Live Funnel Analysis

This project was self built and features an interactive dashboard to simulate and visualise real time e-commerce user progress through a conversion funnel. I aimed to provide a viewpoint of customer behaviour across each stage to help identify repeating drop-off points and opportunities for optimisation.

The project utilises data simulations, an ETL pipeline, and offers analytical insights to solve the issues identified in the data.

Whilst working on a previous financial services site (iPayPDQ), it became evident that businesses face challenges in understanding and improving conversion rates. Without any kind of real-time visibility, it becomes challenging to:

• Detect sudden drop-offs in the funnel
• Better understand customer behaviour across pages
• Quickly improve the site to reduce abandonment
• Filter out noisy traffic that makes analyses challenging

Since I didn't have a live site to deploy the tool to; simulating a live funnel was an obstacle but was neccessary to reliably deploy a fully functioning solution.

The end-to-end solution involved:

• Data Simulation: Generating 30,000 unique users with session data over a simulated week
• Noise and Anomalies: With variable drop-off rates and a simulated checkout error causing a spike in abandonment (useful for testing)
• Data Processing (& ETL): Extracted raw data from a CSV, transformed in Pandas, and loaded into a DataFrame for interactive Streamlit visualisation
• Interactive Dashboard: Visualised a funnel chart showing user flow and drop-offs with time-range selectors, fully deployed to the cloud

Several findings where observed when analysing the dashboard: