Why the hook is the whole game on TikTok
TikTok scores creative on hook rate: what percentage of viewers watch past the first ~3 seconds. Everything after depends on that number. A brilliant demo, script, and CTA get zero reach if 90% of viewers thumb past your opening. The metric performance marketers watch first is 3-second view rate as a share of impressions — anything above roughly 30% is workable, above 40% is strong, above 50% and the algorithm starts giving you cheaper impressions on its own. Getting the hook right isn't polish; it's the single highest-leverage decision in the ad.
The three-second rule and what it actually means
A TikTok viewer scrolls the For You feed at roughly one video per second-and-a-half. Your ad has to earn the next second explicitly. That means: no logo intro, no throat-clearing, no "hey guys today I'm gonna talk about…". The first frame should show a real face already speaking mid-sentence, or a visual pattern break — a product close-up doing something unexpected, a hand movement that promises a reveal. The first spoken word should ideally be a claim, a curiosity gap, or a direct callout to the viewer.
Six hook archetypes that hold
The formats that reliably work on UGC-style ads: (1) Problem-callout — "POV: your skin keeps breaking out no matter what you do." (2) Objection-buster — "I thought this was a scam. Then I actually tried it." (3) Result-first — "Three days on this and my nails literally stopped peeling." (4) Curiosity gap — "Nobody talks about this but every dentist I know uses it." (5) Comparison — "I switched from [category leader] and here's what changed." (6) Native format — "POV / GRWM / Storytime" openers that mirror organic TikTok patterns. Match the archetype to where the viewer is in their awareness journey: cold audiences respond to problem-callouts and result-first; warm audiences convert on objection-busters.
Openings that kill hook rate
Three patterns to strip out. First, brand logo animations — every second of glossy intro reads as "ad" and viewers scroll. Second, narrator-voice product introductions ("Introducing the new…") — feels like a TV commercial that wandered into the wrong app. Third, question hooks with weak stakes ("Ever wondered how to…") — they signal a monologue is coming. If your first three seconds could open a 2010 YouTube pre-roll, they're too polished for the feed.
How to test hooks without burning budget
The efficient testing pattern: same product, same avatar, same script — only the first 2-3 seconds change across variants. Ship four to eight hook variants at once so the ad account can compare them under identical downstream conditions. Cut the bottom half after 48 hours of consistent underperformance and let the winners scale. Because everything else is held constant, the delta you measure is genuinely the hook, not the audio, avatar, or offer. This is where AI-generated variants earn their keep — re-rolling a fresh opening line is minutes of work, versus reshoot cycles.
When to kill a hook and when to iterate
A hook that starts strong and decays fast (day-one CTR high, day-seven CTR halved) is a fatigued winner — spin variations on the same angle. A hook that never earned viewers in the first place is a losing angle — don't tweak wording, change the whole approach. The mistake teams make is treating every miss as a copywriting problem when it's usually a positioning problem: the wrong audience for the wrong angle. Fix the angle first, then optimise the wording.