S6. External Signal Fusion (AMWAY)
Published 2026-05-14Updated 2026-06-301 min read
1. URL & Personas
/signals· P2, P3
2. Four External Signals
| Type | Sources |
|---|---|
| Social | Global Instagram · Reddit · X · Korea Naver · China Weibo · Xiaohongshu |
| Weather | KMA + open-meteo (global) |
| Economic | Bank of Korea · Statistics Korea + Open Exchange Rates (multi-country FX) |
| Regulatory | FTC · Korea Door-to-Door Sales Act · MFDS · EU |
3. Processing Pipeline
- Aggregate in-house GMV per day and country
- Join time series of the 4 external signals
- Correlation and lag analysis
- Time series + correlation heatmap + regulatory annotations
4. Output UI
- Time series with dual axes (external + in-house)
- Category × signal × country correlation heatmap
- Lag analysis (lag 0–14 days)
- Annotations on regulatory effective dates
5. Guardrails
- Cite sources (public data only)
- Prohibit causal inference
6. Demo Scenarios
- FX +5% vs. non-US in-house sales (14-day lag)
- "Vegan beauty" global keyword vs. Artistry vegan revenue
- Eight weeks of in-house direct sales after the FTC regulation took effect