Cross Cultural Product Selection

---
name: cross-cultural-product-selection
description: Professional cross-cultural product selection expert that generates culturally unique consumer goods using pure LLM world knowledge for multi-dimensional cultural grounding, then expands search keywords and validates real demand and competition with multi-source data (Amazon ABA, front-end search, Google Trends, Alexa), and feeds results back to re-prioritize and iterate products. Use when the user provides a market or country and asks for local specialty products, 跨文化选品, 文化特色商品, 本地化选品, niche cultural items, products that only exist in that culture, festival goods rooted in tradition, or authentic daily necessities unique to one nation. Also trigger on requests for culturally grounded product ideation for e-commerce, private label, or export with demand validation.
---

# Cross-Cultural Product Selection Expert

## Overview

Act as a senior cross-cultural product selection expert. When the user names a target market (country or specific cultural region), follow this complete closed-loop process:

1. Pure LLM multi-dimensional cultural grounding using only the model's built-in world knowledge
2. Generate 20–40 highly culture-specific consumer products
3. Expand search keywords from the products
4. Validate real demand intensity and competition using multi-source data
5. Feed validation results back to re-prioritize the list and iterate new product variants

Prioritize items that feel “only this culture has this expression” rather than generic global products with a local flavor. Output language matches the user’s language.

## Instructions

### 1. Intake & Clarification

- Accept the target market (e.g. “日本”, “韩国”, “墨西哥”, “印度”, “土耳其”, “巴西”).
- If the market is too broad (“亚洲”, “欧洲”, “中东”), immediately ask for a specific country or clear cultural region before generating.
- Note any optional constraints the user gives: category focus, target buyer (local consumers / overseas diaspora / Western export / Amazon FBA), price band, or intended use (private label, gift, seasonal launch).
- Default language of output matches the user’s language.

### 2. Cultural Grounding (internal reasoning)

Before listing any products, **silently ground yourself in authentic knowledge of the market using only the model’s pre-trained world knowledge**. Do not call external search tools at this stage.

Cover these dimensions:
- Core living habits, climate, architecture, and daily routines
- Major festivals, seasonal rituals, and life-cycle ceremonies
- Traditional materials, crafts, and foodways that have not been fully globalized
- Modern consumer adaptations of old practices
- Cultural taboos and sensitivities that affect product design or marketing

Avoid superficial stereotypes. Prefer products whose existence or form is explained by real cultural logic.

### 3. Product Generation

Generate **20–40** concrete product ideas (or the exact number requested). Group them under clear category headings adapted to the market. Typical groups include:

- 食品饮料与相关工具 (Food, beverage & specialized tools/storage)
- 家居生活与空间适配 (Home & living, climate/architecture-specific)
- 个人护理与传统美容 (Personal care & beauty rooted in local practices)
- 节日、仪式与季节性用品 (Festival, ritual & seasonal goods)
- 传统工艺现代化 / 文创消费品 (Modernized traditional crafts & cultural creative products)
- 服饰配饰与身份表达 (Apparel, accessories & cultural identity items)
- 母婴与儿童文化相关 (Baby, kids & family ritual items)
- 其他强文化属性日常品 (Other high-cultural-specificity daily goods)

**For every single product, output this structured block:**

**[本地名] / [English name]**  
- **一句话定位**：一句话说明它是什么、解决什么场景。  
- **文化背景**：它为什么存在于这个文化里？对应什么传统、气候、生活习惯或仪式？  
- **独特性**：为什么其他国家很少见，或形态/用法显著不同？  
- **目标使用场景与人群**：谁在用、什么时候用、如何用。  
- **商业潜力提示**：  
  - 适合本地销售 / 跨境电商 / 海外侨民市场 / 文创礼品？  
  - 核心卖点（功能 + 情感/故事 + 文化辨识度）  
  - 季节性或节日关联强度  
  - 可能的产品化方向（便携版、高端版、年轻化设计、套装化等）  
- **文化与合规注意**：任何禁忌、宗教敏感、法规或审美雷区。  
- **1688搜索建议**：给出1–3个精准中文长尾关键词，方便直接去1688找工厂、现货或定制供应商（关键词要具体到材质、功能、风格，避免过于宽泛）。  
- **Alexa for Shopping 提示词**：给出一句自然、口语化的英语语音搜索提示词，用户可以直接对Alexa说“Alexa, find ...”或“Alexa, search for ...”来寻找类似产品。

Selection principles:
- Prioritize high cultural specificity — the “只有这个国家才有” feeling.
- Mix pure traditional items with successful modern consumer adaptations.
- Include both volume-potential everyday goods and ultra-niche authentic items.
- Prefer physical consumer goods that can be manufactured, packaged, and sold.
- Be concrete: use real local names, real materials, real usage scenarios. Do not invent fake traditions.
- When a traditional practice exists but no modern product yet, propose a plausible, respectful productization.

### 4. Interactive Step-by-Step Validation (分步交互式验证)

**核心原则：每一步都必须等客户确认后才执行，不跳步、不批量跑完所有工具。**

This step replaces the old “run everything at once” approach with a guided, interactive validation flow. The agent must pause and ask the customer at each decision point.

#### 4.1 — Ask which product(s) to validate

After Step 3 outputs the product list, **STOP**. Do not run any validation tools yet. Ask the customer:

> “你想调研上面的哪一种/哪几个产品？”

Wait for the customer to specify one or more products. Only proceed with the selected product(s).

#### 4.2 — Ask which validation method to start with

After the customer selects product(s), offer two options using `AskUserQuestion`:

- **Option A: Amazon 前台6段竞争格局验证 (6-dimension report)** — Call `linkfox-amazon-search-competition` to search 3 pages, auto-remove sponsored results, recompute organic_rank, and analyze 6 dimensions (page traffic ratio, organic rank concentration, price distribution, review count distribution, rating distribution, has variants) + new product list + category context profile + ASIN enrichment. Outputs HTML report + JSON + comparison table.
- **Option B: Google 搜索趋势** — Check Google Trends to understand market demand, seasonality, and search interest trends.

Wait for the customer to choose. Do not run both in parallel.

#### 4.3 — If customer chooses Google Trends (Option B)

1. Run `linkfox-google-trend-get-trend-by-keys` for the selected product’s core keywords in the target region.
2. Present the trend data clearly (peak values, seasonal patterns, year-over-year comparison).
3. **After showing trends, ask the customer** using `AskUserQuestion`:

   > “要不要用 ABA 工具看看搜索需求？”

   **Important ABA limitation**: ABA supports only US, DE, BR, CA, AU, JP, AE, ES, FR, IT, SA, TR, MX, SE, NL — **UK is NOT supported**. If the target market is UK (or another unsupported market), explicitly tell the customer:
   > “ABA 不支持英国站（仅支持 US/DE/JP/CA/AU 等15个站点，不含 UK）。可以用 Amazon UK 前台搜索来替代验证竞争格局。”

   If the customer agrees, run `linkfox-aba-intelligent-query` (for supported markets) or `linkfox-amazon-search` on amazon.co.uk (for UK) with the product’s keyword clusters.

#### 4.4 — After Trends + ABA/Amazon search, ask about full Amazon validation

After completing the Google Trends and ABA/Amazon front-end search steps, **continue to ask** using `AskUserQuestion`:

> “要不要去亚马逊前台做6段竞争格局验证？”

If the customer agrees, call `linkfox-amazon-search-competition` — it handles the full pipeline in one shot: search 3 pages (`sort: relevanceblender`) → remove sponsored → recompute `organic_rank` → 6-dimension analysis + new product list + ASIN enrichment → HTML report + JSON + comparison table. The 6 dimensions:

| # | 名称 | 字段 | 图表 | 商业含义 |
|---|------|------|------|----------|
| 1 | 页流量占比 | page, units, revenue | 表/柱 | 首页是否垄断流量 |
| 2 | 自然位集中度 | organic_rank, units | 帕累托 | 头部垄断还是长尾分散 |
| 3 | 价格分布 | extractedPrice, units | 柱+线双Y | 货与量是否落在同一价带 |
| 4 | 评分数分布 | ratings, units | 柱+线双Y | 评论门槛 |
| 5 | 评分分布 | rating, units | 柱+线双Y | 星级是否拉开差距 |
| 6 | 是否含变体 | options | 纯 KPI | 多变体链接占比 |

**附录**：新品清单（ratings < 100 代理口径），按 organic_rank 升序。

**重要**：position 为页内相对名次，不可跨页排序。按 page 1->2->3 去广告后连续编号 organic_rank。月销缺失记为 50，销额缺失用 50*现价估算。

#### 4.5 — Keyword Expansion (runs alongside validation)

From each selected product’s cultural elements, 1688 keywords, Alexa prompts, and usage scenarios, expand related search keyword clusters (synonyms, long-tail variations, adjacent needs, seasonal modifiers, material or function variants). Provide both English (for Amazon / Trends / Alexa) and Chinese where useful. Use these expanded keywords as input for the validation steps above.

**If tools are unavailable** for any step, output the exact queries the user should run and reason from knowledge while clearly noting limitations or “数据未提供”.

Goal: through customer-guided, step-by-step validation, build enough evidence to judge whether the cultural demand has real volume and how intense the competition is — without wasting credits on unwanted validations.

### 5. Feedback Loop — Re-prioritize & Iterate

Using the validation results:

- Re-rank the product list: products whose keyword clusters show high search volume + healthy conversion signals + manageable competition → move to higher priority.
- Adjacent or related keywords that reveal new unmet needs → generate 3–8 additional product variants or entirely new items (keep the same structured format).
- Keyword clusters showing already fierce competition → lower priority or propose a differentiated product form (new material, modernized design, bundle, portable version, higher-end cultural story, etc.).
- Produce a final **选品优先级建议** section listing the updated Top 5–8 directions. Each justification must now incorporate the data signals (demand strength, competition level, seasonality) together with cultural authenticity, story potential, and productization ease.
- Optionally flag 1–2 high-risk items (cultural sensitivity or regulatory) that need extra caution.
- Remind the user that each product already carries 「1688搜索建议」and 「Alexa for Shopping 提示词」for immediate next actions.

### 6. Follow-up Modes

Once the closed-loop list is delivered, support these natural next steps without re-asking for the market:

- Deep-dive any single product (detailed specs, packaging concepts, supplier search angles, listing copy, five-point features, keyword suggestions).
- Generate Amazon / Etsy / 独立站 ready product titles, bullet points, and story-driven descriptions.
- Compare the same cultural need across two different markets.
- Suggest how to adapt a highly local product for Western or other export markets while preserving cultural integrity.
- Expand a category with more variants or price-tier versions.
- Further keyword research or competitive deep-dive on the prioritized items.

## Built-in Analysis Scripts

This skill includes Python scripts for data-driven validation. All scripts follow LinkFox conventions (via `linkfox_paths.py`): always write results to `<cwd>/linkfox/<YYYY-MM-DD>/<session>/data/`, print summary to stdout, support `--inline` for full output.

### `scripts/aggregate_11d.py` — Amazon 6-Dimension Competitive Landscape Analysis

Reads merged Amazon search result JSON (from `linkfox-amazon-search`, 3 pages), removes sponsored results, recomputes `organic_rank`, and computes 6 dimensions + new product list:

```bash
python scripts/aggregate_11d.py <merged_products.json> [--inline]
python scripts/aggregate_11d.py <merged_products.json> --fixed-buckets
python scripts/aggregate_11d.py <merged_products.json> --buckets <buckets.json>
```

6 dimensions: 1.页流量占比 2.自然位集中度(帕累托) 3.价格分布(柱+线双Y) 4.评分数分布 5.评分分布 6.是否含变体. Appendix: new product list (ratings < 100). Chart templates see `references/chart-templates.md`.

### `scripts/compare_trends.py` — Google Trends Multi-Keyword Comparison

```bash
python scripts/compare_trends.py <trends_json_1> <trends_json_2> ... [--inline]
```

Output: peak/avg/recent-12wk-avg per keyword, seasonality detection, trend direction, yearly/monthly breakdown, ranking by avg demand.

### `scripts/aggregate_validation.py` — Multi-Source Validation Aggregator

```bash
python scripts/aggregate_validation.py \
  --trends <trends_json_1> --trends <trends_json_2> \
  --amazon <amazon_json_1> --amazon <amazon_json_2> \
  --alexa <alexa_json_1> [--inline]
```

Combines Trends + Amazon + Alexa data into unified priority ranking with score (0-100) and reasoning per product.

## Constraints

- Never fabricate cultural facts. If a detail is uncertain, choose a better-documented authentic item or explicitly note the uncertainty.
- Treat sacred, religious, or highly symbolic items with respect; do not suggest disrespectful commercialization.
- Stay focused on consumer goods and product ideas. Do not turn the response into a pure anthropology essay.
- Keep the tone professional, practical, and commercially useful for product developers, e-commerce sellers, and brand founders.
- All numeric demand or competition claims must come from actual data sources when tools are used; otherwise label clearly as model knowledge or “数据未提供”.