A. Goal and comparability verdict Primary goal: viewers watching the video to the end. For that goal, Post A has the stronger observed completion result at the shared 7-day checkpoint: Post A: 58% full-watch rate → 1,000 × 0.58 = 580 estimated full watches, equivalent to 580 full watches per 1,000 views Post B: 21% full-watch rate → 2,500 × 0.21 = 525 estimated full watches, equivalent to 210 full watches per 1,000 views Comparability: partial. Both posts came from the same TikTok account, were organic, published two days apart and measured after 7 days. However, they differ substantially in duration (8 seconds vs 28 seconds) and in the underlying idea. So this is not a clean hook test, and the opening cannot be isolated as the cause of the difference. B. Post comparison Post Opening mechanism Goal-relevant total + rate What was observed Main limitation A — 20 Sept Small, specific routine: “The tiny kitchen reset I do before work” 58% full watch → 1,000 × 0.58 = 580 estimated full watches → 580/1,000 views 8-second video; 5.5s average watch time = 68.8% of video length. 12 saves. Much shorter than Post B, so completion cannot be attributed to the opening alone. Other engagement, traffic source and audience location unknown. B — 22 Sept Lifestyle invitation: “Come along for a slow Sunday morning” 21% full watch → 2,500 × 0.21 = 525 estimated full watches → 210/1,000 views 28-second video; 9.0s average watch time = 32.1% of video length. 35 saves. Longer video and broader montage concept. Same account and checkpoint, but not the same idea or duration. C. Observations, hypotheses and unknown Observations At 7 days, Post A had a substantially higher full-watch rate: 58% vs 21%. Post A also retained viewers for more of its runtime: 5.5 ÷ 8 = 68.8%, compared with 9 ÷ 28 = 32.1% for Post B. Post B’s longer average watch time of 9 seconds does not make it stronger for this goal, because the video itself was much longer. The relevant completion result was lower. Hypothesis 1: a very small, clearly defined everyday action may be easier to follow through to the end than a broader lifestyle montage. Hypothesis 2: on the same 8-second kitchen-reset idea, an opening that creates a little curiosity about the everyday payoff may improve completion without becoming clickbait. Main unknown: we do not know how much of Post A’s stronger completion came from its opening versus its shorter duration and narrower idea. D. ONE next test Chosen idea: the quick kitchen-counter reset before leaving home. Variable to change: the opening only. Keep as similar as practical: same TikTok account, organic distribution, approximately 8 seconds, same kitchen-counter clear-up, similar shot order, pacing and visual style. Choose one opening: “The 8-second kitchen reset I do before heading out.” Immediate follow-through: start clearing the first item from the counter immediately. “One tiny thing I sort before I leave the house.” Immediate follow-through: show the untidy counter, then begin the reset straight away. “Nothing fancy — I just don’t want to come home to this.” Immediate follow-through: show the counter as it is, then immediately clear it. 24-hour measure: record: full-watch percentage views average watch time estimated full watches = views × full-watch percentage Do not make the repeat/revise decision at 24 hours, because the existing baseline was measured at 7 days. 7-day repeat/revise criterion: Current matching baseline from Post A: 58% full watch at 7 days Repeat the opening direction if the new post exceeds 58% full watch at 7 days Revise it if it is clearly below 58% at 7 days If it lands close to 58%, treat it as inconclusive and test another opening on the same idea The 24-hour result is still useful to log, but the decision should wait for the matching 7-day checkpoint. This test explores the hypothesis that the opening can improve completion on an already strong short-format idea; it does not prove the original opening caused the 58% result. E. Log opening | idea | date | 24h results | 7d results | observation | repeat/revise/stop