Track Success and Guardrail Metrics for Push Notifications

Quick Overview

Evaluates success and guardrail metrics for a push-notification feature in a networked travel app. Strong answers measure incremental itinerary engagement and retention, monitor opt-outs and fatigue, and design a cluster or graph-aware experiment to reduce spillovers.

Track Success and Guardrail Metrics for Push Notifications

Company: Meta

Role: Data Scientist

Category: Analytics & Experimentation

Difficulty: hard

Interview Round: Technical Screen

##### Scenario Designing and evaluating push notifications for a travel-recommendation mobile app (TripAdvisor-like). ##### Question What primary success metrics would you track for a new push-notification feature and which guardrail metrics would you include (e.g., uninstalls, unsubscribes)? The app has strong network effects (users share itineraries). How would you design an A/B test for the notification while mitigating interference? Describe the unit of randomization and why. ##### Hints Think engagement, retention, negative‐impact guardrails, and cluster-based randomization to reduce spillover.

Quick Answer: Evaluates success and guardrail metrics for a push-notification feature in a networked travel app. Strong answers measure incremental itinerary engagement and retention, monitor opt-outs and fatigue, and design a cluster or graph-aware experiment to reduce spillovers.

|Home/Analytics & Experimentation/Meta
Meta logo
Meta
Jul 12, 2025, 6:59 PM
hardData ScientistTechnical ScreenAnalytics & Experimentation
93
0

Track Success and Guardrail Metrics for Push Notifications

You are designing and evaluating a new push-notification feature for a travel-recommendation mobile app where users can create and share itineraries. Because users influence each other through sharing and engagement, the product may have network effects.

Constraints & Assumptions

  • Treat this as a product metrics and experiment-design question.
  • Assume logs exist for notification eligibility, send, delivery, open, click, itinerary actions, sharing, sessions, retention, unsubscribes, and uninstalls.
  • The notification should create incremental user value without causing notification fatigue or harming the broader network.
  • Experiment design should account for spillovers between connected users.

Clarifying Questions to Ask Guidance

  • What notification is being tested: itinerary reminder, recommendation, share prompt, or reactivation message?
  • Who is eligible, and how often can users receive it?
  • What is the main objective: itinerary creation, sharing, bookings, retention, or engagement?
  • How strong are social or network interactions in the product?

Part 1 - Define Success Metrics

What primary success metrics would you track for the new push-notification feature?

What This Part Should Cover Guidance

  • Incremental session starts, itinerary views, saves, edits, shares, bookings, return visits, and retention.
  • Notification-level metrics such as delivery, open rate, click-through, and downstream action rate.
  • User-level durable value rather than notification opens alone.

Part 2 - Define Guardrail Metrics

Which guardrail or negative-impact metrics would you include?

What This Part Should Cover Guidance

  • Push opt-outs, unsubscribes, notification mutes, uninstalls, churn, spam reports, complaints, dismissals, and app rating impact.
  • Fatigue metrics such as notification frequency, repeat exposures, declining open rate, and long-term retention.
  • Cannibalization of organic sessions or other channels.

Part 3 - Design a Network-Aware Experiment

Given the app's network effects, how would you design an A/B test to mitigate interference or spillovers?

What This Part Should Cover Guidance

  • Why individual-level randomization may create spillovers when users share itineraries or affect friends.
  • Cluster, geo, household, or graph-based randomization options and trade-offs.
  • Exposure logging, cluster balance, sample size, analysis of direct and network effects, and guardrail thresholds.

What a Strong Answer Covers Guidance

A strong answer defines success as incremental durable travel-planning value, monitors fatigue and harm, and proposes a randomization strategy that reduces spillovers while preserving enough power to make a launch decision.

Follow-up Questions Guidance

  • How would you choose clusters for randomization?
  • What if notification opens rise but retention falls?
  • How would you set a notification frequency cap?
Loading comments...