Francesca Monte | Marketing & Data Analyst

SENTRY.mark

Watchdog

Automated anomaly detection engine for marketing data — Z-Score statistical modeling on GA4 + BigQuery pipelines, with real-time Slack alerts and Kill-Switch logic.

GA4 events analyzed via BigQuery 61.7K

Anomalies detected (5.8% rate) 3,605

Rolling Z-Score window per channel 7 days

Real-time Slack alert + Kill-Switch Python

Anomaly Types

5

Bot attack · Tracking break

Detection Rate

100%

↑ Synthetic anomalies caught

Python: Causal Inference

IPW

Measured the true incremental lift of ad exposure using Inverse Probability Weighting — isolating causal effect from selection bias.

Propensity score via Logistic Regression IPW

Selection bias removed — SMD < 0.1 ✓ Balanced

Raw lift vs causal lift delta −0.88pp

True incremental ad lift +5.17%

BI: Churn Analysis

Power BI

3-page interactive dashboard analyzing telecom churn drivers for 7K customers — KPI overview, behavioral deep dive, financial impact.

Churn by Tenure & Monthly Charges

Monthly Charges

Active Churned

Tenure (months)

Churn Rate

26.54%

↑ 2K churned customers

Revenue at Risk

$139K

↓ Monthly exposure

Analytics: Funnel Audit

GA4

Full-funnel GA4 tracking framework to eliminate data gaps, fix attribution leakage and align revenue data.

Funnel Stage Tracking Accuracy — Before vs After

Before audit

After fix

Data Leakage Fixed

3

↑ Checkout stages

Attribution Accuracy

+41%

↑ Multi-channel

Amsterdam Airbnb Pipeline

ETL

End-to-end data pipeline on 10K listings — Python ETL, BigQuery cloud storage, Power BI dashboard. Uncovers Amsterdam's short-term rental market dynamics.

Raw CSVs → BigQuery (3 tables) 4.3M rows

Top revenue neighbourhood €60K/yr

Price vs review score above €500 No corr.

Review trend: Covid → full rebound 2020–25

SQL: HVC Segmentation

BigQuery

Identified High-Value Customers via Hero Category purchases — building a high-ROI audience for retention and lookalike campaigns.

CTEBigQuery

-- High-Value Customers via Hero Category purchases WITH user_profitability AS ( SELECT u.id AS user_id, u.email, u.traffic_source, p.category, oi.sale_price - p.cost AS net_profit FROM `thelook_ecommerce.order_items` AS oi JOIN `thelook_ecommerce.users` AS u ON oi.user_id = u.id JOIN `thelook_ecommerce.products` AS p ON oi.product_id = p.id WHERE oi.status = 'Complete' ) SELECT user_id, traffic_source, COUNT(category) AS high_margin_items, ROUND(SUM(net_profit), 2) AS total_profit FROM user_profitability WHERE category IN ('Blazers & Jackets', 'Suits', 'Outerwear') GROUP BY 1, 2 HAVING total_profit > 150 ORDER BY total_profit DESC LIMIT 10;

Tables Joined

3

↑ Unified pipeline

Audience Identified

HVC

↑ Profit > €150

Francesca
Monte

Strategic Data Solutions

Academic & Professional Journey

Technical Certifications

Ready to scale?