Design a Multi-Organization Advertising Platform with Performance Statistics

Design an advertising platform that displays ads from multiple organizations and collects statistics to compare which ads perform better or worse. ### Constraints Ad formats, selection policy, traffic, billing, targeting, and the definition of performance are unspecified. Clarify the required features before adding auctions or personalization. Preserve organization boundaries and distinguish a valid ad display from a raw request or duplicate event. ### Clarifying Questions - Which organizations can create and manage ads, and what approval or eligibility rules apply? - What counts as an impression, click, conversion, or other success event? - Is performance compared by raw counts, rates, revenue, or another objective? - How are duplicate, delayed, or invalid events handled? ```hint Compare rates with their denominators An ad with more clicks may simply have received more exposure; the event definitions and opportunity counts matter. ``` ### What a Strong Answer Covers - Ad and organization management, eligible-ad serving, and clear event instrumentation. - Durable collection, deduplication, aggregation, and tenant-aware reporting. - Correct performance interpretation, delayed data, and separation of association from causal impact. ### Follow-up Questions - How would you reconcile a click that arrives before its impression event? - What experiment would support a causal claim that one ad performs better?