"Walk me through one of your projects" is asked in almost every data analyst interview. A clear, confident answer often matters more than the project's complexity. Many candidates lose marks by listing tools instead of telling a story.
A simple structure: six parts
1. The problem (one or two sentences)
Start with the business question, not the tools. "A fictional telecom company, SignalOne, saw prepaid churn rise after a tariff change. I wanted to find which customers were leaving and why."
2. The data
What data, roughly how big, and its main tables or columns. "About a year of customer, recharge and complaint data across four tables."
3. Your approach
The steps you took, briefly:
- Cleaned duplicates and missing recharge dates.
- Defined churn as no recharge for 30 days after plan expiry.
- Compared churn by plan, region, tenure and complaint history using SQL.
- Built a Power BI dashboard for the retention team.
4. Findings
Two or three specific insights. "Churn was highest among customers on the lowest plan in their first three months, and those with a network complaint were much more likely to leave."
5. Recommendation or impact
What should the business do? "Target first-quarter low-plan customers with a retention offer and prioritise fixing complaint hotspots."
If it is a practice project, say what impact you would measure, rather than inventing results.
6. Learnings
One honest point: "I learned to agree on a churn definition early because it changed the numbers significantly."
Aim for about two minutes in total.
Follow-up questions to prepare for
- Why that churn definition? Show you considered alternatives.
- How did you handle missing data? Be specific.
- What was the hardest part? Pick something real.
- What would you do with more time? For example, a predictive model or an A/B test of the offer.
- Write the query for one of your findings. Know your own SQL.
- How would you check your numbers are right? Reconciliation, spot checks, totals.
Common mistakes
- Starting with "I used Python, pandas, Matplotlib and Power BI" before explaining the problem.
- Not knowing details of your own project, which suggests you copied it.
- Claiming invented business impact.
- Talking for ten minutes without pausing for questions.
Prepare two projects
Have one SQL or dashboard project and one Python or more analytical project ready. Pick the one closest to the company's domain.
Practise aloud
Record yourself, check the length, and remove filler. Then practise with a mock interview. If you need a strong project to talk about, pick an industry case in our live projects, and see how to present it on paper in resume projects.
Practise what you just read
Reading helps; writing the code yourself is what sticks. Start here: