Test Sun for higher-priority uploads
strong
High priority | publishing
Finding:
Sun had the strongest median views per video among weekdays with enough observations.
Recommendation:
Consider scheduling higher-effort uploads on Sun when the topic and production schedule fit.
Evidence:
Median views per video were 20,774 (32% vs the weighted weekday baseline) across 20 videos.
Caveat:
This is descriptive, not causal. Weekday effects can be confounded with topic, event timing, format, or creator availability.
Test more around topic: music
strong
High priority | content
Finding:
`music` has the strongest average views per video among groups with enough examples.
Recommendation:
Use `music` as a candidate for repeatable testing, while checking fit with audience and production constraints.
Evidence:
Average views per video: 31,323 (28% vs the weighted group baseline) across 31 videos.
Caveat:
Groups can overlap with format, publishing timing, collaboration, and title effects. This rule is descriptive.
Watch tag signal: shorts
strong
High priority | content
Finding:
`shorts` has the strongest average views per video among groups with enough examples.
Recommendation:
Use `shorts` as a candidate for repeatable testing, while checking fit with audience and production constraints.
Evidence:
Average views per video: 39,773 (50% vs the weighted group baseline) across 71 videos.
Caveat:
Groups can overlap with format, publishing timing, collaboration, and title effects. This rule is descriptive.
Evaluate collaboration lift
strong
High priority | content
Finding:
Collaborative uploads are not clearly outperforming non-collaborative uploads on average views per video.
Recommendation:
Do not increase collaborations based on this signal alone; prioritize fit and concept quality.
Evidence:
Collaborative average: 16,811; non-collaborative average: 27,968; lift: -40%.
Caveat:
Collaboration labels can be imperfect, and collaborations may cluster in specific topics or event types.
Repeat format tests: short
strong
High priority | performance
Finding:
short has the strongest average views per video among formats with enough examples.
Recommendation:
Use this format as a candidate for repeatable testing, while checking whether the advantage holds by topic and timing.
Evidence:
Average views per video: 39,773 (62% vs format baseline) across 71 videos.
Caveat:
Format performance can be influenced by topic mix, upload age, collaboration, and promotion.