Completion Rate Decline Analysis: Investigating Viewership Drop-Off on weather.com

This project examines patterns behind declining video completion rates on weather.com, aiming to identify potential platform-specific, content-specific, or behavioral factors contributing to audience drop-off.

---

📌 Purpose

The project has three primary objectives:

• Explore Completion Rate Trends: Understand current state of completion rate across platforms and forms of content
• Draw User Insights: Find strongest correlating variables that have effected completion rates in this target period
• Support Editorial and Platform Strategy: Provide data-driven insight to optimize video content performance across platforms.

---

🔬 Research Foundation

This analysis builds upon platform-sourced behavioral metrics from Amplitude, including completion rates, video views, and watched ads, segmented by platform (Desktop, iOS, Android, etc.). It aims to pinpoint underperforming conditions or patterns through time-based and categorical trends.

---

🧪 Methodology

• Amplitude Data Processing: Imported multiple Amplitude exports, merged them by video title or UUID, and computed platform-specific completion rates
• CMS Metadata Integration: Used regex to extract uuid values from CMS URLs, merged CMS metadata with Amplitude metrics, parsed, and reformatted publish dates.
• Collection Extraction: Parsed the content type (e.g., “science”, “news”, “forecast”) from the structure of the Live Link URL using regex and mapped it into a new Collection column.
• Monthly Platform Averages: Grouped videos by month and platform to compute average platform-level completion rates. These were used to normalize performance.
• Content Score Calculation: Created a Content Score metric for each video by summing its completion rates across platforms, each weighted relative to that platform’s monthly average.
• Daypart Alignment: Compared the hour each video was published (Publish Hour) to the average hour of peak performance for its content type to assess whether timing aligned with viewer behavior.
• Visualization & Debugging: Created histograms, bar plots, and scatter plots to visualize trends in drop-off rate by platform, hour, and content type. Also displayed outliers and edge cases for editorial review.

---

📚 Report

• Link to report can be found here

---

📂 Repository Structure

• Completion Rate Decline Analysis.ipynb: Contains the full analysis in Jupyter format.
• requirements.txt: Requirements file to download all imbedded libraries and tools.

---

🔧 Requirements

To run this project:

pip install -r requirements.txt