Tools · Marketing
Growth Experiments
Describe your funnel — get an ICE-scored backlog of growth experiments.
How it works
Corpus-grounded (growth marketing / experimentation / AARRR via the marketing cluster). Produces 5–8 experiments — hypothesis, lever, metric, cheap test, and impact/confidence/effort + ICE priority each — plus a run sequence (quick wins first), guardrail metrics, and the riskiest assumptions.
You bring
{ context, cluster? }
You get
{ context_summary, experiments[]{hypothesis, lever, metric, test, effort, confidence, impact, ice_priority}, sequencing[], guardrail_metrics[], riskiest_assumptions[], grounded_in, provenance }
Use it for
- →Marketing-guide reader: a prioritized experiment backlog instead of random tactics
- →Rank ideas by ICE and sequence the quick wins first
- →Attach a metric + guardrails to each growth bet
See it work
example outputContext: a B2C language-learning app with strong signups but weak day-7 activation and high early churn.
Growth Experiments — LinguaLoop (language-learning app)
Context: Strong top-of-funnel signups, but day-7 activation is weak and early churn is high. Backlog scored by ICE (Impact · Confidence · Effort).
Experiment backlog
| # | Hypothesis | Lever | Metric | Cheap test | I / C / E | ICE |
|---|---|---|---|---|---|---|
| 1 | A 60-second first lesson before signup lifts activation | Activation | Day-1 lesson completion | Gate-free trial lesson on the landing page | H / H / L | High |
| 2 | A daily streak + reminder push raises day-7 return | Retention | Day-7 active | Add a streak counter + one push | H / M / L | High |
| 3 | A personalized goal in onboarding ("travel in 30 days") improves activation | Activation | Onboarding completion | 3-question goal picker | M / M / L | High |
| 4 | Social proof on the paywall lifts trial→paid | Revenue | Trial conversion | Add "2M learners" + testimonials | M / M / L | Medium |
| 5 | A referral ("teach a friend, both get a free week") drives signups | Referral | Invites sent | Post-lesson invite prompt | M / L / M | Medium |
| 6 | A day-3 inactivity win-back email reduces churn | Retention | Reactivation rate | One lifecycle email | M / M / L | Medium |
Run sequence (quick wins first)
1 → 2 → 3 (all low-effort, high-leverage on the activation/retention leak) → then 4, 6 → 5 last (referral only pays once retention holds).
Guardrail metrics
- Trial→paid conversion (don't trade activation for unqualified signups)
- Uninstall rate (push notifications can backfire)
- Average revenue per user
Riskiest assumptions
- That activation, not acquisition, is the binding constraint — validate against the funnel before building.
- That push opt-in rates are high enough for #2 to move the metric.
Grounded in: growth-experimentation / AARRR / ICE-prioritization (marketing cluster).
Run it now
Design growth experiments
Get an ICE-scored backlog of growth experiments — hypothesis, lever, metric, and test for each — plus a sequence to run them and guardrail metrics.
Prefer code? Call it over the API or hand it to your AI agent via MCP — POST /api/bicycle/growth-experiments · design_growth_experiments. API & agent access →