Snap Data Analyst Case Interview — Product Analytics, AR Engagement, and Experimentation

This case simulates a real Snap data analytics problem aligned with Snapchat's camera-first, AR-driven product. You will be asked to investigate an engagement change tied to Lenses or Stories, read an experiment, and recommend an action while demonstrating Snap-style communication that is kind, smart, and creative. What the case covers: - Context: You are the analyst for the Lenses team. After a carousel update, the Lens apply rate and camera session length appear to have dropped in North America. You have event data and a recent A/B test on the carousel ranking logic. - Clarifying and metric design: Precisely define core metrics such as DAU, camera opens, Lens try-on vs apply, Snaps created, Snaps sent, Story views and completion rate, Spotlight watch time, retention (D1, D7), ARPDAU, and key guardrails like crash rate and latency. Align success metrics to user love and privacy-by-default principles. - Diagnostics plan: Propose a structured breakdown by platform, app version, country, new vs existing users, session type, and entry point to the camera. Build a funnel from camera open to Lens try-on to apply to snap creation and send. Consider seasonality, novelty effects from new Lenses, and potential logging regressions. - Experiment reading: Interpret summarized A/B results for the carousel update. Compute absolute and relative lifts with confidence intervals, discuss p-values or Bayesian posteriors, power and MDE, sequential peeking risks, novelty and fatigue, and whether to treat latency as a non-inferiority guardrail. Address CUPED or stratification if variance reduction is relevant. - SQL- and schema-level reasoning: Given a simplified events table (user_id, ts, country, platform, app_version, event_name, lens_id, story_id, send_type), outline how you would calculate apply rate, sessionized funnels, retention, and cohort analyses. Call out deduplication, time zones, late-arriving data, and bot filtering. - Data quality and experimentation pitfalls: Check event drops around release times, attribution windows for Lenses applied to Snaps vs saved to Memories, treatment contamination across features, and metric movement in related surfaces such as Stories or Spotlight. - Product sense and recommendation: Weigh trade-offs between engagement, wellbeing, and reliability. Provide a clear go or no-go call (ship, rollback, or iterate) and propose targeted follow-ups (e.g., revert on specific platforms or versions, re-rank for first-time users, or holdout by region). Define success criteria and next steps for monitoring post-launch. - Communication style: Synthesize the story crisply for PM, Engineering, and Design partners, using one-page or three-slide narrative: the problem, the evidence, the decision and risk. Reflect Snap values by being collaborative, direct, and user-centered. What interviewers evaluate: - Analytical structure and product intuition for camera, AR, Stories, and Spotlight - Statistical rigor in experiment design and inference - Practical SQL reasoning and data hygiene awareness - Clear, kind, decision-oriented communication tailored to cross-functional partners