YouTube Title and Thumbnail Quality Optimisation
Backlog project to scrape YouTube metadata and evaluate title and thumbnail quality using machine learning.
Notes
Overview
This project aims to evaluate the quality of YouTube titles and thumbnails using historical performance data.
Description
The system would scrape public YouTube metadata including video titles, thumbnails, publish dates, and view counts. This data would then be used to train models that attempt to correlate visual and textual features with engagement outcomes.
The intent is not to predict virality, but to provide relative quality signals that help creators assess whether a proposed title and thumbnail are likely to underperform or align with successful patterns.
This project is currently in the backlog and remains exploratory, with open questions around dataset bias, creator niche separation, and platform dynamics.