agentleFS
lukesw55lukesw55/ai-pm-toolkit/.agents/skills/pm-archetype-growth/SKILL.md

pm-archetype-growth

Skill1 starsChanged 42 days agoMITView on GitHub

Growth PM archetype lens. Invoke when the focus is acquisition, activation, retention, expansion, monetisation, or funnel conversion — experimentation-heavy, analytical before prescriptive. Covers AARRR funnels, North Star → KPI tree cascades, hypothesis-driven A/B tests, primary metric + guardrail design, decision rules (ship/iterate/kill), winner's-curse traps, novelty effects, and segment interactions. Trigger on "AARRR", "funil", "experimento", "como melhorar a ativação?", "ship/iterate/kill", and any ask about funnel drop-offs, cohort retention, pricing tests, or A/B experiment design. For non-growth launch readouts (release notes, generic post-launch monitoring, A/B tests on features that aren't funnel-shaped), use pm-phase-deliver instead — this skill is the lens that brings growth-experimentation discipline.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Do not invoke for non-growth launch readouts — release notes, generic post-launch monitoring, A/B tests on features that aren't …

    line 43

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

PM Archetype — Growth products

Product-type lens. Pairs with phase skills when the work is funnel / experimentation / monetisation. The phase skills cover when and how to sequence; this skill covers the discipline of running many small bets where most will fail.

Shape: a compositional single-file lens by design — one SKILL.md that composes the phase and transversal references linked below and adds only the archetype-specific concerns, anti-patterns and output format; no references/ of its own until a real need appears; its evals/evals.json (standard, adversarial, negative control) is the quality contract and is graded like any full skill.

Prime directive

Learning rate > shipped count. In growth, most tested ideas fail. The job is to set up the test apparatus so the 10% that work compound, and the 90% that don't are killed quickly without leaving technical or product debt.

When to invoke

The work is about moving a funnel-stage metric or a monetisation outcome:

  • acquisition (signups, traffic sources, CAC)
  • activation (% signups hitting first-value within activation window; activation-funnel drop-offs)
  • retention (week-N / month-N cohort retention; habit-formation indicators)
  • expansion (upgrade rate, seat growth, feature adoption leading to plan change, NRR)
  • referral (viral coefficient, invitation conversion)
  • revenue (MRR growth, ARPU, LTV:CAC)

Also invoke when the team is about to run an A/B test and the metric / decision-rule design needs to be tightened.

Trigger phrases that route here include "acquisition", "activation", "retention", "expansion", "monetisation", "AARRR", "funil", "funnel", "conversion", "drop-off", "onboarding flow", "upgrade flow", "pricing test", "lifecycle", "habit formation", "cohort retention", "viral coefficient", "NRR", "ARPU", "LTV/CAC", "ship/iterate/kill", "experiment design", "A/B test", "experimento", "primary metric", "guardrail metric", "novelty effect", "winner's curse", "novel funnel hipótese", "como melhorar a ativação?".

Do not invoke for non-growth launch readouts — release notes, generic post-launch monitoring, A/B tests on features that aren't funnel-shaped. Those belong to pm-phase-deliver; this skill is the lens that brings growth-experimentation discipline.

Required reading before output

  • .ai/rules.md; for project work, read .ai/memory/active-context.md first when present and resolve <slug> before opening relevant project files. If no pointer exists, do not infer a project; initialize only when durable project context is needed and the project name is known.
  • relevant project memory — prior experiment logs are load-bearing; without them the team re-tests dead ideas
  • live PostHog data (via the PostHog MCP tools, e.g. query-run) when sizing the bet — never propose without baseline
  • shared org context in .ai/memory/org/ when present (personas.md, competitors.md, goals.md); open the file the task needs, not all four

References this skill chains to

  • ../pm-phase-define/references/kpi-tree.md — growth trees cascade from North Star into funnel stages
  • ../pm-phase-define/references/pricing-packaging.md — monetisation is often growth territory
  • ../pm-phase-develop/references/tracking-plan-design.md — growth lives or dies on instrumentation
  • ../pm-phase-deliver/references/experiment-interpretation.md — daily reference
  • ../pm-phase-deliver/references/metric-quality-guardrails.md — second-daily reference
  • ../pm-phase-deliver/references/product-analytics.md — funnels, cohorts, retention
  • ../pm-transversal-analysis/ — especially triangulation (why does the number move?)
  • data-science-analyst skill — for validating SQL, A/B maths, leakage checks

AARRR + guardrails

Pair every growth metric with at least one guardrail:

StagePrimaryGuardrail
Acquisitionsignups, CACquality of signup (activation rate downstream)
Activation% first-value within windowretention week-2 (activation theatre check)
Retentioncohort retention week-Nsupport load, NPS / sentiment
Expansionupgrade rate, NRRdowngrade / churn signal
Referralviral coefficientinvitation spam complaints
RevenueMRR, ARPUgross-margin, refund rate

Workflow

  1. State the AARRR layer — which stage of the funnel is this about?
  2. Read the current data — baseline + segment patterns. Pull from PostHog directly when possible.
  3. Form a hypothesis — if [change], [segment] will [behaviour], moving [metric] by [magnitude], because [mechanism].
  4. Design the minimum test — cheapest way to learn; A/B only when it fits (see ../pm-phase-deliver/references/experiment-interpretation.md).
  5. Pre-declare metrics + decision rule — primary, guardrails, ship / iterate / kill thresholds.
  6. Stress-test the metric setup — loop in your QA lead for tracking-plan QA + experiment validity. For statistical correctness, chain to data-science-analyst.
  7. Run + interpret — triangulate with quali if something surprises.
  8. Decide + document — ship / iterate / kill; log learning.
  9. Update memory — experiments.md, decisions.md, any triangulation memos.

Growth-specific anti-patterns

  • Local funnel optimisation that hurts retention. Pushing users through step 3 at the cost of step 5.
  • Over-testing low-value tweaks. 50 button-colour tests; 0 user-outcome tests.
  • Activation theatre. Moving activation rate by redefining "activation" instead of improving product.
  • "Everything is a conversion problem." Some product problems are not about conversion; they're about product-market fit.
  • Winner's curse. Stat-sig "wins" that don't replicate at scale.
  • Pricing changes without migration plans. New price, old customers, no communication → churn bomb.
  • Ignoring long-horizon outcomes. Shipping on week-1 metrics; week-N metric regresses.
  • Single-segment obsession. Helping free users; harming paid; net zero or worse.

Output format

## pm-archetype-growth recommendation

### AARRR layer
...

### Baseline + current state
metric, segment, trend (from PostHog or warehouse)

### Hypothesis
if [change], [segment] will [behaviour], moving [metric] by [magnitude], because [mechanism]

### Minimum test
method, sample, duration, decision rule

### Primary metric + guardrails
...

### Expected learning
what we'll know if green / yellow / red

### Risks to validity
confounds, novelty, interaction with other tests

### Follow-up experiments (if green)
...

### Memory updates
...

Integration

  • Upstream: pm-phase-define (KPI tree, cascade, North Star, pricing & packaging).
  • Build phase: pm-phase-develop (instrumentation — tracking plan + feature flag strategy + experiment scaffolding).
  • Launch phase: pm-phase-deliver (experiment interpretation, metric quality, post-launch monitoring, decision).
  • Transversals: pm-transversal-analysis (triangulating quali+quant when a number moves and you need to know why), pm-transversal-docs (experiment log structure in Confluence, Jira ticket hygiene around variants).
  • Engineering pairings: your QA lead (tracking-plan QA, experiment integrity), data-science-analyst (validate SQL / A/B maths), your engineering architecture partner when feature-flag plumbing crosses services.
  • Copilot mirror: .github/agents/pm-growth.agent.md.

Success criteria

  • funnel metrics move because of shipped work (not seasonality)
  • kill rate on experiments is high — bad ideas drop quickly
  • retention improves alongside activation (no local optimisation)
  • the experiment log teaches the team what works and what doesn't
  • instrumentation is trustworthy; no surprise metric drift

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.