S6. External Signal Fusion (4 Types)

📅Published 2026-05-14·🔄Updated 2026-06-30·⏱2 min read

Use all 4 external data types — one of the core values of the marketing innovation PoC.

1. URL Path

• /signals

2. User Stories

P2 (Insights) — Wants to see, on one screen, the correlation between the external environment (trends · weather · FX · competitors) and first-party category sales.

P3 (D&A) — Statistical validation of category × external variable correlation coefficients.

3. Input UI

• 4 external signal toggles: social · weather · economy · competitor
• First-party data: BU/Category/Segment/Channel
• Period slider

4. Data Mix

External Signal	First-Party Mapping	Use
Social trends (Naver · Google · X · Instagram · Olive Young)	Category · keyword mapping	Time series + SOV
Weather (KMA · Air quality)	Region · date key	Scatter + correlation
Economy (FX · CPI · Employment)	Month · first-party GMV	Time series + lag
Competitor (Amorepacific · Aimo · Yuhan-Kimberly)	Category mapping + event dates	annotation + comparison

5. Processing Pipeline

1. Daily (region, category) first-party GMV aggregation
2. Load 4-type external signal time series
3. Join by date/region/category key
4. Correlation · lag analysis (cross-correlation)
5. Visualization: time series + scatter + correlation heatmap + competitor annotation
6. LLM comments (per persona)

6. Output UI

• Dual-axis time series (external + first-party)
• Category × external signal correlation heatmap
• Lag analysis line (lag 0~14 days)
• Competitor event annotation (launch · campaign · controversy)

7. Guardrails

• Mandatory source attribution (social · weather · economy · competitor — all)
• No causal inference (express "correlation" only)
• Competitor information only from public sources

8. Demo Scenarios

1. Temperature -5°C vs. hot pack · hand cream → R² 0.78
2. Instagram "vegan beauty" keyword vs. first-party vegan SKU sales
3. FX vs. imported category GMV (7-day lag)
4. Change in first-party category sales 4 weeks after competitor new product launch