Root-causing a steady decline in notification delivery for an e-commerce app

Quick Overview

A metric-diagnosis question for a product manager at an e-commerce company whose app notification delivery is falling steadily. It tests defining the delivery metric, ruling out tracking errors, mapping where notifications get lost, using the gradual shape of the decline, and ordering the checks.

Root-causing a steady decline in notification delivery for an e-commerce app

Company: Google

Role: Product Manager

Category: Analytics & Experimentation

Difficulty: medium

Interview Round: Other

You are the product manager at an e-commerce company. Lately you have noticed that notification delivery for the company's app has been declining steadily. Identify the plausible causes of the decline, and explain how you would narrow them down to the real cause or causes. ```hint Follow one notification Trace a single notification from the event or campaign that triggers it to the moment it shows up on a phone, and ask at which step the numbers start to fall short. ``` ### Clarifying Questions - Which channel is declining: push notifications, in-app messages, email or SMS? - What does "delivered" measure here (accepted by the push service, confirmed on the device, or shown to the user), and has that definition or its tracking changed lately? - Is the decline in the total number delivered or in the delivered share of sends, and is it concentrated in one platform, app version, region or notification type? ### What a Strong Answer Covers - A precise definition of the delivery metric, and a check that the decline is real rather than a tracking or definition change - A map of each point where a notification can be lost, covering causes the company controls and causes it does not - Reasoning about why the decline is gradual rather than sudden, and which causes fit that shape - An order of checks, with the data or segment cut that confirms or rules out each hypothesis - A fix for the likely cause and monitoring that would catch the next decline early ### Follow-up Questions - Suppose the decline is confined to Android users on the newest OS versions. What would you check next, and what would you change in the product? - If users switching notifications off turns out to be the cause, how would you tell whether volume or relevance is driving it, and what would you change? - How would you set up alerting so the next decline is caught within days instead of weeks?

Overview: A metric-diagnosis question for a product manager at an e-commerce company whose app notification delivery is falling steadily. It tests defining the delivery metric, ruling out tracking errors, mapping where notifications get lost, using the gradual shape of the decline, and ordering the checks.

Google logo
Google
Jan 28, 2024
mediumProduct ManagerOtherAnalytics & Experimentation
0
0

You are the product manager at an e-commerce company. Lately you have noticed that notification delivery for the company's app has been declining steadily. Identify the plausible causes of the decline, and explain how you would narrow them down to the real cause or causes.

Clarifying Questions Guidance

  • Which channel is declining: push notifications, in-app messages, email or SMS?
  • What does "delivered" measure here (accepted by the push service, confirmed on the device, or shown to the user), and has that definition or its tracking changed lately?
  • Is the decline in the total number delivered or in the delivered share of sends, and is it concentrated in one platform, app version, region or notification type?

What a Strong Answer Covers Guidance

  • A precise definition of the delivery metric, and a check that the decline is real rather than a tracking or definition change
  • A map of each point where a notification can be lost, covering causes the company controls and causes it does not
  • Reasoning about why the decline is gradual rather than sudden, and which causes fit that shape
  • An order of checks, with the data or segment cut that confirms or rules out each hypothesis
  • A fix for the likely cause and monitoring that would catch the next decline early

Follow-up Questions Guidance

  • Suppose the decline is confined to Android users on the newest OS versions. What would you check next, and what would you change in the product?
  • If users switching notifications off turns out to be the cause, how would you tell whether volume or relevance is driving it, and what would you change?
  • How would you set up alerting so the next decline is caught within days instead of weeks?