The First 3 Seconds: Writing Hooks That Make People Stop Scrolling
Written byNereida Enriquez • Communications Associate
Video - November 2025
Every video is really two separate pieces of content stacked on top of each other. There's the video you spent a week planning, shooting, and editing. Then there's the three seconds at the very front that determines whether anyone ever sees the rest of it. Most brands pour nearly all their effort into the first piece and improvise the second, which is the equivalent of catering an elaborate dinner and forgetting to send out any invitations.
Here's what the data actually shows. TikTok's own numbers put sixty three percent of the highest performing videos as hooking viewers within the first three seconds. Separately, roughly seventy one percent of viewers make their stay or leave decision within those same first few seconds. But the mechanism underneath that number is what actually matters operationally. When someone swipes away at second two, you haven't just lost that one viewer. You've sent the system a signal that your content failed its initial test, and the system responds by not showing it to the next pool of strangers at all.
There are three recurring failure patterns worth naming specifically, because they show up across nearly every brand account we've audited. The first is the logo animation or brand sting, the kind of opening that reads as polish on a website and reads as an ad the instant it appears in someone's feed. A logo has never once stopped a scroll, structurally speaking, because it contains no information a viewer needs and no reason to keep watching.
The second is the warm up opening. "Hey everyone, so today we wanted to talk about our new spring collection." By the time that sentence finishes playing, the viewer has already scrolled past two other videos. Any slow buildup where the first three to five seconds go to pleasantries functions as a retention killer, regardless of how good what follows actually is.
The third is the vague tease. "This changed everything." "You have to see this." These mimic the rhythm of a real hook without delivering any of its substance, because they promise nothing specific enough for a viewer's brain to register as worth staying for. A hook has to communicate, instantly, what the viewer gets in exchange for the next ten seconds of their attention.
Lead with the payoff and explain it afterward. Don't build toward your result the way a traditional narrative would. Open on the result itself. A skincare brand doesn't need four seconds establishing the bottle before the reveal. Show the reveal first, then earn it retroactively with the explanation. This inverts the storytelling structure most people were trained on in school, and it's consistently the highest leverage single change most brand accounts can make to their opening.
Make a specific claim you immediately follow through on. "This posting habit quietly killed my reach last year" works structurally because it names a concrete consequence and implies you're about to reveal exactly what caused it. The follow through has to actually happen fast, because a hook that overpromises and underdelivers tanks completion rate, and completion rate is precisely what determines whether the system expands your reach to the next test pool.
Interrupt the established visual pattern of the feed itself. Feeds have a rhythm, and viewer brains scroll on a kind of autopilot until something breaks that pattern. A sudden movement, an unexpected cut, bold on screen text, even an abrupt moment of silence, functions as a pattern interrupt that jolts someone out of autopilot scrolling.
Design deliberately for silence, because roughly half of any given audience watches with sound off by default. If your hook lives entirely in the audio track, it functionally doesn't exist for half your potential audience. The promise needs to live on screen as text, in the first frame, in a font legible at arm's length without needing to lean in.
Turning this into a repeatable system is the harder half of the work. Writing one great hook is a creative act. Writing a reliably great hook for every single video, indefinitely, is an operations problem, and this is where most in-house teams actually break down over time.
The structural fix is separating hook production from body production entirely. Shoot the body of the video once. Shoot three to five distinct openings for that same body. Same content, same edit downstream, different first three seconds each time. Post these as variants, or run them as a small paid test, and let the resulting data decide which one wins instead of relying on internal opinion. This removes guesswork almost entirely, because instinct about hooks is wrong roughly half the time even among people who do this professionally. After twenty or so tests, a clear pattern usually emerges specific to your actual audience, whether that's a response to blunt cost claims or a total lack of response to aspirational lifestyle framing that looked promising on paper.
Keep a real log alongside this process. Date, platform, hook text, first frame description, and outcome metrics including watch time, completion, and click through. Reading the retention curve for each video tells you exactly where a problem actually lives. A dip in the first two seconds signals a weak promise in the hook itself. A cliff appearing in the middle of the video signals dead air in the body, unrelated to the hook at all. Most brand teams only look at total views, which tells them almost nothing actionable about where to actually fix the content.
There's a real limit to what a hook can fix, and it's worth stating plainly. A strong opening cannot rescue a video that has nothing to say once it's been opened. If the body is a product feature in search of an audience, a clever hook just gets more people to the exact moment they realize their time got wasted, which trains that specific viewer to skip your account next time regardless of the opening. Chasing hooks too aggressively can also deform an account's identity over time. If every video opens with a provocation purely because provocation tested well once, the account slowly becomes defined by provocation rather than by whatever it actually does well, and that pattern is exhausting to sustain and hard for an audience to trust after enough repetition. The strongest accounts hold a consistent voice underneath and vary the hook within that voice, not the other way around.