# Reduce Stale Listings in a Job Marketplace Design a system that detects and resolves stale job postings in a large marketplace. A stale listing may have been filled, closed on the employer's own site, abandoned by its owner, duplicated, or left open despite no longer accepting applications. Cover freshness signals, classification, employer workflows, user experience, automatic actions, measurement, and abuse or false-positive controls. ### Constraints & Assumptions - Listings arrive from direct employers, feeds, crawlers, and external applicant-tracking systems. - Sources vary in reliability and update frequency. - Removing a valid job harms employers and applicants; retaining dead jobs also harms trust. - Employers may not respond to verification requests. - Decisions and source evidence must be auditable and reversible. ### Clarifying Questions to Ask - What business state counts as stale versus merely old? - Which sources permit active status checks and at what rate? - How quickly should a high-traffic listing be reverified? - Can the marketplace pause applications without deleting the listing? ### What a Strong Answer Covers - Source-specific freshness identity and positive and negative evidence - Risk-based verification scheduling and an explainable stale-score model - Graduated actions such as warning, deprioritizing, pausing, and closing - Employer confirmation, appeals, reversible state transitions, and audit - Precision-focused evaluation and applicant and marketplace outcome metrics ### Follow-up Questions 1. How would you handle an employer feed that repeatedly reopens closed jobs? 2. What signals distinguish an old evergreen role from an abandoned listing? 3. How would you test automatic closure without hiding valid inventory? ```hint Separate evidence from action Maintain a freshness state and confidence score, then choose an action whose severity matches the evidence and cost of a false positive. ```