The 12 hook types that stop a scroll
A hook is the first spoken line of a clip, and it decides more than everything that follows it. This is the taxonomy ClipAI’s scoring model uses in production to classify every clip it detects: twelve archetypes that earn attention, and four opener patterns that reliably lose it. No formulas, no “guaranteed viral” templates, just what each pattern is, when it works, and how it fails.
How this list was made
These twelve archetypes are not editorial opinion. They are the exact classification set ClipAI’s clip scorer applies to the opening line of every candidate clip, distilled from public hook research and calibrated against hand-scored transcripts. Every clip the product detects gets one of these labels, so the taxonomy has to survive contact with real footage every day.
One honest note on grading: most raw, unedited long-form content scores in the middle of the range. A genuinely elite hook is rare, and a scoring system that hands out 90s to everyone is flattering you, not helping you. ClipAI’s scores are deliberately calibrated so the difference between a 45 and a 78 means something.
The archetypes
Twelve ways a first line earns the next ten seconds
- 01
Specific number
The first line contains a concrete, surprising figure: an amount of money, a count of days, a percentage. Specificity reads as evidence of a real story rather than a generic lesson.
“I turned $200 into $12,000 in ninety days, and the method is embarrassingly simple.”
Works when The number is unexpected and the viewer immediately wants the mechanism behind it.
Fails when Ordinary numbers do not help. "I saved 10% on groceries" is specific and still boring.
- 02
Contrarian take
The opening names a belief the viewer probably holds, then flatly reverses it. The tension between what they think and what you said keeps them watching for the argument.
“Most people think meal prep saves money. It's actually costing you.”
Works when The reversed belief is genuinely common, so most viewers feel personally contradicted.
Fails when Contrarian about something nobody believes is just a strange claim with no tension.
- 03
Bold claim
A strong assertion delivered without hedging. The confidence itself is the hook: the viewer stays to see whether you can back it up.
“Every productivity app you have installed is making you slower.”
Works when The claim is falsifiable and the clip actually delivers the backing.
Fails when True-but-generic advice reads as bold and scores mid-low. "Consistency matters, show up every day" is a bold claim with nothing to prove.
- 04
Curiosity gap
The first line opens a loop and deliberately withholds the resolution. The viewer's need to close the loop does the retention work.
“There's one clause in almost every lease that landlords hope you never read.”
Works when The withheld thing is concrete and the payoff arrives inside the clip.
Fails when Gaps that never close feel like clickbait, and the scorer's clarity dimension punishes a payoff the clip does not contain.
- 05
Before / after
Two states of the same subject, far apart, stated back to back. The distance between them implies a method, and the method is the video.
“Six months ago I couldn't run a mile. Last Sunday I finished a marathon.”
Works when The transformation is large, believable, and time-boxed.
Fails when Small or vague deltas ("I used to be bad at this, now I'm better") have no gap to explain.
- 06
POV scenario
The opening drops the viewer inside a situation instead of describing one. Second-person framing makes the stakes theirs.
“Imagine you wake up tomorrow and every client you have has cancelled.”
Works when The scenario is one the audience actually fears or wants, so immersion is instant.
Fails when Scenarios outside the viewer's life play as fiction and get swiped.
- 07
Experiment
A self-imposed test with defined rules: same meal for 30 days, no phone for a week, one cold email a day. The structure promises a result, and results are watchable.
“I ate the same meal every day for thirty days. Here's what happened to my blood work.”
Works when The rules are strict enough that the outcome could have gone either way.
Fails when Experiments with an obvious outcome are a setup without suspense.
- 08
Price reveal
The cost of something, stated plainly and early. Money is the most universally legible number there is, and a surprising price in the first line is an instant re-anchor.
“This entire home studio cost me $180.”
Works when The price is far from what the viewer would have guessed, in either direction.
Fails when Expected prices confirm rather than surprise, and confirmation does not stop a scroll.
- 09
Mistake confession
The speaker opens with their own failure and what it cost. Vulnerability plus a named consequence buys trust and attention at the same time.
“The biggest mistake I made was hiring my best friend. Here's what it cost me.”
Works when The mistake is one the viewer could plausibly make, and the cost is concrete.
Fails when Humble-brags disguised as confessions ("my mistake was working too hard") read as insincere.
- 10
Question / challenge
A direct question or dare aimed at the viewer. Done well, the viewer starts answering in their head, and now they are participating instead of scrolling.
“Can you name one subscription you actually used every day last month?”
Works when The question has a specific, slightly uncomfortable answer.
Fails when Rhetorical filler like "have you ever wondered..." is so overused the scorer treats it as a weak opener, not a hook.
- 11
Command
An imperative first line: stop doing X, delete Y, never sign Z. The directness is jarring in a feed full of throat-clearing, and jarring earns a beat of attention.
“Stop keeping your savings in a checking account.”
Works when The command contradicts the viewer's current behavior, so it demands a justification.
Fails when Commands about things the viewer already does not do land on nobody.
- 12
Credential drop
A specific, relevant credential stated up front to make the next sentence carry weight: years in the field, volume of work, a named role.
“I've edited four thousand short-form videos. Only one pattern survived every trend.”
Works when The credential is quantified and directly earns the claim that follows.
Fails when Credentials without a payoff are just bragging, and vague ones ("as an expert...") are noise.
The scroll-killers
Four openers the scorer caps on sight
Greetings and self-introductions
"Hey guys, welcome back to the channel." The first seconds are spent on ritual instead of content, and viewers who do not know you have no reason to stay for it.
"Have you ever wondered..."
A curiosity-gap formula so worn out it now signals template content. The scorer treats it as a weak opener regardless of what follows.
Filler starts
"Okay so...", "So basically...". Verbal throat-clearing that tells the viewer the real first line has not arrived yet. They rarely wait for it.
Trust-dependent claims
Openers that only land if the viewer already knows and believes the speaker. Cold viewers, which is most of a feed, bounce off them.
The subtler version of this problem: in raw conversation the hook almost never falls on the first word of a talking turn. When ClipAI validated its boundary detection on a 48-minute interview, roughly a third of the detected clips needed their start point moved, sometimes by more than 20 seconds, so the clip opens on the hook line instead of the wind-up. If you cut clips by hand, the single highest-leverage edit is the same one: start at the hook.
Questions
What is a hook in a short-form video?
The hook is the first spoken line or visible moment of a clip, the part a viewer judges before deciding to keep watching or swipe. On TikTok, Reels, and Shorts that decision usually happens in the first one to two seconds, so the hook carries more weight than any other part of the clip.
Which hook type performs best?
There is no universally winning archetype; specificity is what separates strong hooks from weak ones inside every type. That said, in ClipAI's scoring rubric the highest calibration anchors are specific numbers and mistake confessions, because both signal a concrete, real story in the first sentence.
How long should a video hook be?
One to two sentences. ClipAI scores opening line strength on exactly that basis: would the first sentence or two, heard in isolation, stop a scroll? If the interesting part arrives in sentence three, the clip should start at sentence three.
How do I know if my hook actually works?
Test it before posting. ClipAI's free score tool grades any clip's hook quality from 0 to 100 and names which archetype the opening uses, with no sign-up required. After posting, the score is checked against the clip's real tracked views, so the grade is accountable rather than a guess.
Do these hook types work for podcasts and interviews?
Yes, but raw conversation rarely opens on the hook line. In ClipAI's validation on a 48-minute interview, roughly a third of detected clips needed their start point moved, sometimes by 20 seconds or more, so the clip begins where the hook actually lands rather than at the start of a talking turn.