Cross-Border Product Growth

Goal

Help a Chinese small seller decide where a product should go internationally and produce a practical launch package.

This skill is designed for ecommerce AI tools that already have product notes, rough photos, or merchant descriptions. It does not promise automatic sales or actual marketplace publishing. It produces decision support, localized copy, creative briefs, and validation tasks that the seller can execute or pass to an agent.

Default to Chinese output for Chinese sellers unless the user asks for English. Keep platform-facing listing copy in the target market language, but explain decisions, risks, and next steps in Chinese.

• target market and platform recommendation
• quantified readiness score and evidence gaps
• competitor teardown and positioning wedge
• localized buyer persona and use scenarios
• compliant product claims and forbidden-claim warnings
• listing title, bullets, description, search terms, ad hooks
• image and short-video creative briefs
• market-data checklist and manual research steps
• 3-day validation plan with completion standards
• review loop that improves the plan after feedback
• follow-up iteration plan when the user is not satisfied

中文新手入口

当用户不知道怎么用时，先让用户按这个最小模板输入，不要要求一次填全：

商品是什么：
主要卖点：
价格/成本：
现在在哪个平台卖：
想去哪个海外平台：
有没有竞品：
有什么不能乱说的点：

如果用户只给一句话，也继续工作，但必须标出“我基于哪些假设生成”和“还缺哪些信息会影响质量”。

Why This Skill Is Different

• Market-first, not translation-first: choose a first launch wedge before writing copy.
• Evidence-scored: distinguish facts, assumptions, risky claims, and tests.
• Competitor-aware: convert competitor pressure into a narrower wedge instead of generic advice.
• Compliance-aware: separate safe claims from claims that need evidence.
• Creative-ready: create image and short-video briefs that can be sent to image/video models when available.
• Validation-driven: give scripts, sample sizes, pass standards, stop rules, and pivot rules.
• Iterative: use the previous output plus user feedback to produce a sharper version.

Use This Workflow

1. Clarify product facts
• Separate observed facts, seller-provided claims, and assumptions.
• Never invent certification, medical effects, material, capacity, origin, safety claims, or platform eligibility.
• If facts are missing, mark them as needs_verification.
• If critical information is missing, ask at most 3 targeted questions only when the missing data blocks a safe recommendation.
• If the user wants speed, proceed with assumptions and put them in Missing Info & Assumptions.

1. Score market fit
• Score 3 to 5 likely markets/platforms.
• Consider buyer demand, logistics difficulty, regulation risk, visual appeal, price room, competition intensity, and content fit.
• Prefer a narrow first wedge over a broad global plan.
• Use references/market-heuristics.md when platform selection is uncertain.
• Use references/market-data-checklist.md to tell the user exactly which public signals to collect when real market data is missing.

1. Choose one launch wedge
• Select the highest-probability market/platform combination.
• Explain why in plain language.
• Name the riskiest assumption that must be tested first.

1. Build an evidence ledger
• Mark each important claim as verified, seller_claim, inferred, or needs_verification.
• Convert unverified claims into safer wording or test tasks.
• For electronics, children, food, beauty, health, pet, safety, and regulated categories, make risk visible early.

1. Analyze competitors
• If competitors are provided, compare them directly.
• If not, infer the likely competitor set and label it as an assumption.
• Output 3 competitor angles: price/value, trust/proof, and creative style.
• Identify what not to copy and where the seller can be meaningfully different.
• When competitor URLs are not available, output a 15-minute manual search plan instead of pretending to have live data.

1. Localize content
• Rewrite benefits for local buyers, not literal Chinese translation.
• Replace Chinese platform idioms with local shopping language.
• Keep claims conservative and evidence-based.
• Produce variants for organic listing, paid ad, and social post.

1. Create visual briefs
• Give image/video generation prompts that preserve product identity.
• Specify aspect ratio, scene, buyer context, text overlay, props, and forbidden elements.
• For products with safety, health, children, cosmetics, food, or electronics risk, add compliance cautions.

1. Plan validation
• Give tasks that can be done today.
• Include exact outreach scripts, sample size, target signal, and pass/fail threshold.
• If the result is weak, propose a pivot: market, persona, price, bundle, or creative angle.
• Include one "how to actually do it" line for each task, so a beginner can execute without guessing.

1. Iterate until satisfactory
• When the user says the result is not good enough, ask what is wrong only if the feedback is unclear.
• Otherwise regenerate using the previous output plus feedback.
• Make the next version more specific, not merely longer.
• Preserve useful validated facts and only change the weak parts.

Input Contract

Use any available fields. If the user gives only a rough product note, continue with explicit assumptions.

product_name:
category:
target_customer:
rough_notes:
visible_photo_facts:
seller_claims:
price_or_cost:
available_assets:
preferred_platforms:
constraints:
competitors:
evidence_available:
market_data:
seller_stage:

Output Shape

Return concise Markdown with these sections. Do not claim that images, videos, ads, listings, or orders were actually created unless a tool has really done so.

## Market Choice
- Best wedge:
- Why:
- Main risk:

## Missing Info & Assumptions
- Known facts:
- Assumptions used:
- Ask next if available:

## Readiness Score
| Dimension | Score | Evidence | What would raise the score |

## Market Scores
| Market / Platform | Score | Why | Risk |

## Market Data Checklist
| Signal to collect | Where to check | How to use it | If missing |

## Evidence Ledger
| Claim | Status | Safe wording | Verification needed |

## Competitor Teardown
| Competitor / Substitute | Their angle | Weak spot | Our wedge |

## Localized Positioning
- Buyer:
- Core pain:
- Promise:
- Proof needed:

## Listing Copy
- Title:
- Bullets:
- Description:
- Search terms:

## Creative Briefs
### Main Image
### Benefit Image
### Short Video

## Outreach Scripts
- Buyer DM:
- Seller interview:
- Creator brief:

## Compliance Watchlist
- Safe to say:
- Do not say unless verified:

## 3-Day Validation Plan
| Day | Task | Script / Output | Pass Standard |

## Beginner Execution Steps
1.
2.
3.

## Decision Rules
- Continue if:
- Revise if:
- Stop or pivot if:

## Next Iteration
- If users dislike price:
- If users dislike visual style:
- If users do not understand the product:

Strong Defaults

• For first-time users, output a shorter "beginner mode" result first: market choice, main risk, 3 immediate actions, then detailed sections.
• For broad consumer products, compare: United States Amazon, United States TikTok Shop, Etsy, Shopee Singapore/Malaysia, Lazada Southeast Asia, Shopify DTC.
• For handmade/design products, include Etsy.
• For cheap impulse products, include TikTok Shop and Temu-style positioning but warn about margin pressure.
• For beauty, supplements, food, children, electronics, medical, pet health, and safety products, raise compliance risk.
• For products with obvious short-video demonstration value, make TikTok Shop or creator-led validation visible even if Amazon is the eventual sales channel.
• For products that depend on trust, certification, installation, sizing, warranty, or after-sales support, score Amazon/Shopify lower until proof is provided.
• For unclear product photos, focus on discovery questions and conservative copy.

Quality Bar

The answer is not good enough if it only translates text. It must reveal:

• which market to test first
• what the buyer cares about there
• what claims are risky
• how it differs from competitors
• what evidence is missing
• what market data should be collected next
• what images/videos should be generated
• what the seller should do today to validate demand
• how a beginner should execute each step

Read supporting files only when useful:

• references/market-heuristics.md: platform selection heuristics.
• references/market-data-checklist.md: real market-data collection checklist and fallback rules.
• references/beginner-playbook.md: step-by-step usage manual for first-time users.
• references/validation-scorecard.md: scoring dimensions and pass/fail thresholds.
• references/use-cases.md: concrete usage scenarios for new users.
• references/faq-anti-pitfall.md: FAQ, wrong-use examples, and anti-pitfall guidance.
• references/example-output.md: sample output style.
• scripts/generate_brief.mjs: quick local draft for rough product notes.

Do Not Overclaim

Do not say the skill can:

• directly publish to Amazon/TikTok Shop/Etsy
• guarantee sales or conversion lift
• generate final images/videos without an external image/video model
• verify certifications, customs rules, taxes, or legal compliance
• scrape private platform data

When those are needed, output a checklist or handoff instruction instead.