← Practice Arena

Product Sense

Frameworks are scaffolding, not scripts — interviewers can smell a recited framework. Learn the structure, practice until it's internalized, then let it fade into the background of a natural conversation.

CIRCLES

Product design questions — 'Design X' or 'Improve Y'
  1. C
    Comprehend the situation. Ask clarifying questions: what's the goal (revenue? engagement?), what platform, what constraints? Never answer the question you assume — answer the one they meant.
  2. I
    Identify the customer. List 2–3 user segments, then pick ONE with a stated reason (size, underserved, strategic fit).
  3. R
    Report needs. For your chosen segment, list their needs/pains as user stories: 'As a ___, I want ___ so that ___.'
  4. C
    Cut through & prioritize. Rank the needs (impact × frequency × underserved-ness) and pick the top one or two. Say what you're NOT solving.
  5. L
    List solutions. 3 ideas minimum, at least one non-obvious. Range across ambition levels: quick fix, solid feature, moonshot.
  6. E
    Evaluate tradeoffs. Pick one solution and be honest about its costs, risks, and what you'd need to believe for it to work.
  7. S
    Summarize. 30-second recap: user, problem, solution, and the metric that tells you it worked.
Worked example: 'Design a fitness app for new parents' → Comprehend (goal: engagement, mobile), Identify (choose new mothers 0–6 months postpartum), Report (no time, unpredictable schedule, guilt), Cut (need: workouts that survive interruption), List (5-min resumable sessions / baby-inclusive workouts / sleep-synced plans), Evaluate (resumable micro-workouts: low build cost, direct fit), Summarize with adoption + weekly retention as metrics.

AARM / Metrics funnel

Metric questions — 'How would you measure X?'
  1. A
    Acquisition. How users arrive: installs, signups, by channel. Watch for quality, not just volume.
  2. A
    Activation. The 'aha moment' — the action that correlates with sticking around. Define it precisely (e.g., 'created first playlist within 24h').
  3. R
    Retention. Do they come back? Cohort curves, DAU/MAU. Almost always the metric that matters most.
  4. M
    Monetization. Conversion to paid, ARPU, LTV. Plus guardrails: refunds, churn, support tickets.
Worked example: 'Metrics for Instagram Reels' → Acquisition: % of IG users entering Reels. Activation: watched 5+ reels in a session. Retention: weekly Reels repeat rate. Monetization: ad load tolerance. North star: time spent OR creation rate — argue for one. Guardrails: creator retention, report rate, cannibalization of Stories.

Metric-drop investigation

Execution questions — 'Metric X dropped, investigate'
  1. 1
    Validate the data. Is the drop real? Check logging changes, tracking bugs, dashboard definitions. Say this first — it signals seniority.
  2. 2
    Clarify the shape. Sudden cliff or gradual decline? One-time or recurring? A cliff implies an event; a slope implies erosion.
  3. 3
    Internal causes. Releases, experiments, pricing changes, outages. Correlate timing with the deploy log.
  4. 4
    External causes. Seasonality, holidays, competitor launches, platform/OS changes, news events.
  5. 5
    Segment. Slice by geo, platform, user cohort, acquisition channel. A 15% global drop is often a 100% drop in one segment.
  6. 6
    Conclude & act. State most likely cause, immediate mitigation, and the monitoring you'd add to catch it earlier next time.
Worked example: 'Uber rides down 12% in one city' → data valid? → cliff on Tuesday → no release that day → external: city bus strike ended Monday (commuters returned to buses) → segment: drop concentrated in morning commute hours, confirming → action: commuter-hour promos, and add a local-events feed to anomaly monitoring.

Market entry / strategy

Strategy questions — 'Should company X do Y?'
  1. 1
    Clarify the goal. Growth? Profit? Defense? The right answer depends entirely on what the company is optimizing for.
  2. 2
    Market attractiveness. Size, growth, margins, competition. A quick guesstimate here earns credibility.
  3. 3
    Right to win. What assets transfer: users, brand, distribution, tech, data? No right to win = probably don't enter.
  4. 4
    Risks & costs. Cannibalization, focus dilution, regulatory, capital. What's the cost of being wrong?
  5. 5
    Recommend. Take a position. 'It depends' is a failing answer; a justified 'no' is a strong one.
Worked example: 'Should Swiggy enter medicine delivery?' → Goal: growth + order frequency. Market: large, growing, thin margins, regulated, entrenched players. Right to win: fleet density + app traffic, but no pharma supply chain or trust. Risks: regulatory, focus. Recommendation: partner with an existing pharmacy chain rather than build — gets frequency without the supply-chain build-out.

Watch real answers first

Exponent — PM Mock Interviews20–40 min each

Channel · Real candidates answering real product sense questions with interviewer commentary. Watch 3, then notice you can predict the feedback.

Product design answer walkthrough (CIRCLES in action)~20 min

Search · See the framework used naturally instead of recited — the difference interviewers care about.

Shreyas Doshi on product sense~60 min

Search · The best articulation of what 'product sense' actually is beyond frameworks.

Execution & metrics interview walkthrough~25 min

Exponent (search) · A full metric-drop investigation answered live — the exact structure from our Root-Cause framework.

A/B testing explained properly~15 min

Search · Significance, sample size, and guardrails — enough statistics to survive follow-up probes.

Drills

Set a 25-minute timer, answer out loud, then compare your structure to the linked framework. For feedback on your actual answer, take it to the AI Coach.