Hooks that travel, and hooks that don't

The hook formula itself is platform-agnostic. "3 reasons you're sleeping wrong" is a number-led hook everywhere it ships. What changes between TikTok, Reels, and Shorts is the surface around the hook: how fast the scroll is, whether audio is on, how much commitment the viewer has already made before the first frame, and what other metadata (title cards, hashtags, channel names) is visible alongside the video. Each of those levers tilts which formulas land hardest.

Scroll speed is a large variable. TikTok's FYP swipe is famously fast: the viewer is in flick-decide-flick mode, often watching muted, and the hook has very little time before the thumb commits. Reels and Shorts viewers often arrive in different contexts (Shorts viewers frequently come from a long-form YouTube session) and may tolerate a longer setup. None of these dwell patterns are published platform constants; they are working observations that vary by audience, and the same hook in three different scroll contexts can perform differently for purely mechanical reasons. Test against your own retention curves.

Audio defaults reshape which sensory channel the hook can use. Sound behavior varies by platform, placement, audience, device settings, and viewing context; no current published figure pins a platform-wide mute rate for any of the three surfaces. The safe design rule: a hook should read silently everywhere, and a hook that relies on a spoken twist or a voiceover punchline ports badly to any muted context. Re-stage the idea as on-screen text to survive the move, then let voice add lift where audio is on.

Surface metadata adds the third lever. Shorts can show a video title alongside the player, so a number-led hook ("3 reasons…") gets a head start from the title text alone. TikTok's caption sits below the video during playback and is rarely the first thing the viewer reads. Reels shows hashtags and caption text under the video but the eye lands on the visual hook first. These different metadata environments are why the rubric's hook category calibrates differently for each surface even when the script is identical. Read more on the upstream signals in /learn/algorithm-signals and on muted viewing in /learn/sound-off.

As a starting hypothesis, the formulas that win a fast, silent scroll decision travel best to TikTok. The pattern-interrupt opener (an unexpected visual, a jarring cut, a frame that doesn't look like the feed around it) is the most TikTok-native hook in the library. It works because the viewer hasn't committed yet and the only signal they have is the first frame. A visual surprise stops the swipe before the brain has a chance to keep flicking.

The Ad Bench methodology: The per-platform hook-score differences in this article (for example, pattern-interrupt and contrarian openers scoring meaningfully higher under the TikTok calibration than under the Shorts calibration, or an identical script landing at 0.78 on Reels and 0.54 on TikTok) reflect The Ad Bench's current scoring rubric and creative-review approach. They are designed to identify creative risk before media spend, not to state platform rules or predict campaign performance.

The contrarian-claim opener is the other TikTok-tilted formula. "Everyone's wrong about creatine" or "Stop using retinol" works on TikTok because the viewer's default posture is skeptical curiosity; the feed trains viewers to expect each video to challenge something. The contrarian claim hijacks that posture and earns the second frame. In the rubric's hook category, contrarian openers also calibrate higher on TikTok than on Shorts, where viewers arriving from long-form YouTube may read contrarian openers as clickbait rather than discovery. Treat that as a testing hypothesis.

What these two formulas share: both work without audio, both land in the first frame, and both create a tension the viewer has to resolve by watching. That trio of properties is exactly what a fast scroll-and-decide loop selects for. Formulas that need a beat to develop (a curiosity gap that opens at second 4, a social-proof anchor that lands at second 6) show up later in the article because the TikTok swipe has usually already happened. For the broader early-distribution mechanics see /learn/tiktok-fyp-cold-start.

Reels viewers often tolerate longer hook setups than TikTok viewers do. That extra runway is exactly what curiosity-gap formulas need. A curiosity-gap hook opens with a question or a setup, withholds the payoff for four to six seconds, then lands the resolution. On TikTok the viewer often scrolls before the gap closes; on Reels they are more likely to wait. This is an observed tendency, not a platform rule, so validate it against your own audience.

Social-proof openers ("100K people already use this" or "15K saved this in a week") pair naturally with Reels because saves are a valuable high-intent signal there, alongside watch time, shares, and viewer satisfaction, and save-anchored hooks signal correctness in the same currency. The viewer reads "saved by 15K" and the implicit ask is "you should save this too." The save economy is the underlying mechanic; see /learn/reels-save-economy for how saves fit among Reels signals and how that flows back into hook design.

The portability cost: a curiosity-gap hook tuned for Reels often under-performs on TikTok because the gap stays open past the swipe horizon. The fix when porting Reels-to-TikTok is to compress the gap into the first two seconds (either close it earlier or reveal the question and answer simultaneously on-screen). A social-proof hook ports better in either direction but may lose some lift on Shorts, where subscriber intent and repeat viewing matter alongside saves and the social-proof anchor doesn't hit the same currency.

Shorts has a structural advantage the other surfaces mostly lack: the video title can be visible alongside the player. A number-led hook ("3 reasons you're sleeping wrong", "5 things I wish I knew at 25") gets a head start from the title text alone, then the first frame just has to confirm the promise. On TikTok and Reels the same hook has to do the title's work itself, which spends a beat the silent-fast surfaces don't have to give. The rubric reflects this: number-led openers calibrate higher on Shorts than on the other two.

Problem-agitate-solution hooks also land well on Shorts because the format wants compressed structure. Repeat viewing is a positive signal worth designing for when it improves the idea, and a tight PAS arc (problem in frame 1, agitation in frames 2–4, solution by frame 8) maps cleanly onto a 15–30 second runtime that's designed to be watched twice. The viewer hits the problem on the first pass, registers the agitation on the second, and the solution feels earned. The repeat-view mechanic is covered in detail in /learn/shorts-repeat-view-economy.

The portability tradeoff runs the other way too. A number-led hook ported from Shorts to TikTok loses the title-card head start and has to earn the same number-of-reasons promise from frame 1 instead of from pre-play metadata. PAS hooks port to Reels with minor compression edits; they port to TikTok only if the "problem" can be landed visually inside the first second, otherwise the agitation beat blows past the swipe horizon.

The workflow when you have a hook that won on one surface and want to ship it on another: first identify which platform lever made the original work. If it was pattern-interrupt on TikTok, the lever was silent visual surprise. If it was curiosity-gap on Reels, the lever was the longer runway. If it was number-led on Shorts, the lever was the title-card preview. Then swap that lever for one the target surface favors. Don't try to make the original lever work harder; replace it.

Concrete example. A TikTok hook that won with a pattern interrupt (say, a creator dropping a phone in a glass of water, frame one) wants to ship to Shorts. The original lever was the silent visual surprise; on Shorts that lever is weaker because the title card has already set context and the slower scroll dilutes the shock. Swap it for a Shorts-leaning mechanic: convert the same script into a number-led frame. "3 things this phone survives that yours can't" as the title card, frame one shows the same glass-of-water moment but now reading as item 1 of 3 instead of as standalone shock. The visual lift carries; the framing adapts to the surface that's receiving it.

The same trick runs in every direction. Reels-to-TikTok: compress the curiosity gap so the question and answer land in the first two seconds instead of at second four. Shorts-to- Reels: trade the number-led title-card frame for a social- proof anchor that leans into the Reels save behavior. TikTok-to-Reels: extend the pattern-interrupt setup with a one-beat curiosity gap so the gain isn't spent on the first frame. The library at /library/hooks has all 30 formulas with templates; the portability map in this article tells you which one to reach for when the rewrite is for a different surface than the original.

The mistake to avoid: shipping the same hook unchanged across all three platforms and assuming the rubric will score them equally. It won't. The hook category in the rubric calibrates per surface, and an identical script can land at 0.78 on Reels and 0.54 on TikTok in the current rubric purely because the lever that made it work doesn't exist on the receiving platform. Those are Ad Bench scores, not performance predictions. Treat each port as a rewrite with one variable changed, then test.

1. TikTok Support. "How TikTok recommends content." Accessed August 24, 2026. support.tiktok.com
2. Meta Transparency Center. "Our approach to explaining ranking." Accessed August 24, 2026. transparency.meta.com
3. The Ad Bench methodology, current scoring rubric and hook-category calibration. Reviewed August 24, 2026.