The hook gets drafted from the words already in your take

The word generator oversells slightly, and it is better to say so on the way in. Cutroom transcribes your take and drafts the opening headline out of what you actually said. You pick a draft, or you write your own.

It does not invent a claim you never made. A headline promising what the take cannot deliver buys three seconds of attention and spends them on annoyance, which trains an audience to distrust the next ad from the same account.

That grounding is the useful property. An opener written from the transcript can only promise things the ad goes on to say, so hook and body cannot drift apart. Drifting apart is the standard failure of writing hooks in a separate tool.

The rest of this page is the part that transfers anywhere: eight patterns, each with a worked example.

The callout names the viewer so precisely the right person feels caught. Weak: 'Attention all business owners.' Worked: 'You run a salon and your last ad was the price list.' The test is whether the wrong person scrolls away immediately. They should.

The bold claim anchors on a number that sounds counted rather than rounded. Weak: 'Save tons of time on editing.' Worked: 'Eleven ads came out of one three minute take.' Odd, specific numbers read as records. Round ones read as copywriting.

The open loop asks a question and holds the answer back one beat. Weak: 'Want better ads?' Worked: 'Why did the ugly version outspend the pretty one nine to one?' A loop only works when the ad genuinely closes it.

The pattern interrupt breaks the feed's texture in the first frame: an odd image, a negative, a sentence started mid-thought. Worked: 'Delete your best performing ad. Here is why.' Spend it carefully. A viewer interrupted twice by the same trick is a viewer trained to leave.

The villain tease blames a culprit the viewer already resents. Weak: 'The industry is broken.' Worked: 'The real reason your ads die on day three is the platform rewarding your worst habit.' It earns attention by taking a side.

Proof first pulls the result to the front and explains afterwards. Worked: 'This took four minutes. Watch.' Then the ad shows the four minutes. Nothing outperforms evidence delivered before the pitch, when you have evidence.

The damaging admission opens on the honest flaw. Worked: 'This will not fix a bad offer, and it costs $19.99 a month.' A confession buys the credibility every later claim spends. It is also the pattern this page's own product uses on itself, deliberately.

The takeaway tells part of the audience not to buy. Worked: 'If you post once a month, do not buy this.' Exclusion reads as confidence. The viewers who remain lean in, because a seller who turns people away appears to have buyers enough.

Whichever pattern you pick, the machine protects it structurally. The first two and a half seconds of every ad are a guarded zone.

The first caption group is capped at four words, whatever caption pack you chose, because the opening line is read before it is heard and a wall of text is not a hook.

No fullscreen footage may appear before 1.2 seconds, so the opening frames belong to a face. The one exception is the pattern interrupt, which is allowed to open on the visual because being visual is its whole mechanism.

None of this needs configuring. It is the difference between a hook generator that hands you a sentence and one that also builds the three seconds around it.

Run the eight patterns against a take with no number, no result, no named audience and no admission, and every draft comes back as arranged emptiness. The generator can only shape what the take gives it.

The fix costs thirty seconds and happens before recording. Write down one claim a stranger could check. A number you counted, a before and after, a thing you refuse to promise. Say that on camera and every pattern above suddenly has material.

This is the honest reason hook tools disappoint people. They are asked to compensate for the script, and no arrangement of eight words compensates for having nothing to say.

You will write many of these either way. Winners run at roughly 5 to 8 percent of creatives, per Motion's analysis of 550,000+ Meta ads. The patterns make the writing faster. Only the claim makes it work.

The transcription, the drafted headline and the whole first cut are one batch at 100 credits. Rewriting the headline afterwards costs nothing, and export is 20 credits per minute of finished video, so a thirty second variant is about 10 credits to download.

That arithmetic is the point for hook testing. The batch is paid once. Every new headline on the same take is another cheap export rather than another edit.

Lite is $19.99 a month for 1,250 credits, nothing on the picture, cancel in one click.