Design a metric tree for improving long-term engagement in content discovery feed

Problem Statement Description

You are working on a large-scale content discovery feed used by international audiences across many languages, regions, devices, and network conditions. The feed’s purpose is to help users repeatedly discover relevant, trustworthy, and engaging content over time, not just drive short-session consumption spikes.

Your task is to design a metric tree for improving long-term engagement. The interviewer is looking for how you define the core engagement outcome, break it into measurable drivers, account for differences across international markets, and ensure the metrics are useful for product decisions rather than vanity reporting.

You should assume the feed includes personalized ranking, recommendations, creator or publisher content, notifications or re-entry surfaces, and user actions such as viewing, clicking, saving, sharing, following, hiding, or reporting content. The metric tree should help teams diagnose whether changes improve durable user value while protecting quality, trust, and accessibility.

The experience should consider:

- A clear definition of “long-term engagement,” including the time horizon, eligible user population, denominator, and retention or repeat-usage framing.

- How to separate acquisition, activation, short-term engagement, repeat engagement, and durable habit formation.

- International cohorts by country, language, maturity of market, content supply, device type, bandwidth, and new versus returning users.

- Instrumentation needs for feed impressions, sessions, content interactions, re-entry events, negative feedback, and downstream retention.

- Leading indicators versus lagging indicators, and how each would support product decision-making.

- Guardrail metrics for content quality, user trust, misinformation or policy violations, fatigue, notification abuse, creator ecosystem health, and accessibility.

- How the metric tree would help diagnose whether a decline or improvement is driven by ranking quality, content supply, localization, onboarding, notifications, or external seasonality.

The goal is to produce a structured measurement framework that a product team could use to evaluate experiments, prioritize improvements, and monitor whether the content discovery feed is building sustainable engagement for international users without sacrificing user trust or content quality.