Google·Product Manager·Onsite - Product Sense / Strategy·Senior

SeniorPrefer not to say

May 2026

Summary

Got a troubleshooting scenario at Google for a PM role, basically a 'nothing changed but things are broken' situation. The question had a lot of moving parts and I think I handled the structure okay but probably went too deep into the weeds on DB stuff when they wanted higher-level thinking.

Questions Asked (1)

A client says their web app has been intermittently slow during report generation and searches for the past few months, even though no code has changed. Walk through how you'd troubleshoot this: how do you figure out if it's one user or many, what metrics and logs you'd pull, and how you'd prioritize fixes after forming hypotheses.

Root Cause AnalysisProduct Analytics & MetricsTechnical Trade-offs

Author's notes

I started by scoping the blast radius, is this one user's browser or are multiple people hitting the same slowness.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by scoping the problem: determine whether the slowness affects one user, a segment, or all users, and correlate with time patterns. Then systematically gather metrics and logs to form hypotheses, prioritize fixes based on impact and effort, and validate with experiments.

Pro tip: Always tie technical symptoms back to business impact—quantify how many users are affected and how it impacts key metrics like report completion rate or user retention. This shows you think like a PM, not just an engineer.

1. Scope the Issue

Determine if the slowness is isolated to one user, a specific customer segment, or widespread. Check if it correlates with time of day, data volume, or specific report types.

2. Gather Data

Pull application performance metrics (e.g., response times, error rates), infrastructure metrics (CPU, memory, DB load), and logs (slow query logs, user activity logs). Segment by user, report type, and time.

3. Form Hypotheses

Based on data, hypothesize root causes: e.g., increased data volume, database contention, third-party API latency, or infrastructure scaling issues. Prioritize hypotheses by likelihood and impact.

4. Prioritize Fixes

Evaluate potential fixes based on user impact, effort, and risk. Consider quick wins (e.g., query optimization) vs. long-term solutions (e.g., architecture changes). Align with business goals.

5. Validate and Monitor

Implement fixes incrementally, measure impact via A/B tests or before/after metrics, and set up monitoring to prevent recurrence. Communicate findings and next steps to stakeholders.

Key Points to Mention

User segmentation: distinguish between single-user, cohort, and global issues
Key metrics: p95/p99 latency, error rates, throughput, database query times, CPU/memory utilization
Log analysis: slow query logs, application logs, and user session traces
Hypothesis-driven approach: prioritize based on data, not guesswork
Prioritization framework: impact vs. effort, RICE or similar
Business impact: tie technical issues to user experience and revenue

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.