Your goal: design, size and read out an A/B test the way a product data scientist would.
What this is (20–25 minutes)
- One main product case: you lead the design; Daniel answers clarifying questions.
- Follow-ups on metrics, randomization, sample size, novelty and network effects.
- A results table appears on the notepad: read it out and make a recommendation.
- If time allows, a shorter second case.
How it works
- Daniel behaves like a real interviewer: he probes, but he does not teach, hint or tell you whether you are right.
- Your scored report arrives after the call.
What scores well
- Clarifying the goal before naming metrics
- Primary, guardrail and north star metrics with clear definitions
- A randomization unit you can defend, and sample-size reasoning with numbers
- Checking the validity of results before reading them, and naming pitfalls unprompted
- Ending with a clear decision and its trade-off
Tips
- Ask clarifying questions; state assumptions when Daniel says to.
- Rough math out loud is fine; the reasoning matters more than exact digits.