Execute qualquer Skill no Manus
com um clique

Execute qualquer Skill no Manus com um clique

Começar

conversion-optimization

Estrelas0

Forks0

Atualizado18 de fevereiro de 2026 às 08:20

Optimize conversion with evidence-tiered layout and copy experiments.

Instalação

Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.

Executar no Manus

Fonte

abpaul

abpaul/codex-compound-engineering

Abrir repositório GitHub Ver repositórios do creator

Download

Executar no Manus

Ocupações relacionadasSOC

Baseado na classificação ocupacional SOC

Analistas de pesquisa de mercado e especialistas em marketingNegócios e Operações Financeiras·SOC 13-1161

Explorador de arquivos

5 arquivos

SKILL.md

readonly

name	conversion-optimization
description	Optimize conversion with evidence-tiered layout and copy experiments.

Conversion Optimization (Layout + Copy)

Use this skill when the goal is to move a measurable conversion metric (signup, checkout, purchase, lead capture, donation) via layout/IA and/or messaging changes.

Required Inputs

surface + page type (landing, PDP, pricing, onboarding, email, in-app prompt)
funnel stage + primary conversion event
primary KPI + guardrails (bounce, complaints, unsubscribe, refund, churn, lead quality)
constraints (brand/legal/analytics/engineering) + traffic/device split
baseline metric (or explicitly labeled assumptions)

Evidence Tiers

Tier 1: experiments/benchmarks/peer-reviewed behavioral evidence
Tier 2: measured practitioner case studies or mixed evidence
Tier 3: heuristics only (low-confidence)

Workflow

Choose lane(s): layout, copy, or both.
Pull 5–8 candidate levers/principles:
- layout: references/layout-evidence-matrix.md
- copy: references/copywriting-evidence-matrix.md
Convert each into one hypothesis: if -> then -> because, with expected KPI direction + likely moderators.
Draft variants:
- layout: concrete sections/components to change + rationale
- copy: 3–5 variants (control-safe, clarity-first, motivation-first, optional blended)
Run ethics/truth gate:
- no fabricated social proof or scarcity
- no deceptive defaults/dark patterns
- no regulated claims beyond provided constraints
Build experiment spec:
- randomization unit + traffic split
- KPI definitions + guardrails
- runtime/sample sizing assumption + stop rule (no peeking stop)
Return prioritized recommendations with citations + confidence.

Output Contract

Prioritized Hypotheses (P1/P2/P3) with tier + citation(s)
Implementation Plan (layout diffs and/or copy variant table)
Experiment Spec (KPI/guardrails/split/stop rule)
Conflicts + Unknowns (null/backfire conditions)

Assets and References

examples: assets/examples.md
new-study extraction template (optional): assets/extraction-template.md

Mais deste repositório

mesmo repositório

agent-native-suite

abpaul/codex-compound-engineering

Design, audit, and review agent-native systems to ensure action parity and robust agent workflows.

2026-02-180

cloud-transfer

abpaul/codex-compound-engineering

Upload and sync files to cloud providers using reproducible rclone-based workflows.

2026-02-180

design-fidelity

abpaul/codex-compound-engineering

Drive implementation fidelity against design intent through compare-analyze-iterate loops.

2026-02-180

docs-publishing

abpaul/codex-compound-engineering

Produce and publish project-facing documentation artifacts (changelogs, walkthroughs, release docs).

2026-02-180

frontend-design

abpaul/codex-compound-engineering

Create distinctive tokenized UI for Rails server-rendered apps using Phlex, Stimulus, and Turbo/Hotwire Native.

2026-02-180

frontend-ship

abpaul/codex-compound-engineering

Orchestrate frontend design, implementation, review, and browser evidence with conditional depth.

2026-02-180

Conversion Optimization (Layout + Copy)

Use this skill when the goal is to move a measurable conversion metric (signup, checkout, purchase, lead capture, donation) via layout/IA and/or messaging changes.

Required Inputs

surface + page type (landing, PDP, pricing, onboarding, email, in-app prompt)

funnel stage + primary conversion event

primary KPI + guardrails (bounce, complaints, unsubscribe, refund, churn, lead quality)

constraints (brand/legal/analytics/engineering) + traffic/device split

baseline metric (or explicitly labeled assumptions)

Evidence Tiers

Tier 1: experiments/benchmarks/peer-reviewed behavioral evidence

Tier 2: measured practitioner case studies or mixed evidence

Tier 3: heuristics only (low-confidence)

Workflow

Choose lane(s): layout, copy, or both.

Pull 5–8 candidate levers/principles:

layout: references/layout-evidence-matrix.md
copy: references/copywriting-evidence-matrix.md

Convert each into one hypothesis: if -> then -> because, with expected KPI direction + likely moderators.

Draft variants:

layout: concrete sections/components to change + rationale
copy: 3–5 variants (control-safe, clarity-first, motivation-first, optional blended)

Run ethics/truth gate:

no fabricated social proof or scarcity
no deceptive defaults/dark patterns
no regulated claims beyond provided constraints

Build experiment spec:

randomization unit + traffic split
KPI definitions + guardrails
runtime/sample sizing assumption + stop rule (no peeking stop)

Return prioritized recommendations with citations + confidence.

Output Contract

Prioritized Hypotheses (P1/P2/P3) with tier + citation(s)

Implementation Plan (layout diffs and/or copy variant table)

Experiment Spec (KPI/guardrails/split/stop rule)

Conflicts + Unknowns (null/backfire conditions)

Assets and References

examples: assets/examples.md

new-study extraction template (optional): assets/extraction-template.md