AARRR Growth Framework & Metrics Questions

Understand the AARRR model (Acquisition, Activation, Retention, Revenue, Referral) as a mental model for thinking about growth. Know how to map growth problems to specific stages: acquisition challenges differ from retention problems. Learn standard metrics for each stage (CAC, LTV, activation rate, retention cohorts, viral coefficient). Practice identifying which stage is most relevant for a given problem.

EasyTechnical

56 practiced

Define Customer Lifetime Value (LTV). Given ARPU (average revenue per user per month) = $10 and monthly churn rate = 5%, compute the simplified LTV using LTV = ARPU / churn. Explain the assumptions behind this formula and two ways you would refine the LTV estimate for business decision-making.

HardTechnical

72 practiced

An A/B test shows a 3% increase in conversion (p=0.04) but a 1.5% decrease in average order value (AOV) (p=0.06). How would you decide whether to ship the change? Discuss calculation of combined business impact, statistical significance and power considerations, possible segmentation, guardrails, and practical rollout approaches.

Sample Answer

Situation: We ran an A/B where Variant B shows +3.0% conversion (p=0.04) and −1.5% AOV (p=0.06). As a PM I must decide based on net business impact, robustness of stats, and risk controls.

1) Calculate combined business impact- Use expected revenue per visitor (RPV) = conversion_rate * AOV.- Baseline RPV = CR0 * AOV0. New RPV ≈ CR0*(1+0.03) * AOV0*(1−0.015) ≈ Baseline * (1+0.03)*(1−0.015) ≈ Baseline * 1.01455 → ≈ +1.455% RPV.- Translate to projected monthly/annual revenue to quantify dollars and compare to costs.

2) Statistical significance & power- Conversion p=0.04 is statistically significant at α=0.05; AOV p=0.06 is a near-miss. But AOV typically has higher variance → underpowered for small effect.- Compute confidence intervals for both metrics; compute CI for RPV or use delta method / bootstrap on per-session revenue to get direct CI for business metric.- Check sample sizes and power: if power <80% for AOV change, consider more traffic or longer test.

3) Segmentation & sanity checks- Break down by device, geography, new vs returning, traffic source. If positive conversion concentrated in low-AOV segments, aggregate RPV could differ.- Check funnel behavior, cancelations, returns, and downstream metrics (LTV, retention). Ensure no signal of fraud or instrumentation bug.

4) Guardrails & rollout plan- If combined RPV CI > 0 and business impact material, roll to progressive rollout (e.g., 10% → 50% → 100%) with kill-switch and monitoring.- If combined RPV CI includes zero but point estimate positive, run an extended test with power calculation targeting AOV variance, or do a controlled ramp with revenue/returns guardrails.- Implement health metrics (support tickets, refund rate, retention) and alerting.

5) Decision framework (practical)- If RPV boost is positive and robust across segments and downstream metrics, ship with progressive rollout.- If RPV ambiguous: continue test/collect more data or run targeted experiments to recover AOV (e.g., pricing messaging) before full launch.- Always present stakeholders with dollar impact scenarios, CI, and contingency plan.

This approach balances statistical rigor with business practicality and risk mitigation.

HardTechnical

61 practiced

Build a high-level model to forecast MRR over the next 12 months using current retention cohorts, expected new acquisition by channel, and an upcoming pricing change. Describe required inputs, equations or approach (cohort build-up), key assumptions, and how you'd run sensitivity analysis to show best/worst cases.

Sample Answer

High-level approach:- Build a monthly cohort-based MRR forecast: start with existing cohorts (by signup month and channel), propagate retention and upgrades/churn monthly, add projected new acquisitions by channel and apply pricing change when it takes effect.

Required inputs:- Current MRR by cohort (cohort = month acquired × channel)- Monthly retention curve (survival rate r_t for month t since acquisition) per cohort/channel- ARPU / price per plan and distribution by cohort- Forecasted new customers by channel per month (or marketing spend → CAC → conversion)- Timing and magnitude of pricing change (percent change by plan)- Upgrade/downgrade probabilities, expansion MRR rates, and churn lift/drag from pricing change- Seasonality factors, macro scenarios

Equations / cohort buildup:- For each cohort c acquired in month m: Customers_c,m,t = Customers_c,m,0 × r_c,t (t = months since m) MRR_c,m,t = Customers_c,m,t × ARPU_c,t- Total MRR_month = sum over all cohorts (historical + new cohorts for that month)- Apply pricing change: ARPU_c,t = ARPU_c,t × (1 + delta_price) from effective month; adjust r_c,t if price affects retention.

Key assumptions to document:- Retention curve shape and whether it’s stationary or improving- Elasticity of churn/expansion to price changes- New acquisition ramp by channel and CAC limits- No major product launches or market shocks unless modeled

Sensitivity analysis / scenarios:- Build three baseline scenarios (Base / Upside / Downside) by varying: acquisition volume ±X%, retention curve ±Y ppt, price elasticity of churn ±Z%- One-way sensitivity: vary each input (e.g., retention month-1 to month-12) and show impact on 12‑mo MRR- Tornado chart to rank drivers- Monte Carlo (if available): sample distributions for top 3 uncertain inputs to generate confidence intervals

Outputs & visualization:- Monthly MRR waterfall and cohort heatmap (MRR by cohort over time)- KPI table: starting MRR, net new MRR, churn, expansion, ending MRR- Scenario-band chart with 10/50/90 percentiles and key assumptions annotated

Execution plan:- Pull cohort data from analytics (segment by channel, plan)- Fit retention curves, estimate price elasticity from A/B or historical price changes- Build model in spreadsheet/Looker Studio; validate with backtest (forecast past 6 months)- Deliver scenarios and recommendations (e.g., phased pricing, communications to mitigate churn).

HardTechnical

57 practiced

Your referral program is being gamed, resulting in suspicious spikes in invites and low-quality signups. Propose a cross-functional plan to responsibly increase the viral coefficient: include detection heuristics for fraud, product flow changes to improve legitimate invites, incentive design favoring quality, and key metrics to monitor referral health.

Sample Answer

Situation: We’re seeing suspicious spikes in referral invites and many low-quality signups, harming LTV and product trust. Goal: responsibly raise viral coefficient while reducing fraud and improving signal on invite quality.

Plan overview (cross-functional): product, data/analytics, engineering, trust & safety, marketing, legal.

1) Fraud detection heuristics (Data & T&S):- Real-time score combining: rapid invite bursts per account, multiple invites to disposable/email domains, identical message patterns, device/IP pools, velocity on activation, short session time, no engagement within 7 days.- Use labeled examples + unsupervised clustering (anomaly detection) to find new attack patterns.- Flag tiers: soft-watch (monitor), challenge (email/phone verification, captcha), hard-block (suspend/refuse reward).

2) Product flow changes (Product & Eng):- Throttle invites per time window with graceful UX (e.g., “You’ve reached your weekly invite limit — try later”).- Require lightweight friction on bulk invites: optional contact import but limit batch size; progressive verification when thresholds hit.- Add contextual invite channels: deep-links, social share cards that personalize message and make organic invites more appealing.

3) Incentive design favoring quality (Growth & Finance):- Move from instant reward on signup to milestone-based rewards: reward only after invitee completes meaningful actions (e.g., 7-day retention, first purchase, or completes onboarding).- Use split rewards: give small immediate acknowledgment to referrer and larger deferred reward when quality criteria met.- Cap total referral rewards per user per month; introduce non-monetary prestige badges/leaderboards to encourage organic behavior.- Offer invitee-focused incentives (discount for both) to improve conversion fidelity.

4) Monitoring & metrics (Analytics & PM):- Immediate health metrics: viral coefficient, invites per active user, invite-to-activation rate, activation-to-7-day retention, fraud rate (flagged invites/confirmed fraud), reward payout ratio, CAC for referred vs organic.- Alerting: sudden deviation in invites-per-user, drop in invitee retention, spike in disposable domains.- Dashboards split by channel, cohort (new vs power users), geography, device.

5) Implementation roadmap & governance:- Week 0–2: deploy detection heuristics and monitoring dashboards; soft-watch mode.- Week 2–6: implement throttles, invite UX changes, progressive verification.- Week 6–12: migrate incentives to milestone-based payouts; A/B test reward structures.- Continuous: feedback loop between fraud team and data to refine heuristics; legal review for privacy; quarterly policy updates.

Trade-offs and reasoning:- Adding friction reduces some legitimate volume short-term but raises quality and lowers abuse costs long-term.- Deferred payouts align incentives to long-term value, reducing incentive for gaming.- Maintain transparent messaging to users to preserve trust.

Expected outcomes:- Lower fraud rate, improved invitee retention, reduced wasted payouts, and a healthier viral coefficient driven by higher-quality referrals.

EasyTechnical

59 practiced

Explain common retention metrics used in growth: cohort retention curves, DAU/MAU, rolling retention, churn rate, and survival curves. For each metric, state when it is most useful and one limitation a PM should be aware of when using it to make decisions.

Unlock Full Question Bank

Get access to hundreds of AARRR Growth Framework & Metrics interview questions and detailed answers.

Join thousands of developers preparing for their dream job.