MicrosoftVerified interview challenge
Experiment assignment bug hunt
Check whether an A/B assignment helper produces stable groups and trustworthy exposure data.
Unlock with Full AccessThe challenge
The same user should keep the same assignment for an experiment across processes, deployments and repeat visits. Different experiments must remain independent, and recording an assignment is not the same as showing that variant. Review the helper for faults that undermine allocation or experiment measurement.
What you'll practise
Stable assignment
Check whether the same user and experiment produce a consistent result over time.
Rollout boundaries
Inspect percentage handling and independence between experiments.
Exposure measurement
Distinguish looking up a variant from actually showing it to a user.
Experiment assignment helperCode included in the full challenge
Your review
Explain the assignment and measurement failures, with a fix for each.
The code, instructions and feedback are included with Full Access.