Engineering Growth Strategies
Given a product description and goal (e.g., "grow MusGo's weekly active users"), produce four deliverables in order: Growth Strategy → Acquisition Plan → Retention Plan → Experiment Matrix. Always ground recommendations in the AARRR funnel (Acquisition, Activation, Retention, Referral, Revenue) and prioritize experiments by ICE score (Impact, Confidence, Ease).
Example invocation output skeleton:
[North star metric, growth loops, funnel diagnosis]
[Channels, tactics, budget/effort allocation, CAC targets]
[Onboarding, engagement loops, churn triggers, win-back]
| # | Hypothesis | Funnel Stage | ICE Score | Owner | Timeline |
Progress:
- Clarify North Star Metric (NSM) and current funnel stage that's weakest
- Diagnose growth bottleneck (acquisition, activation, retention, or virality)
- Draft Growth Strategy (loops, positioning, NSM, key bets)
- Draft Acquisition Plan (channels ranked by CAC/LTV fit)
- Draft Retention Plan (onboarding, habit loops, churn prevention)
- Build Experiment Matrix with ICE-scored, testable hypotheses
- Sanity-check: every recommendation ties back to a metric and a testable hypothesis
Step details:
-
Diagnose first. Never jump to tactics before identifying which stage of the funnel is leaking users most. Ask for or infer: signup rate, activation rate, D7/D30 retention, referral coefficient (k-factor).
-
Growth Strategy must define:
- North Star Metric (one metric that reflects value delivered)
- 1-2 core growth loops (not just funnels) — e.g., content loop, referral loop, paid loop
- The single biggest bottleneck to attack first
-
Acquisition Plan must rank channels by expected CAC, using a simple table (Channel, Tactic, Target Audience, Est. CAC, Est. Volume, Priority).
-
Retention Plan must separate:
- Activation (first-session "aha moment")
- Habit formation (trigger → action → reward → investment, Hook Model)
- Churn intervention (win-back triggers, lifecycle emails/push)
-
Experiment Matrix must have 8-15 experiments, each with a falsifiable hypothesis ("If we do X, metric Y will increase by Z%"), scored ICE (1-10 each), sorted by total score descending.
Example 1: Input: "MusGo is a ride-hailing app in Indonesia losing users after first ride. Design a growth strategy."
Output:
NSM: Weekly Active Riders (WAR) Bottleneck: Activation→Retention drop-off after ride #1 (D7 retention 12%, industry benchmark 25%+) Core Loop: Referral loop (rider invites rider for ride discount) + habit loop (daily commute reminders)
| Channel | Tactic | Audience | Est. CAC | Priority |
|---|---|---|---|---|
| Referral | In-app double-sided discount | Existing riders | Rp15k | High |
| Local partnerships | Campus/office kiosks | Students/office workers | Rp25k | Medium |
| Paid social | Lookalike audience from top 10% riders | Urban 18-35 | Rp40k | Low (test only) |
- Activation: Guarantee first-ride ETA <5min or refund, to nail "aha moment"
- Habit: Push notification at user's historical commute time
- Churn intervention: Auto-send 20% off voucher if no ride booked in 10 days
| # | Hypothesis | Stage | ICE | Timeline |
|---|---|---|---|---|
| 1 | Commute-time push increases D7 retention by 15% | Retention | 8/7/9=24 | 1 week |
| 2 | Double-sided referral bonus increases k-factor by 0.1 | Virality | 7/6/8=21 | 2 weeks |
- Always propose a North Star Metric before any tactic — tactics without a metric are noise.
- Prefer growth loops over funnels when virality/referral is viable; loops compound, funnels don't.
- Every retention tactic should map to the Hook Model (Trigger → Action → Reward → Investment).
- Size experiments for speed: prioritize 1-2 week tests before big-bet 3-month plays.
- Use local context (payment methods, channels, culture) when the product is region-specific (e.g., Indonesia: WhatsApp, GoPay/OVO, local influencers).
- Don't recommend acquisition tactics before diagnosing the real bottleneck — scaling top-of-funnel when retention is broken wastes CAC.
- Don't give vague hypotheses like "improve onboarding" — always quantify expected impact and metric.
- Don't skip the Experiment Matrix's ICE scoring — unprioritized lists aren't actionable.
- Don't conflate acquisition channels with growth loops — paid ads are linear spend, loops are compounding and should be emphasized when possible.