Design an A/B test for a recommendation algorithm
Last updated: June 8, 2026
Quick Overview
Design an experiment to test the impact of new pricing tiers. Include sample size calculation, metrics, and analysis plan.
Compass
June 8, 20266
5
3,511 solved
Design an experiment to test the impact of new pricing tiers. Include sample size calculation, metrics, and analysis plan.
Compass asks this during the Onsite to assess your experimentation skills. They want to see how you define success metrics, design controlled experiments, and interpret results with appropriate statistical rigor.
What the Interviewer Expects
- Define clear success metrics aligned with business goals
- Propose a basic experimental design with control and treatment groups
- Interpret results correctly and draw reasonable conclusions
- Identify obvious confounding variables
Key Topics to Cover
How to Approach This
- Define success metrics carefully. A good metric is measurable, actionable, and aligned with business goals.
- Run experiments long enough to account for novelty effects and weekly seasonality.
- Use funnel analysis to identify where users drop off for maximum optimization impact.
- Segment results by key dimensions (platform, country, user cohort) to catch hidden patterns.
- Consider network effects and interference between treatment and control groups.
Possible Follow-up Questions
- What would you do if a stakeholder wants to end the experiment early because initial results look good?
- How would you handle seasonality in your experiment?
- What if you discover a bug in the logging during the experiment?
Sharpen Your Skills on Codemia
Practice similar problems with our interactive workspace, get AI feedback, and track your progress.
Browse Analytics QuestionsSample Answer
Problem Setup
The analytical question we want to answer is: "What is the impact of the new pricing tiers on user engagement and conversion rates for Compass's recommendation algorithm?" To effectively analyze this,...
Methodology
For the A/B test, we will randomly assign users to either the control group (existing pricing tiers) or the treatment group (new pricing tiers). The statistical approach will involve:
- **Sample Siz...
Submit Your Answer
