Lab log · Batch 2 · posted 30 September 2026
Question hooks against statement hooks
- Test
- Hook: a question (A) against a statement (B). The voice and the top bar also varied, which muddied it.
- Result
- Statements did better (30% stayed against 21%), but the opening pictures differed too. Not a clean result.
What we tested
Whether opening with a question beats opening with a statement. We also tried five different voices and two versions of the top bar in the same batch, which we now know was a mistake: too many things changed at once.
The numbers (6–11 hours after posting)
“Stayed to watch” is the share of people shown the Short who watched instead of swiping.
| Video | Views | Stayed to watch | Hook | Opening picture |
|---|---|---|---|---|
| Dog bed | 1,045 | 39.7% | Statement | A man asleep in a giant dog bed |
| Beer holder (from batch 01) | 655 | 30.7% | Statement | A can flying out of a drink holder |
| Junk Teens | 1,016 | 28.1% | Question | Teens on a truck piled with sofas |
| Car stickers | 1,085 | 24.9% | Question | A car covered in stickers |
| Pasta sauce | 841 | 19.8% | Statement | Two guys cooking |
| Party nights | 648 | 11.2% | Question | A crowded party |
What we decided
- Every video got a similar first push from YouTube (650–1,100 views), then flattened within about three hours. “Stayed to watch” decided which ones went further.
- Statement hooks averaged 30% stayed, question hooks 21%. That’s the opposite of what we expected.
- But the opening picture lined up with the results even better. The top three open on something odd you understand instantly; the bottom two on ordinary scenes.
- So we tested the hook again in batch 03, and wrote down every difference between videos before posting.