Product Manager — GE CX Essentials

The altitude test: answer as an SVP, and stay there — the documented failure mode is opening strategic and sliding into tactical feature-talk halfway through.

Likely shapes

• Should Google offer a StubHub competitor — sell sports, concert, theater tickets? Official
• How would you grow [product]? · How would you monetize [product] more effectively? Official
• How would you 10x Google Cloud IoT? · What are the biggest trends in [industry]?
• You're the CEO of Uber — what's your 10-year strategy?
• $100M to build an OLTP database product: cost-neutral and growing in 5 years — how?
• Where does the creator economy go in 10 years, and what should Google build there?

Method

• Business objective first → then name which signals you're weighing (market, competitors, users, trends, metrics) and why.
• Cover monetization, pricing, and cost angles explicitly.
• Bring a real opinion on the AI-agents CX market: Salesforce, Sierra, Decagon vs Google.

Practice set — drill out loud, 4-step method: objective → structured options → trade-offs → recommendation

1. Should Google offer a StubHub competitor — sell sports, concert, theater tickets? Official
2. How would you turn Facebook Events around?
   Sourced model answer

   Clarify what's broken → engagement, whole funnel. Structure first: three engagement levels — awareness → basic (RSVPs) → deep (creating/commenting). Ideas attach to levels: event suggestions from status posts; countdown-urgency UI; suggest uninvited friends; creation-tips pop-ups ("adding a picture increases comments 20%"); auto-populated events; Live streaming in event pages. Interviewer hint (drop-off is basic→deep) redirects focus. Trade-offs named: tips risk clutter; auto-events have grim edge cases (suggesting an anniversary event for someone deceased); Live integration costs another team and risks cannibalization. Recommendation: creation tips first.

   IGotAnOffer strategy guide (full answer) — source · archived in vault sources/

3. Imagine you're the CEO of Uber — what's your 10-year strategy?
   Sourced model answer (Mark, ex-Google Sr PM)

   Porter's Five Forces, worn loosely: competition (rideshare + logistics → edge is network effects and execution) · customers (loyalty follows convenience → AI for routing/prediction/personalization: "the best AI doesn't just respond; it anticipates") · suppliers (drivers Uber doesn't own; autonomous is the long-term bet) · substitutes (weak for people, serious for goods) · new entrants (decentralized transport — defend with product quality). Synthesis: four pillars — AI-first, autonomous, network effects, product quality; names what he skipped (brand, M&A, margins); prioritizes AI; lands on an invented North Star: Monthly Move Units — people + packages + goods moved. "If MMU is growing, Uber is winning."

   IGotAnOffer 8-types guide · video version

4. How would you 10x Google Cloud IoT? (reported Google Cloud PM question)
5. How would you grow [pick: Customer Engagement Suite]? How would you monetize it more effectively? Official
6. You have $100M VC investment to build an OLTP database product — cost-neutral and growing within 5 years. (reported Google Cloud)
7. What are the biggest trends you see in the customer-service/AI-agents industry? (adapt of reported "security industry" question)
8. How do you see the creator economy evolving over ten years — what should Google build there?
9. Pretend Google wants to acquire iRobot — what do you look for, how do you position?
10. How would you double YouTube's user base? · How would you map an unmapped area?

Two competencies in one session — expect a design question wearing a stakeholder-navigation layer.

Product vision shapes

• How would you improve restaurant search? · Improve Google Maps? Official
• Next great feature for Google Search? Official
• Design an alarm clock for a person with a visual impairment Official
• Design X for Y: Maps for kids · elevator for blind users · shopping for elderly + TAM
• Design an AI assistant for [segment] — possibly followed by “now prototype it”

Problem-space shapes

• Conflicting product requirements — what or who decides which takes the hit? Official
• Latent field failure driving up returns and support contacts — manage it Official
• Largest customer demands an off-roadmap feature; Sales went straight to Engineering Official
• Ad-ranking change dropped CTR but raised browse time; Marketing and Eng disagree on root cause

What scores

• A specific user segment — “people who want to be productive” fails.
• One unexpected insight; narrow v1 → 2–3-year north star; explicit deprioritization.
• “A vision specific enough to be proven wrong.” Problem framing IS part of the answer.

Practice set — BUS method: objective → segment + pain points → scored solutions (value 1–3 × ease 1–3)

1. How would you improve restaurant search? Official · How would you improve Google Maps? Official
2. Design an alarm clock for a person with a visual impairment. Official
3. Design a computer keyboard.
   Sourced model answer (full BUS walkthrough)

   Clarify → gamers, no mouse, maximize revenue. Segment: casual vs pro → casual (bigger market fits the revenue goal). Seven pain points listed; prioritize the two tied to winning (slow/dusty keys; weak shortcut software) because a gamer's primary objective is performance. Solutions scored value+ease: lighter-press keys (6), per-game shortcut sets (5), macro buttons (6)… Trade-offs: sensitivity vs mis-presses; customization vs setup complexity. Recommendation: sensitive keys + shortcut software + macro buttons. The scoring table is the move — it turns brainstorm into decision.

   IGotAnOffer design guide — source · archived in vault sources/

4. Design something for Google Meet · a camera for the elderly · a product to find a doctor · a first-time-flyer booking experience. (reported Google)
5. What would you design as PM of Google Flights if the strategy had to be monetized? (reported Google)
6. Design an AI assistant for [segment], then: "now prototype it — why did you prompt it that way?" (reported Gemini loop)
7. How do you resolve conflicting product requirements — what or who decides? Official
8. Your largest customer demands an off-roadmap feature; Sales went straight to Engineering. What do you do? Official
9. Ad-ranking change dropped CTR but raised browse time; Marketing and Eng disagree on the root cause — how do you approach the problem space?
10. Leadership chose a direction you disagree with but must implement — how do you get yourself and your team on board?
    Sourced model answer (ex-Google EM)

    State assumptions out loud (direction non-negotiable; no layoffs) → understand leadership's reasoning before reacting → deliberately take time to get personally on board ("use time as a tool instead of being subject to it" — credited to a former manager) → meet leads first so they become advocates, then the broader team → stay available for 1:1s. Don't drift into storytelling — it's a hypothetical; stay in it.

    IGotAnOffer Google hypothetical guide (method stated as based on Google's official prep video) — source

11. How would you convince a Google Cloud customer to expand their cloud services? (reported Google GCA — closest to this role's world)

Product analysis — expect pushback by design

• How many messages per second does Gmail receive? Official
• 30% usage change — what do you do? · Gmail down 15% over the weekend — what do you do? Official
• What are the key metrics for an API in a cloud? — most role-relevant reported question
• First hypothesis gets a flat “no”: exhaust the space — bug → UI change → rollout → behavior shift → seasonality → external.

Execute with judgment — where the AI material lands

• App is 1 month from launch; internal feedback says not ready, CSAT below target — what do you do? Official
• VC offers $20M for any tech product — walk through: Problem, Solution, User, Monetize, TAM Official
• What's in your monthly business review deck? · When and how do you sunset a product? Official
• “Users complain Gemini is confident but wrong — fix it.” Follow-ups: north star? guardrails? A/B design? model issue vs UX issue? roll back vs iterate?
• AI fundamentals to be fluent in: model selection criteria · evaluating a prompt · RAG · RLHF · DPO · rule-based vs NN vs LLM · what tasks suit LLMs vs don't.

Googleyness & leadership — five scored dimensions

• Intellectual humility · comfort with ambiguity (primary emphasis for AI roles) · collaborative spirit · conscientiousness · integrity.
• Conflict with eng on a slipping timeline · buy-in for a controversial change · a time data reversed your decision · impactful project you led.
• STAR is Google's officially recommended structure; quantify results; end with what changed in you.

Practice set A — analysis (estimation: clarify → validate plan → round + calculate aloud → self-sense-check · metric change: DEC · metric definition: GAME)

1. How many messages per second does Gmail receive? Official
2. You notice a 30% change in usage — what do you do? · Gmail received-mail down 15% over the weekend — what do you do? Official
3. How much revenue does YouTube make per day?
   Sourced model answer (estimation, 4-step)

   Clarify → US-only video ad revenue, per-1,000-views. Validate the plan first: users → ad views → revenue. 300M ÷ 8 age bands; watch rates 75/75/50/10% each justified from personal experience ("my niece watches Shorts about that often") → ~150M daily users × 10 videos × 50% with ads × $10 CPM ≈ $7.5M/day. Self-sense-check: "half the country watches daily — reasonable." ⚠️ The source's own annual cross-check figure is inconsistent — the method is the practice target, and saying "this number looks off" out loud is itself the step-4 skill.

   IGotAnOffer estimation guide — source · useful anchors: ~$14.90 avg CPM; ~300M US pop; 80-yr life expectancy

4. YouTube watch-time per session is down 5% — investigate and report to the executive team.
   Sourced model answer (DEC root-cause — best rehearsal for Google's pushback style)

   Define: exact metric, window, segment → mobile only, MoM. Explore a MECE internal/external tree, interrogating each branch: data accuracy (cross-check vs time-on-page) → seasonality (past >1% drops were outages; uptime 100%) → product changes → found it: player UI change made "Send to device" 2× bigger and "Full screen" half-size → mis-taps end sessions early. Then — the key move — still sweeps the remaining branches before concluding. Conclude: cause, mechanism, ruled-out alternatives, next steps (revert; revisit the change's original goals).

   IGotAnOffer metrics guide (full dialogue) — source

5. What are the key metrics for an API in a cloud? (reported — the most role-relevant metrics question found; use GAME + input vs output metrics)
6. Estimate: Google Docs created per day · Gmail's yearly ad revenue · queries answered by Google per second · storage for all Street View imagery. (all reported Google)
7. Instagram feed engagement drops 10% — what do you do? (practice as question-asking: internal → external → user behavior; sourced answer traces it to a hack of large accounts — see practice-bank note)

Practice set B — execute with judgment + AI

1. Your strategically-important app is 1 month from launch; dogfood says it isn't ready (CSAT below target). What do you do? Official
   Sourced building block: phased-rollout answer

   From the behavioral bank's "decision with limited information": flaky third-party dependency two weeks pre-launch → gathered impact input from eng/QA/support → proposed phased rollout to 10% of users with monitoring + a prepared fallback plan → error rates stayed acceptable, issue patched in a week, marketing deadline met. Reuse this structure: quantify the gap → smallest reversible launch → explicit rollback trigger.

   IGotAnOffer behavioral guide, sample answer 5 — source

2. I'm a VC offering $20M for any tech product — walk me through getting started (Problem, Solution, User, Monetize, TAM). Official
3. Pick a product: its goals, and what's in your monthly business review deck for leadership? Official · When and how do you sunset a product? Official
4. Prioritize three competing projects.
   Sourced model answer (RICE, fully numbered)

   Photo app: messaging update (A) vs desktop app (B) vs crop tool (C). Agree objective (engagement) → Reach 250/350/600, Impact 0.5/3/2, Confidence 80/50/80%, Effort 1/6/3 → scores 100/~88/320 → raw order C-A-B. Then judgment overrides the formula: B was promised to stakeholders early → final C-B-A, with the cannibalization trade-off on the free crop tool named. Framework drafts, judgment decides — that IS "execute with judgment."

   IGotAnOffer 8-types guide — source

5. Users complain Gemini is confident but wrong — how do you fix it? Follow-ups: north star? guardrails? A/B design? model vs UX issue? roll back vs iterate? (reported Gemini loop — THE AI-judgment drill)
6. Your launched model is underperforming — triage and prioritize next steps. (asked by an eBay Sr Technical PM; wants structured diagnosis: data quality → model accuracy → user adoption)
7. AI fluency rapid-fire: criteria for selecting a model · how do you evaluate a prompt · RAG · RLHF · DPO · rule-based vs NN vs LLM · what tasks suit LLMs vs don't. (reported AI-PM bank; answer from your prep cheat-sheets)
8. Design the high-level system for how Gemini answers a query — in Google Drawings. (reported Gemini loop; model the flow on the sourced notification-system walkthrough in practice-bank)

Practice set C — Googleyness & leadership (STAR + an explicit lessons close, ~2 min, quantified)

1. What's one accomplishment you're most proud of? · Describe a project you managed start to finish, including tools and stakeholder communication. (both reported Google)
2. Tell me about a time you faced conflict with engineering when a release timeline slipped — and: you've committed to a release but eng says it can't ship in time. (reported Google; the 8 sourced SPSIL sample answers in practice-bank are the templates — rebuild each with a Newly story)
3. Tell me about a time you failed and what you learned. · A time data reversed your decision.
4. Get buy-in for a controversial change. · Lead without authority.
5. What would happen if Google fired all of its PMs tomorrow? (reported Google curveball)
6. If you had coffee with Sundar Pichai, what would you talk about? (reported Google)