month in metros, with experienced analysts and those with strong domain knowledge earning more. Benchmark against similar roles in your city.

Good sources include graduates in statistics, economics, engineering and commerce who have built projects, people moving from MIS or reporting roles, and analysts in adjacent industries. A resume database search for "SQL" plus your domain, such as "retail" or "lending", can surface candidates who will not see your post otherwise.

Many first analysts spend months fighting for data access. Before they join, arrange read access to the main databases, exports from Tally or your ERP, and a contact in each team who owns the data.

Agree on two priority questions for the first 60 days, and a weekly 30-minute review with the business owner of those questions. Early, visible answers build trust in data across the company and make the analyst's job far easier.

Also agree on definitions early. Teams often disagree about what counts as an "active customer" or a "delivered order". Ask the analyst to write a short glossary of key metrics in the first month and get it signed off by the relevant managers, so every report uses the same numbers and meetings stop turning into arguments about whose figure is right.

A long list of tools on a resume is not evidence of skill. Many candidates have completed online courses covering Python, Tableau, statistics and machine learning but have rarely worked with real, messy business data. The SQL test usually reveals this quickly.

Be wary of candidates who present only polished dashboards and cannot explain what decision the dashboard led to. Equally, be careful of candidates who jump to conclusions without checking data quality. Both habits can lead managers to act on wrong numbers.

On the employer side, the biggest mistake is hiring an analyst and then using them only to produce the same reports every week. Repetitive reporting should be automated over time so the analyst can work on questions that actually change decisions.

Keep the process itself short. Three steps are usually enough: a screening call, the SQL and Excel test, and a combined case-study and presentation round with the hiring manager. Move between steps within a few days, because strong analysts are often in several processes at once.

• Cannot write a basic join or group-by query without help.
• Treats a correlation in the data as proof of cause.
• Ignores duplicates, missing values or inconsistent names in the test data.
• Uses heavy jargon with non-technical interviewers.
• Has no questions about your business, customers or data sources.