📊 OmniTrack | Enterprise Marketing Analytics & Multi-Touch Attribution Engine

An end-to-end data science and analytics platform designed to solve the digital marketing measurement crisis: cross-channel attribution distortion, walled-garden double-counting, privacy signal loss, and unmonitored capital misallocation.

---

🌟 Key Capabilities

1. Multi-Touch Attribution (MTA) Engine:

   • Evaluates conversion and revenue distribution across First-Touch (FTA), Last-Touch (LTA), Linear (MTA), and Position-Based (40-20-40) models.
   • Eliminates last-click bias that structurally penalizes top-of-funnel discovery channels (Display & Social) while artificially inflating lower-funnel channels (Paid Search & Email).

2. Automated Anomaly & Spend Bleed Alerting:

   • Real-time rule-based & statistical Z-score anomaly detection engine.
   • Detects sub-target ROAS (< 2.2x), high CAC, and negative ROI campaigns with automated severity classification (SEV-1 Critical, SEV-2 Warning, Opportunity).

3. Interactive Executive Web Dashboard:

   • Executive Overview: Real-time KPI scorecards, cumulative media spend vs. revenue pacing curve, and channel portfolio allocation.
   • Campaign Matrix & Alerts: 4-quadrant efficiency matrix (Spend vs. ROAS) and incident-style triage feed.
   • Attribution & Journey Waterfall: Cross-model lift analysis and common customer journey pathways.
   • What-If Budget Reallocation Simulator: Live sliders to model cutting the $144.3k competitor search bleed and reallocating funds into high-performing channels.
   • Dual Themes: Executive Dark Mode & Boardroom Light Mode toggle.

4. Production Data Warehouse SQL Pipeline:

   • Fully optimized ANSI SQL queries utilizing analytical window functions (COUNT() OVER(), ROW_NUMBER(), SAFE_DIVIDE) compatible with Google BigQuery, Snowflake, and Databricks.

---

🏗️ Architecture & Data Flow

flowchart LR
    subgraph Ingestion["1. Multi-Channel Touchpoints"]
        A1["Paid Search"]
        A2["Paid Social"]
        A3["Display Programmatic"]
        A4["Email & Retention"]
        A5["Organic Search"]
    end

    subgraph Engine["2. Attribution & Analytics Engine"]
        B1["Customer Journey Sequences"]
        B2["MTA Weight Calculations<br/>(First, Last, Linear, U-Shaped)"]
        B3["Anomaly & Z-Score Detection"]
    end

    subgraph DataOutputs["3. Analytical Datasets"]
        C1["campaign_summary.csv"]
        C2["mta_attribution_summary.csv"]
        C3["campaign_alerts.csv"]
        C4["customer_journeys.csv"]
    end

    subgraph Presentation["4. Executive Web Application"]
        D1["OmniTrack Portal (index.html)"]
        D2["Power BI / Tableau Integration"]
    end

    Ingestion --> Engine
    Engine --> DataOutputs
    DataOutputs --> Presentation

Loading

---

📈 Empirical Analysis & Key Findings

Marketing Channel	Total Spend	Last-Touch Rev	Linear MTA Rev	Attribution Variance	Strategic Diagnostic
Display Programmatic	$36,875.66	$3,665.96	$6,707.47	+83.0% Lift 🚀	Originated 87 first-touch journeys vs. 39 last clicks. Last-click misses 45% of Display's true pipeline value.
Social Media (Meta)	$72,566.04	$7,647.80	$8,622.19	+12.7% Lift 📈	Prime upper-funnel discovery engine (99 first-touch conversions) driving mid-funnel consideration.
Organic Search	$0.00	$4,200.90	$5,236.19	+24.6% Lift	High-intent organic baseline that paid search brand bidding frequently cannibalizes.
Paid Search (Google)	$180,873.09	$12,958.62	$9,907.80	-23.5% Over-credit 📉	Over-indexed on last-touch (106 last vs 75 first). Captures demand initiated by Display and Social.
Email & Retention	$2,691.58	$6,861.82	$4,861.45	-29.2% Closer	The consummate transaction closer (64 last-touch vs 16 first-touch). Delivers 62.95x ROAS.

---

📁 Repository Structure

├── index.html                   # Interactive Executive Web Application & Dashboard
├── attribution_engine.py         # Python Analytics, MTA, and Anomaly Alert Engine
├── multi_touch_attribution.sql  # Production BigQuery/Snowflake SQL Engine
├── campaign_summary.csv         # Aggregated campaign metrics & costs
├── mta_attribution_summary.csv  # Cross-model attribution comparison dataset
├── campaign_alerts.csv          # Flagged campaign anomalies & prescriptive actions
├── customer_journeys.csv        # Simulated multi-touch customer touchpoint sequences
├── requirements.txt             # Python dependencies (pandas, numpy)
├── .gitignore                   # Standard ignore rules
└── README.md                    # Comprehensive documentation

---

🚀 Quickstart Guide

1. Run the Python Analytics Engine

# Clone the repository
git clone https://github.com/<your-username>/marketing-analytics-attribution.git
cd marketing-analytics-attribution

# Setup virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run engine
python attribution_engine.py

2. Launch the Interactive Web Dashboard

Simply open index.html in any modern web browser, or serve it locally:

python -m http.server 8085
# Visit http://localhost:8085

---

🌐 Deploy to GitHub Pages

1. Push this repository to GitHub.
2. Go to Settings → Pages.
3. Under Build and deployment, select Deploy from a branch.
4. Set branch to main and folder to /(root).
5. Click Save. Your dashboard will be live at https://<your-username>.github.io/<repo-name>/.

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.