Question Hooks and 5 Other Examples That Survive the Three-Second Scroll

You post a short-form video. Somewhere between second two and second three, a thumb decides whether it keeps going or flicks past you forever. That thumb doesn't read your caption, doesn't care about your edit, and definitely doesn't wait for you to get to the point. It just scrolls — and on most feeds, it scrolls on more than 80% of videos before the three-second mark.

That's the actual problem. Not "make better content." Not "post more consistently." The first line you say, and the first words on screen, decide whether anyone sees the other 27 seconds you spent editing.

Why the opening line carries all the weight

Every short-form platform ranks on watch time before anything else happens — before a like, before a comment, before a follow. A video that loses most viewers in the first three seconds never gets a chance to prove the rest of it is good. The algorithm already moved on.

That puts an odd amount of pressure on a single sentence. Not the thesis of the video. Not the summary. Just the first thing you say and the first thing someone reads on screen, working together to make a thumb pause.

There are a handful of hook types that keep showing up in videos that actually hold attention:

• Contrarian — say the opposite of the obvious advice, then earn it.
• Question — ask the thing the viewer was already half-wondering.
• Mistake — name a specific error the viewer is probably making right now.
• Stat/shock — lead with a number that doesn't match intuition.
• Story cold-open — drop into the middle of a scene, no setup.

None of these are exotic. What separates a hook that works from one that doesn't is specificity — "day 1 is where you quit" beats "fitness tips," every time, because it names a moment the viewer recognizes instead of a topic they're supposed to care about.

Writing a question hook that actually lands

A weak question hook restates the topic as a question: "Did you know gyms are hard?" A strong one surfaces something the viewer has felt but hasn't said out loud: "Why does everyone quit by January 20th?" The difference is specificity again — a date, a number, a named moment — not just a question mark at the end of a generic sentence.

The same rule holds for the other types. A contrarian hook needs something to push against ("stop going to the gym every day" only works because it contradicts what everyone's already been told). A mistake hook needs to name the mistake precisely enough that the viewer recognizes it as theirs, not a generic category of error.

Getting five or six of these — one contrarian, one question, one mistake, a couple of variations — and picking the one that fits the specific audience is the actual skill. Most creators either write one hook and ship it, or write five and can't tell which one is strongest. Both are expensive mistakes when your topic is competing with every other video in the feed.

What the tool returns

Hook Lab, a Capafy skill, takes a plain description of the video you're making and returns a full hook set instead of one guess. Here's a real, unedited example — the kind of input you'd actually type and the output that came back:

You send: "Hooks for a 30s TikTok about why most people quit the gym in January — energetic, for beginners."

You get back (shortened):

• Positioning: "The gym isn't the problem — your first week is." You own the beginner-who-keeps-quitting angle.
• Hooks (5+, each with spoken line + on-screen text + type):

1. Contrarian — 🗣 "Stop going to the gym every day." / 🖥 "Day 1 is where you quit" 2. Question — 🗣 "Why does everyone quit by January 20th?" / 🖥 "The January 20th wall" 3. Mistake — 🗣 "You're not lazy. You're doing week one wrong." / 🖥 "Week 1 mistake"

• Recommended pick: #3, with one line on why it fits a beginner audience.
• Script: 0–3s hook · 3–10s the mistake · 10–22s the fix in 3 steps · 22–30s CTA.
• Caption: short caption + CTA + hashtags, plus one filming tip.

That's the whole deliverable in one pass: a positioning line, a ranked set of hook options by type, a pick with a reason attached instead of a bare recommendation, a timed script structure, and a caption. The part that matters most for the scroll problem is the spoken-line-plus-on-screen-text pairing on every hook — most creators write only the thing they'll say and leave the on-screen text as an afterthought, when it's doing half the work of stopping the scroll.

Try it here: https://capafy.ai/agent/8123079349?ct=hooklab_blog

The part that's still on you

No tool picks up a camera. The fix in the script above — three steps, delivered between second ten and twenty-two — still has to be true and specific to your thing, not generic advice. A hook set gets you past the three-second wall with a viewer who's curious; whether they stay for the rest of the video still depends on whether what you say next is worth their time.