Katkat
Back

Define metrics and data model for product features

MetaData Engineer · Technical Screen
MediumData EngineerTechnical ScreenSystem Design

Last updated

This problem evaluates your ability to translate product goals into actionable metrics and design a scalable, reliable data schema to support those metrics.

Scenario

Meta is launching a new feature called 'Group Highlights,' which allows group admins to pin a summary of the most popular posts from the last 7 days. As a Data Engineer, you are responsible for the data infrastructure supporting this feature.

Requirements

  1. Metrics: Define 2-3 key metrics to measure the feature's success and explain why they matter.
  2. Data Modeling: Design the event logging structure required to calculate these metrics. Assume this is a high-traffic feature with millions of daily active groups.
  3. Architecture: Outline how these events flow from the client to an analytics-ready table (e.g., in a data warehouse like Hive or Snowflake).
  4. Metric Drop: If the 'daily active users interacting with highlights' metric suddenly drops by 20%, describe your step-by-step approach to debugging the root cause.

Constraints & Expectations

  • The data model must account for event attributes (e.g., user_id, group_id, timestamp, interaction_type).
  • Consider the storage implications of high-frequency events.
  • Your answer should address both the 'what' (metrics) and the 'how' (pipeline/storage) of the data lifecycle.

Hints

Hint 1

Focus on metrics that capture both the creation of highlights by admins and the consumption by end-users.

Hint 2

Think about how to store event data in a schema that supports efficient aggregation by group, date, and user segment.

Hint 3

When diagnosing a metric drop, distinguish between technical failures (e.g., instrumentation bugs) and product-level changes (e.g., UI updates or external seasonality).

Follow-up questions

  • How would you handle late-arriving data or data quality issues in your pipeline?
  • If we wanted to personalize the highlights for different users, how would your data schema need to evolve?
  • What strategies would you implement to ensure data privacy and compliance (e.g., GDPR/CCPA) when logging user interactions?