This page is part of the ForgeSDLC knowledge base — an AI-assisted, human-directed methodology for taking product work from concept to production. For the core operating model and vocabulary, see Forge SDLC overview and What is ForgeSDLC?.

Growth Engineering & Funnel Optimization

Overview: Growth engineering treats the funnel as a system — stages, conversion rates, guardrails, and feedback loops — improved through prioritized experiments rather than one-off campaigns. The goal is repeatable learning: each change should be falsifiable, measured against pre-registered metrics, and safe for users and the brand.

AARRR (“pirate metrics”) framework

Stage	Definition	Key metrics (examples)	Example KPIs	Typical ownership
Acquisition	Users arrive via channels	CAC, installs, signups, traffic by source	Blended CAC, paid vs organic mix, channel LTV	Marketing, Growth
Activation	Users reach first meaningful value	Time-to-value, activation rate, onboarding completion	% completing “aha” event in session 1	Product, Growth
Retention	Users return or stay subscribed	D1/D7/D30, churn, cohort curves	Net revenue retention (NRR), WAU/MAU	Product, CS, Growth
Revenue	Monetization and expansion	ARPU, expansion, trial-to-paid	LTV, paywall conversion, upgrade rate	Product, Finance, Growth
Referral	Users bring others	K-factor (use carefully), invite rate, NPS → referral	Viral coefficient by cohort, referral % of signups	Growth, Product

AARRR funnel with conversion rates

flowchart TB ACQ[Acquisition visitors / signups] ACT[Activation aha completed] RET[Retention repeat use / renewal] REV[Revenue paid / expanded] REF[Referral invites / shares] ACQ -->|"c₁"| ACT ACT -->|"c₂"| RET RET -->|"c₃"| REV REV -->|"c₄"| REF REF -.->|loop| ACQ

Growth process cycle

flowchart LR ID[Ideate] --> PR[Prioritize ICE / RICE] PR --> EX[Experiment] EX --> ME[Measure] ME --> LE[Learn] LE --> ID

ICE / RICE: Score ideas on Impact, Confidence, Ease (ICE) or add Reach for RICE — use consistently so the backlog is comparable week to week.
Learning: Document hypothesis, design, result, decision — even null results reduce duplicate work.

Funnel instrumentation and guardrails

Layer	Examples
Stage definitions	Mutually exclusive rules (e.g. “activated” = completed event X within Y hours of signup)
Event quality	Schema versioning, deduplication, identity stitching for logged-in vs anonymous
Guardrail metrics	Refund rate, chargebacks, support tickets per thousand users, spam signups — must not regress while optimizing conversion
Qualitative triangulation	Session replay (privacy-safe), interviews, and support themes to explain why a metric moved

Without guardrails, teams often “win” experiments that damage trust, compliance, or operational load — then pay the cost in the next quarter.

Acquisition optimization

Lever	Notes
Channel diversification	Reduce single-channel dependency; compare cohort quality, not just CPA
CAC optimization	Improve creative, landing, and activation together — cheap clicks that do not activate raise blended CAC
Organic vs paid mix	Paid for learning velocity; organic for durability — align budget to PMF stage
Virality (k-factor)	Treat K as a sketch; model invites per active user, saturation, and incentive distortion

Activation optimization

Lever	Notes
Time-to-value	Remove steps between signup and the first successful outcome
“Aha moment”	Infer from retention correlates (events that split retained vs churned cohorts) — validate with experiments
Onboarding	Checklists, templates, guided setup, and smart defaults
Progressive disclosure	Surface depth after the first win — avoid walls of configuration up front

Activation sequence (conceptual)

sequenceDiagram participant U as User participant P as Product participant E as Events / analytics U->>P: Signup P->>U: Guided setup (minimal path) U->>P: Completes first value moment P->>E: Emit activation event E->>P: Trigger habit loop nudge (contextual) Note over P: Next session: reinforce loop

Retention optimization

Topic	Practice
Cohort analysis	Group by signup week or campaign; compare curves, not single snapshots
Retention curves	Flattening curves suggest PMF in a segment; steep drops flag onboarding or value gaps
Engagement loops	Notifications, in-product triggers, and content that tie to recurring jobs
Re-engagement	Push, email, and in-app win-back — respect frequency caps and consent
Habit (Hook model)	Trigger → action → variable reward → investment — ensure the “investment” stores future value (data, content, workflow)

Revenue optimization

Lever	Notes
Pricing experiments	Grandfather fairly; test packaging and presentation before destructive list-price wars
Upgrade triggers	Usage thresholds, feature gates, and success moments (not arbitrary paywalls)
Expansion	Seats, usage tiers, add-ons — align sales and PLG motions
LTV / CAC	Targets depend on payback period and capital efficiency; define guardrails (support load, refunds)
Paywall UX	Clarity of value, trial design, and payment friction materially affect conversion

Referral optimization

Element	Guidance
Loop design	Make inviting part of a natural workflow (collaboration, sharing output)
Incentives	Two-sided rewards reduce friction; watch fraud and low-quality referrals
Tracking	De-duplicate invites; attribute assisted vs direct referral paths
NPS	Promoters are a pool for referrals — pair surveys with concrete invite CTAs

A/B testing infrastructure

Topic	Guidance
Experiment design	Hypothesis, primary metric, guardrails, minimum detectable effect
Sample size / power	Pre-calculate before launch; avoid peeking without sequential rules
Statistical significance	Frequentist p-values or Bayesian probability — pick one approach per program
Sequential testing	Safe interim reads when volume is high
Multi-armed bandits	Good for short-lived optimization (headlines, creatives) with clear reward
Feature flags	Tie exposure to user/account IDs; support kill switches and gradual rollouts
SRM checks	Sample ratio mismatch invalidates many results — monitor allocation drift

Experimentation culture

Pillar	Behaviors
Velocity	Small batches, clear WIP limits on running experiments
Learning backlog	Ideas ranked; “done” includes write-up
Team shape	PM + engineer + designer + analyst (or shared analyst pool) for full-stack tests
Ethics	No dark patterns; informed consent where required; vulnerable users protected

Anti-patterns

Anti-pattern	Consequence
Premature growth	Scaling before PMF burns capital and damages reputation
Vanity metrics	Optimizing signups without activation or revenue
No statistical rigor	False wins and thrashing roadmap
Growth hacking without ethics	Regulatory risk, churn, and brand erosion

External references

Sean Ellis & Morgan Brown, Hacking Growth — growth process and case patterns
Reforge — Growth Series and advanced retention / monetization material (paid programs)
Alistair Croll & Benjamin Yoskovitz, Lean Analytics — metrics, stages, and focus for startups

Index: growth/README.md · Marketing map: ../MARKETING.md

Keep project-specific marketing plans in docs/product/marketing/ and GTM documents in docs/product/, not in this file.

Navigate