Interview questions in Product Management, page 49
Define and calculate Weekly Active Users (WAU) for Slack
Tests translating a business metric to a technical spec. Define 'active' by key actions (sending messages, not just opening), then COUNT(DISTINCT user_id) on an events table, filtering out bots and background syncs. A red flag is a generic definition.
How would you define and calculate Weekly Active Users (WAU)?
This tests your product sense and technical precision in defining a core business metric. A great answer defines 'active' with specific user actions, outlines the SQL/event-based calculation, and discusses pitfalls like bots and background events.
How do you fix a disconnect between team velocity and value?
This tests your ability to distinguish output from outcome. A good answer investigates the "why" with the PO, proposes experiments to test value hypotheses, and suggests refining processes like sprint planning, rather than just blaming the PO or trying to…
Designing a tiered entitlements backend
Plans map to features and limits, subscriptions link account to plan, an entitlement check enforces access; handle upgrade/downgrade with effective dates and proration.
Cross-platform stateful onboarding sync
Store onboarding state server-side keyed to the user, expose idempotent step-completion APIs, and push updates to other clients.
How do you shift analytics from growth to profitability?
This tests your ability to translate business strategy into technical changes. A great answer connects profitability drivers to specific updates in event taxonomy, data models, and dashboards. A red flag is ignoring core financial metrics like LTV and CAC.
How do you prioritize tech debt against new features?
This tests your ability to translate engineering risk into business impact. A great answer frames debt as a business cost (slowdown, bugs), quantifies the impact, and proposes a specific capacity allocation. A red flag is framing it as an 'us vs. them' battle.
Event-driven sync between billing and CRM
Billing emits a tier-changed event via outbox to a broker, CRM consumes idempotently with retries and DLQ.
Dynamic personalized onboarding architecture
A segmentation pipeline, a serving layer choosing task order per segment, an experimentation engine, and a feedback loop measuring activation.
Define idempotency in data processing and give an example
Tests your grasp of distributed systems reliability. Define idempotency (N>1 runs = 1 run), explain its role in fault-tolerant retries, and provide a concrete example using transaction IDs. A red flag is confusing it with immutability.
Team struggles with Sprint spillover. How do you investigate?
This tests your ability to diagnose systemic issues, not just recite Scrum rules. A great answer prioritizes data gathering, categorizes root causes (refinement, tech debt, etc.), and proposes team-led experiments.
Building a cross-product case for shared-lib debt
Map each product's exposure and blast radius, quantify risk and migration cost, propose a coordinated phased plan with an owner.
Cold-start to personalized feed transition
Start with popularity or onboarding-declared interests, collect implicit signals like dwell and clicks, then blend toward personalized as confidence grows.
How would you visualize three years of monthly revenue?
This tests your grasp of time-series visualization and data integrity. A strong answer picks a line chart, insists on a zero-based Y-axis and clear labels, and adds context like seasonality.
How do you handle a critical bug mid-sprint?
Tests your pragmatism and ability to navigate crisis. A good answer involves triaging the bug's impact, assessing the cost to the Sprint Goal, empowering the Product Owner to make a trade-off, and transparently adjusting the plan.
Monorepo vs polyrepo for a product portfolio
Monorepo eases atomic shared-component changes and consistency but needs tooling; polyrepo gives autonomy but version drift; tie choice to deploy independence.
Peeking in A/B tests and how to mitigate it
Peeking is checking significance repeatedly and stopping at the first significant result, which inflates false positives; mitigate with fixed sample sizes or sequential…
Visualize two continuous and one categorical variable?
Tests your ability to map data to visual encodings. A great answer starts with a scatter plot, then adds the categorical data using color, shape, or faceting, explaining the tradeoffs. A red flag is suggesting a 3D chart, which is difficult to read.
Team Velocity Dropped for 3 Sprints. How Do You Diagnose?
This tests your ability to use data for diagnosis, not blame. A good answer gathers quantitative (cycle time, unplanned work) and qualitative data, then presents hypotheses to the team.
Strict tenant isolation in a multi-tenant data layer
Choose silo, pool, or bridge by risk; enforce tenant scoping at multiple layers with RLS; encrypt and audit.
We are hiring for this. Every open role lists the topics its interview covers, so you can prepare for the real thing rather than guessing.
See open roles