Design an A/B test for a recommendation algorithm
Last updated: June 12, 2026
Quick Overview
Design an experiment to test the impact of a redesigned homepage. Include sample size calculation, metrics, and analysis plan.
Mastercard
June 12, 2026113
15
3,553 solved
Design an experiment to test the impact of a redesigned homepage. Include sample size calculation, metrics, and analysis plan.
Statistics questions at Mastercard test your ability to reason quantitatively and design rigorous experiments. This Technical Screen question evaluates your understanding of statistical inference and its application to business decisions.
What the Interviewer Expects
- State the correct formula or theorem with clear definitions
- Apply the concept to the given scenario step by step
- Interpret the result in plain language
- Identify assumptions and when they might be violated
Key Topics to Cover
How to Approach This
- Define your hypotheses (H0 and H1) clearly before performing any test.
- Calculate required sample size BEFORE running an experiment, using power analysis.
- Remember the Central Limit Theorem: sample means become approximately normal with large n.
- Watch for Simpson's paradox. Always segment data by key dimensions.
- Distinguish between statistical significance and practical significance.
Possible Follow-up Questions
- How would you handle multiple comparisons?
- What if the sample size is very small?
- What alternative statistical method could you use here?
Sharpen Your Skills on Codemia
Practice similar problems with our interactive workspace, get AI feedback, and track your progress.
Browse Statistics QuestionsSample Answer
Problem Formulation
We want to evaluate the impact of a redesigned homepage on user engagement for Mastercard's recommendation algorithm using an A/B test. Our null hypothesis (H0) states that there is no difference in u...
Solution Approach
- Define Metrics: We will use click-through rate (CTR) as our primary metric for user engagement. 2. Sample Size Calculation: We need to determine the necessary sample size to detect a meanin...