How to Create Video Hooks That Stop the Scroll
How to create video hooks that stop the scroll: the psychology, proven opening-line formulas, the visual-hook rule, and how to A/B test the first three seconds.
The first one to three seconds decide whether anyone watches the rest of your video. I have sat through enough edit reviews to know the pattern by heart: a sharp script and a clean cut, all of it wasted because the opening gave nobody a reason to stop their thumb. So I build the hook before I worry about anything downstream, and I judge it on one number most teams never pull. This is the process I use, and how I prove the hook actually did its job instead of hoping it did.
If you only want the essentials:
- Fix the hook before anything else. Early watch-through is a ranking signal, so a weak opening buries the whole video, not just the first viewer.
- Open on your single strongest frame, with the hook as on-screen text. Most feed video plays on mute.
- Make a promise you can actually pay off, then A/B test three to five openings on hook rate, never on raw view counts.
Why the hook decides your reach, not just your first viewer
A weak opening does more damage than losing one viewer. It suppresses the whole video. Platforms treat early watch-through (the two or three second view rate, then overall completion) as a core ranking signal, so a soft first three seconds tells the algorithm to stop pushing the video to anyone at all. I think of the hook as a distribution lever more than a courtesy to the person watching.
Two metrics I keep separate, because teams constantly blur them:
- Hook rate: the share who watch past the opening threshold (roughly 2 seconds on TikTok, 3 on Meta). A healthy TikTok hook rate sits near 30 percent, and the strongest work I see clears 40 to 45.
- Hold rate: how far the rest of the audience gets before they drop.
I fix the hook first every single time, because a strong body cannot rescue an opening nobody reaches. There is real comfort in the data, too: even viewers who bail inside three seconds still show measurable lifts in recall and intent, and roughly three quarters of a campaign’s impact lands inside the first 10 seconds. The opening seconds carry the weight.
The levers that actually stop a thumb
Every hook that works is pulling one of a handful of psychological levers, and once you can name them you stop writing openers by feel. These are the ones I reach for, with the kind of line each one produces.
| Lever | Why it stops the thumb | An opening line it produces | Where it fits |
|---|---|---|---|
| Curiosity gap | Opens a loop the brain wants closed | ”Most people get their subject lines completely wrong.” | Education, how-to |
| Pattern interrupt | Breaks the feed’s expected rhythm with novelty | A blender roaring to life over a brand-new phone | Product demos, brand |
| Negativity bias | Loss aversion outweighs the promise of gain | ”Stop doing this if you want your ads to convert.” | Warnings, common mistakes |
| Bold or contrarian claim | The viewer stays to confirm or argue | ”Your homepage video is costing you customers.” | Opinion, thought leadership |
| Direct question | The viewer answers it in their own head | ”When did you last finish a video ad?” | Relatable, top of funnel |
The curiosity gap is the one I lean on most, because it works for education, and that is where most of my clients live. I always open with the lever before I write the sentence; the words come easier once I know which nerve I am pressing.
What I never open with
The fastest way to kill a hook is to start with throat-clearing. “Hey guys” and “In this video” both hand the viewer several seconds of nothing while the thumb is already moving, and a slow logo intro is worse. “Hey guys” was the single most common opening line across an analysis of more than a million videos, which is exactly why it earns nothing now. If a phrase could open literally any video on the platform, it is not a hook.
Put your strongest frame first
Your opening frame should be the single most striking image in the whole video: high-stakes motion or hard contrast, never a title card or a stock plate. Because most feed video gets watched on mute, the words of your hook have to appear as on-screen text from the very first word, so the line lands even in silence. I burn the hook text in by default, and if an opening only works with the sound on, I treat it as broken.
A hook I still pull up on scoping calls: Will It Blend
When a client tells me their product is too boring to hook anyone, I pull up Blendtec’s “Will It Blend?” It is the clearest demonstration of every principle above, made by a company that sells kitchen appliances.
“Will it blend? That is the question.” Then a brand-new iPhone drops into the blender. That one line is a curiosity gap and a pattern interrupt at the same time, and you physically cannot scroll away until you see what happens. Blendtec built an entire demo series on that single repeatable opening. The same mechanism runs under the other famous ones: Dollar Shave Club opens on a founder making a blunt, funny claim in the first line, and Old Spice’s “look at your man, now back to me” is a pattern interrupt so fast the eye cannot leave.
Make a promise, then keep it
A hook is a promise, and the body has to pay it off. Bait that never delivers wrecks retention and trains the algorithm against your whole account, so I only write openings I can actually cash. When the hook implies a payoff the video does not contain, I cut the hook rather than shoot something that lies to the person who stopped for it.
Test the hook, not the whole video
The hook is the highest-return variable you can A/B test, so I test it and hold everything else steady. I run three to five different openings against the same body and judge them on hook rate or swipe-away rate, never on raw view counts, which flatter you for the wrong reasons. One rule I never break: I do not compare hook-rate numbers across platforms, because TikTok, Meta and YouTube each count a “view” at a different threshold, so the comparison means nothing.
The one stat to stop repeating
Retire the “eight-second attention span, shorter than a goldfish” line for good. It is a debunked, misattributed stat, and dropping it into a deck just dates you. Hooks matter because the feed is infinite and scrolling costs nothing, so the viewer always has somewhere better to be. Human brains did not shrink; the competition for them got bigger.
At Moonb I help teams build video engineered to win those first three seconds, cut for every channel their audience actually watches, as an embedded creative team. If your videos keep dying before anyone reaches the good part, see how Moonb works.
Frequently asked questions
The hook is the first line and the first frame together, so it ends the moment you deliver on the promise it made, usually inside the first two to three seconds. On TikTok I want the tension to break by second two. On long-form YouTube I stretch the setup to the first 15 to 30 seconds, because the viewer already chose to click and is giving you a little more rope.
The psychological levers are identical, but the delivery shifts. On short-form, the visual and the first spoken line carry everything. On long-form, the thumbnail and title are the real hook that earns the click, and the opening 30 seconds have to confirm the click was worth it rather than restart the pitch from zero.
I write five to ten hooks first, then build the video that pays off the strongest one. Writing the body first tends to produce a hook that summarizes the video instead of creating tension, and a summary is not a hook. Start with the opening that stops the thumb, then earn it.
