Explainable Brand Decline Intelligence System 📉

Gen AI Hackathon Submission

Problem Statement

Most social media analytics tools tell you what happened (e.g., "engagement dropped by 10%"). They fail to predict when and why a specific brand campaign will collapse.

This project solves the problem of Brand-Specific Campaign Decline Prediction, helping marketing teams intervene before a campaign loses momentum.

Why It Matters

• Proactive Intervention: Detect decline signals before the collapse happens.
• Cost Efficiency: Stop spending ad budget on dying campaigns.
• Brand Protection: Identify negative sentiment shifts early to prevent PR crises.

---

System Architecture

The solution is built on a Unified Intelligence Architecture:

1. Frontend (Streamlit): Business-facing dashboard with KPI cards and interactive charts.
2. Backend (FastAPI): High-performance API serving the intelligence engine.
3. Unified Intelligence Agent: A single, rule-based AI agent that:
   • Analyzes multi-dimensional signals (Engagement, Sentiment, Saturation).
   • Predicts decline probability and lifecycle stage.
   • Generates human-readable explanations.

Explainability First 🧠

Unlike "black box" ML models, our system provides clear, business-focused reasoning for every prediction:

"Engagement is dropping rapidly due to content saturation, indicating audience fatigue."

---

How to Run Locally 🚀

Prerequisites: Python 3.9+

1. Backend Setup

   pip install -r requirements.txt
   uvicorn backend.main:app --host 0.0.0.0 --port 8000

2. Frontend Dashboard (New Terminal)

   python -m streamlit run frontend/main.py

3. Access

   • Dashboard: http://localhost:8501
   • API Docs: http://localhost:8000/docs