Lab log · Batch 3 · posted 1 October 2026
Question hooks against statement hooks, again
- Test
- Hook: a question (A) against a statement (B), on tech and controversial stories
- Result
- The opening picture decided it again: a physical object kept 43.6%, screens kept 13–24%. Several other things changed at once.
What we tested
The hook, again, after batch 02 gave a muddy answer. This time the topics were tech and controversial business wins.
The numbers (5–8 hours after posting)
| Video | Views | Stayed to watch | Opening picture |
|---|---|---|---|
| Satan Shoes | 683, still climbing | 43.6% | A giant sneaker dripping red, held up to the camera |
| HeyReach | 124 | 23.9% | Founders toasting round a laptop |
| Pushups app | 120 | 18.6% | A man mid-pushup behind a padlocked phone |
| Cluely | 55 | 12.9% | A man smirking at a laptop with an AI window |
| Corteiz | — | — | Never uploaded: we hit YouTube’s daily upload limit |
What we decided
- YouTube’s first push was much smaller this time (55–124 views), except for Satan Shoes. Again, the video with the best “stayed to watch” was the one YouTube kept showing.
- The opening picture explained the results better than the hook. The winner opens on a physical object; the losers open on screens, which can’t be read at swiping speed.
- Too many other things changed for every video at once to read the hook result: the topics, the voice, the ending, the tags and description, and the end banner.
- Next: test the first frame on its own, with the script fixed.