A/B Test Analysis — Ship or Kill?
Data & Analytics
What it tests
Statistical reasoning, business judgment, and ability to make a clear recommendation when data is ambiguous
Format
- 1Candidate receives real A/B test results with deliberately conflicting signals: conversion up but revenue down, mobile vs desktop split
- 2Must write structured analysis with clear ship/kill recommendation
- 3Must respond to conflicting stakeholder opinions (PM vs revenue analyst) with their actual position
What to look for
- Do they catch the Simpson's Paradox in the mobile/desktop split?
- Ship/kill recommendation: is it clear or fence-sitting?
- Response to stakeholders: do they take a position or try to please everyone?
- What they'd test next: is it informed by what they learned, or generic?
Adaptation guide
Replace the sample data with a real A/B test from your company. The more ambiguous the result, the more signal you get from the candidate's reasoning.
Full description
Format:
- Candidate receives A/B test results with conflicting signals
- Writes structured analysis with clear recommendation
- Must respond to conflicting PM vs analyst positions
Time: 90 minutes
What to look for:
- Statistical literacy (Simpson's Paradox detection)
- Clear recommendation, not fence-sitting
- Stakeholder communication under disagreement
Adaptation: Use a real A/B test from your company with ambiguous results.
Use this template in NouSpark
Start with this assessment template and customize it for your hiring process.
Create Assessment