Strategy question

Should Facebook launch a feature suggesting to "Unfriend this friend"? If yes, how should it decide which friends to suggest to unfriend?

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Sensitive feature judgment: weighing user value against emotional and trust risk, and designing selection criteria carefully if proceeding.

State the core tension up front: suggesting to unfriend someone touches real social relationships and could feel invasive, hurtful, or presumptuous if done poorly.
Weigh the case for building it: some users have hundreds of stale connections cluttering their feed and privacy settings, and a cleanup tool could genuinely help.
Weigh the case against: incorrectly suggesting to unfriend a close friend or family member, even occasionally, causes real reputational and trust damage.
If proceeding, recommend it be entirely user-initiated and private (a self-serve 'clean up my friends list' tool) rather than an unprompted push notification suggesting specific people.
Propose selection criteria only within that user-initiated context: extremely low or zero interaction over a long period (for example two years), never framed as a judgment about the relationship.
Define success as opt-in usage of the tool and low negative sentiment or complaint rate among users who use it.

Leads with the real emotional and trust risk of this feature rather than jumping straight into selection criteria.
Recommends a meaningfully different, safer version, self-initiated cleanup rather than proactive suggestions, showing good judgment about what to actually ship.
Sets conservative, low-risk selection criteria (very low interaction over a long period) only within that safer, opt-in framing.

Jumping straight to selection criteria without first weighing whether the proactive, unprompted version should be built at all.
Ignoring the reputational risk of the algorithm getting a suggestion wrong for an important but low-interaction relationship, like a family member overseas.

What would you do if user research showed most people found even the opt-in version upsetting?
How would you word or frame this feature to minimize emotional harm?