Video ad hooks that survive the first second: 18 patterns

Meta counts a video view at three seconds. TikTok's feed makes its real decision faster than that: the thumb is moving before the first frame finishes loading, and your ad either interrupts the motion or it doesn't. Everything you spent on the other 27 seconds is conditional on the first one.

The economics are lopsided. An edited short-form ad runs $300-800 from a freelancer or agency. The hook is maybe five percent of the runtime, and it decides whether the rest of the file was produced for an audience or for a folder.

Below: 18 hook patterns that survive the flick, grouped by the mechanism that makes them work, each with an example. Steal the mechanism, not the line.

18 hook patterns, grouped by mechanism

Pain-naming

The viewer stays because you described their Tuesday more precisely than they would have. Precision is the whole trick. "Are you tired of clutter" names a category; a hook names a moment.

The exact moment. Name the instant the problem bites, not the problem. "You've reopened the same tab four times today to check if that invoice got paid."
The embarrassing workaround. People recognize their duct tape faster than their diagnosis. "Stop screenshotting your budget spreadsheet and texting it to your accountant."
The running cost. Put a number on a pain they've stopped noticing. "Your gym membership is $40 a month. You've gone twice since January."

POV:

POV hooks cast the viewer in a scene before they've decided whether to skip. The format is native to TikTok, so it reads as content for a beat longer than it reads as an ad.

POV: inside the problem. Make them the thing that's suffering. "POV: you're the shared spreadsheet, and 14 people are editing you at once."
POV: the future self. Show the after-state as a first-person scene. "POV: it's Sunday night and every meal for the week is already in the fridge."
POV: the bystander. Let someone else's confusion carry the claim. "POV: your coworker watches you log off at 5pm for the third day straight."

Contrarian

A contrarian hook buys you one second of "hang on." It's a loan. The body of the ad has to defend the claim by second four, or the skip comes back with interest.

Attack the standard advice. "Meal prepping on Sundays is the reason you quit by Thursday."
Attack the category. Tell them the shelf they're shopping is the wrong shelf. "Delete your to-do app. You have a calendar problem, not a task problem."
Confess against interest. A limitation, stated first, buys credibility for everything after it. "This posture trainer does nothing if you wear it all day. Fifteen minutes, then take it off."

Number-specific

Round numbers read as marketing. Odd ones read as measurement.

The oddly specific count. "Setup took 11 minutes. I timed it because I didn't believe the box either."
The failure tally. Position the product as the survivor of a process. "I tried nine standing desks. Eight went back."
The price anchor. Swap a familiar recurring cost for a smaller one. "This replaced a $90-a-month habit with a $6 one."

Before/after

Before/after compresses proof into a single cut. The hook's job is to make the gap feel implausible, then start closing it immediately.

Open on the after. Lead with the result, date-stamp the disaster. "This inbox had 4,300 unread on Monday."
Time-compress the before. "Three weeks ago this balcony was a junk drawer with a railing."
Threaten the before. Promise the ugly version is coming and make them wait one beat. "This is my skin today. In ten seconds I'm showing you February. Stay calm."

Overheard

Overheard hooks launder the claim through a third party. A compliment you report lands harder than a compliment you make, because viewers audit a story less than they audit an ad.

The reported question. "My accountant asked what software I switched to. She has never once asked me anything."
The household verdict. "My wife said, and I quote, 'we are not returning this one.'"
The mistaken identity. "Someone at the gym asked if I was a trainer. I started in April."

The hook has to be spoken, not just printed

A common failure mode: the voiceover opens with "hey guys, so I wanted to talk about" while the actual hook sits in a text overlay. That overlay is competing with the username, the sound title, the caption, and the share buttons, in the most cluttered pixels on the screen.

Audio gets in earlier than text. Sound reaches a viewer whose thumb is still mid-scroll and whose eyes haven't committed. A printed line needs focused attention to work; a spoken line works on partial attention, which is the only kind a feed gives you.

Coverage only runs one direction. Burn captions into the video, and a spoken hook reaches sound-on and sound-off viewers both. A text-only hook covers one mode and forfeits the other.

Platforms also transcribe your audio to classify the ad. A hook in the voiceover is signal the delivery system can use. A hook rendered in a font is pixels.

There's a quieter benefit too. A line a human has to say out loud, on camera, in their first breath, can't hide. Weak openers survive in overlays because nobody had to commit to them.

You don't pick the winning hook. You price it.

Nobody pre-selects winners reliably, including the people who write pattern lists. The practical move is to run five to ten hooks against the same body and CTA, then let the feed vote.

That turns hook testing into a production-cost question. At $300-800 per edited ad, six hooks is a $1,800-4,800 experiment. With UGC creators at $60-150+ per video, it's still $360-900 to test six openings of the same ad. Most teams look at that math and ship one hook, untested, which is how $500 of editing ends up riding on a first line nobody pressure-checked.

The test only becomes routine when one more variant costs about as much as the coffee you drink while reviewing it.

That's the cost we built Aitachyon around: paste a URL, get a finished ad with script, voiceover, lip-synced avatar, and captions in about two minutes, at roughly $1.32 a video on Pro. Six hooks against the same product is under $8, and the feed will tell you which one survived the first second.