Will Boyle

YouTube Shorts Analytics Pipeline & Dashboard

GitHub – willboyledev/ytshortsanalytics

Tech stack:
Python · YouTube Data API · YouTube Analytics API · Google Cloud Run · BigQuery · Looker Studio · OAuth 2.0

Project Overview

This project was built to solve a real faced by my brother, who runs the Discovery Dose YouTube Channel (700k+ subscribers, at the time of writing). The problem is that YouTube Shorts performance data is difficult to analyze at scale. While YouTube Studio provides useful metrics, deeper analysis requires him to manually export CSV files, apply formatting, and repeatedly recreate the same views. This makes it hard to spot trends across large volumes of content.

I built an automated analytics pipeline and interactive dashboard specifically focused on analyzing Views, Average Percentage Viewed (APV), and Average View Duration (AVD) across all Shorts in a single dashboard. The goal was not to replace YouTube Studio, but to create a complementary tool designed for comparative analysis, trend discovery, and data-driven decision making.

Purpose & Design Philosophy

YouTube Studio excels at inspecting individual videos and monitoring recent performance. However, it is not optimized for answering questions such as:

  • Which Shorts consistently retain viewers best?
  • How does video length and views relate to APV and AVD over time?
  • What patterns emerge when comparing hundreds of Shorts at once?

This project was designed to serve that distinct analytical purpose.

A core design decision was to present all Shorts in a single table, where each row represents a video and key metrics are visually encoded using a heatmap-style layout. This makes it easy to identify high- and low-performing content, something that is difficult to do within YouTube Studio.

Dashboard Features

KPI Overview

Key performance indicators are displayed at the top of the dashboard, providing an immediate snapshot of overall Shorts performance. All KPIs dynamically update based on active filters and selections.

Two-Page Layout

  • Table Page (see below)
    • Displays all Shorts in a single, comprehensive table
    • Each row represents a video
    • Heatmap-style formatting highlights relative performance for Views, APV, AVD, and Duration
  • Graphs Page (see below)
    • Dedicated to trend analysis and deeper exploration
    • Separates high-level comparison from analytical workflows

Interactive Filtering

  • Date Range Filtering
    • Updates KPIs, graphs, and table values
    • Enables analysis of specific posting windows or experiments
  • Search by Video Title
    • Quickly isolate individual videos or subsets of content
    • All visuals update dynamically based on search results
  • Table-Driven Filtering
    • Selecting videos directly in the table filters all charts
    • Enables fast, exploratory analysis without manual inputs

Visual Analytics

The dashboard includes four core analytical views:

  • Views and APV by Date
    • Drill-down supported
    • Used to analyze performance trends and posting consistency
  • Video Duration and APV by Date
    • Drill-down supported
    • Helps evaluate how content length relates to retention
  • Duration of the Top 10 Highest-Viewed Videos
    • Dynamically updates based on selected date range
    • Used to identify optimal Shorts length
  • APV of the Top 10 Highest-Viewed Videos
    • Dynamically updates based on selected date range
    • Highlights retention quality among high-performing content

Technical Implementation

  • Data Ingestion: Python-based scheduled API ingestion using Google Cloud Run
  • Storage: Structured dataset Google BigQuery
  • Analytics Layer: Cleaned and modeled metrics optimized for reporting
  • Visualization: Interactive Looker Studio dashboard with filters for dates, videos, and performance metrics

Outcome & Impact

  • Eliminated manual CSV exports and repetitive formatting
  • Enabled rapid comparison of Shorts performance at scale
  • Provided clear insight into retention patterns and content effectiveness

This project demonstrates my ability to design purpose-built analytics tools that go beyond surface-level reporting, by translating raw performance data into insights that directly support real-life strategy and decision making.