Design ad click tracking and dashboard at scale
Company: Meta
Role: Software Engineer
Category: System Design
Difficulty: medium
Interview Round: Onsite
## A) Design an ad click/impression tracking system at 100M requests/second
Design a highly scalable system to record ad events (impressions/clicks) at **~100 million requests/second** and support an advertiser-facing dashboard.
Cover:
- APIs / ingestion endpoints
- Validation, abuse/fraud considerations (at a high level)
- Real-time vs batch aggregation
- Storage choices (hot vs cold), retention, and cost
- Serving queries for the dashboard (latency, freshness, filters)
- Consistency/idempotency (dedupe, retries)
- Fault tolerance and backpressure
## B) Design a “Status” feature (WhatsApp-like) and how to search statuses
Users can post ephemeral “status” updates (e.g., text/media) visible to allowed viewers for a limited time (e.g., 24 hours).
Design:
- Data model and APIs for creating/viewing statuses
- Privacy rules (who can see whose status)
- Feed generation for viewing
- **Search**: how users search statuses (by text/hashtags/owner), and how indexing respects privacy and expiration
- Performance: caching, indexing strategy, and scaling
- Deletion/expiration and compliance considerations
Quick Answer: This question evaluates a candidate's skills in designing large-scale distributed systems including real-time data ingestion and aggregation, storage and retention strategies, consistency/idempotency and fault tolerance, abuse/fraud and privacy-aware data modeling, and search/indexing for ephemeral content.