American Express Data Analyst Interview Experience | Round 1

Total rounds: 3 | Round 1: Technical + Analytical Skills

I recently appeared for the first round of the Data Analyst interview at American Express, and honestly, it was a mix of things I had prepared for and a couple of things I wasn't expecting.

The interview started with a brief introduction. I talked about my background, my experience as a Business Analyst at Axis Bank, and some of the projects I had worked on, particularly around propensity modelling and business analytics.

The interviewer then moved into the technical questions. There wasn't much small talk after that!

1. SQL: WHERE vs HAVING

The first question was a pretty straightforward SQL concept: What is the difference between WHERE and HAVING?

I was comfortable with this one. I explained that WHERE filters individual rows before aggregation, whereas HAVING filters groups after the GROUP BY operation.

The interviewer seemed interested in whether I understood the practical difference, rather than just the definitions, so I explained it with an example involving filtering customers and then filtering groups based on their total transaction value.

This was probably the point where I felt most comfortable because SQL is something I use regularly.

2. The Three Ants Puzzle

Then came something I genuinely wasn't expecting: a puzzle involving three ants.

It was a nice change from SQL, but it also made me pause because these questions aren't necessarily about technical knowledge. They're more about how you approach a problem when you can't immediately see the answer.

I took a moment to think through the possibilities instead of jumping to a conclusion. The interesting part was that the interviewer wasn't just interested in the final answer; explaining the reasoning was important too.

This was one of those moments where you realise that preparing only technical questions isn't enough for an interview like this.

3. Outlier Detection and Anomaly Detection

This was probably the most interesting part of the interview.

The interviewer asked how I would identify outliers in a dataset. I discussed using percentiles to understand the distribution and identify unusually high or low values.

We also discussed visualisation. I mentioned using a box plot to identify potential outliers and a scatter plot to understand how observations are distributed and whether any points look unusual.

The discussion then moved towards anomaly detection, specifically DBSCAN.

I explained that DBSCAN is a density-based clustering algorithm that can identify points that don't belong to sufficiently dense clusters as noise. This makes it useful for detecting certain types of anomalies without having to specify the number of clusters beforehand.

What made this question interesting was that it wasn't simply about naming different methods. You had to understand when each approach would make sense.

For example, percentile-based methods can be useful for identifying extreme values in a distribution, while DBSCAN can help detect unusual observations based on their local density and the structure of the data.

I felt reasonably confident here, although it required more explanation than the SQL question.

4. What Are the Two Most Important Skills for a Data Analyst?

This question caught my attention because it was less technical than the previous ones.

I approached it from two angles: analytical thinking and communication.

Analytical thinking helps you break down a business problem, investigate the data, and identify what is actually driving the outcome. Communication is equally important because even a technically correct analysis has limited value if you cannot explain the findings and translate them into actionable business recommendations.

I also felt that this question was a reminder that the role isn't just about writing SQL queries or building models. You need to connect the analysis to the business problem.

5. Propensity Modelling — Questions About My Work Experience

The final major discussion was around propensity modelling, which was particularly relevant to my experience as a Business Analyst at Axis Bank.

The interviewer asked about the propensity model I had worked on, and I had to explain the business objective, how the model was used, and how it supported decision-making.

This was a different kind of challenge. With technical questions, you can often prepare definitions and practise examples. With work-experience questions, you need to be clear about what you actually did, why you did it, and what the outcome was.

I focused on explaining the business problem behind the model rather than getting lost in technical terminology. The key was to connect the analytical work to its business application.

This part of the interview reinforced how important it is to understand every line of your resume. If you mention a model or project, you should be prepared to explain the problem statement, your contribution, the methodology, and the impact.

Overall Experience

The first round covered more ground than I initially expected. It started with a basic SQL question, moved into a logical puzzle, tested my understanding of outlier detection and clustering, and then shifted towards business-oriented questions and my actual work experience.

My strongest moments were the SQL discussion and connecting analytical concepts to practical use cases. The puzzle was more unexpected, while the anomaly detection discussion required me to explain the reasoning behind different approaches rather than simply list algorithms.

One thing I took away from the experience is that a Data Analyst interview isn't just a test of SQL or Python. It tests how you think, how clearly you explain your approach, and whether you can connect data to business decisions.

This was only the first of three rounds, so I knew there was still more to prepare for. But overall, it was a useful experience because it gave me a clearer idea of the balance between technical fundamentals, problem-solving, and business understanding that the role demands.