30 UGC Hook Formulas for AI Ads (Copy-Paste List)

The first three seconds decide whether the algorithm keeps showing your ad or buries it. Everything downstream (the offer, the product shot, the CTA) only matters if the hook earns the scroll-stop first. Below are 30 hook formulas we reach for when generating AI UGC, grouped into five types, each with a filled-in example so you can see the shape. At the end is the part most hook lists skip: how to convert a written hook into a generation prompt for Seedance 2.0 or Kling 3.0 so the visual and the line land together.

Swap the bracketed variables for your product and pain point. Track hook-to-click rate on each, kill the bottom half after 48 hours, and pour spend into the two or three that separate.

The anatomy of a hook that converts

A working hook does three jobs in three seconds:

1. Targets one person. It calls out a specific viewer with a specific problem, not "everyone." The people it excludes are supposed to scroll.
2. Creates a gap. It opens a loop the viewer has to close: a problem unresolved, a claim unproven, a question unanswered. That tension is what stops the thumb.
3. Matches the visual to the emotion. A confident line over a confused face reads as fake. The delivery, expression, and framing have to agree with the words.

Formula, in one line: [specific person] + [tension: problem, curiosity gap, proof, or contradiction] + [emotional match in delivery]. The 30 below are variations on where you put the tension.

Problem-agitate hooks

Open on the pain, stated plainly, so the person who has it feels seen before you say a word about the product.

1. "My [skin] was getting worse every month and I couldn't figure out why."

2. "If you [wake up exhausted after 8 hours of sleep], this is for you."

3. "I was spending [$200 a month] on [skincare] and my [skin] still looked like this."

4. "Nobody warns you that [the reason your hair is thinning] has nothing to do with your shampoo."

5. "I almost gave up on [finding jeans that actually fit] until three weeks ago."

6. "The most frustrating part of [dry winter skin] isn't the dryness. It's [that nothing you buy actually works]."

Each names a specific ache and sits in it for a beat before offering relief. Agitate honestly. Do not manufacture a problem the product cannot solve.

Curiosity hooks

Open a loop the viewer has to stay to close. The tension is information the viewer does not have yet.

7. "I found out why [my supplements weren't working] and it's kind of embarrassing."

8. "There's one thing [every dermatologist does] that nobody talks about."

9. "This is going to sound fake, but [this $30 tool replaced my $400 routine]."

10. "I wish someone had told me [this about retinol] two years ago."

11. "Watch what happens when I [add one drop of this to my morning coffee]."

12. "The reason [your candles burn out in a week] is not the wax."

The rule for curiosity hooks: the payoff has to be real. If the loop you open is more interesting than the loop you close, you trained the viewer to distrust the next ad.

Social-proof hooks

Borrow credibility from numbers, authority, or a crowd. The tension is "other people already know something you don't."

13. "My [dermatologist] actually asked me what I was using on my [skin]."

14. "I tested [14 serums] in [60 days]. Only one made the list."

15. "[47,000 people] bought this last month and now I get why."

16. "I didn't believe the reviews on this until [I ran out and my skin freaked out]."

17. "My [entire group chat] switched to this in one week."

18. "This has a [six-week waitlist] and I finally understand the hype."

Keep the numbers real. A social-proof hook with an invented stat is the fastest way to fail an ad review and lose viewer trust at the same time.

Contrarian hooks

Take the position the audience does not expect. The tension is a contradiction the viewer needs resolved.

19. "[Vitamin C serums] don't fade dark spots. Unless you're using [this form of it]."

20. "Stop [drinking more water] to fix your [dry skin]. It's not the problem."

21. "The [expensive moisturizer] everyone recommends made my skin worse."

22. "Everyone's wrong about [protein powder] and I have the [bloodwork] to prove it."

23. "You don't need [a 10-step routine]. You need [these two things]."

24. "[More expensive] does not mean [better] and here's what I switched to."

Contrarian hooks earn the strongest stops and the strongest skepticism. Back the claim in the next five seconds or the credibility gap kills the conversion.

POV and story hooks

Drop the viewer into a moment or a first-person point of view. The tension is narrative: what happens next.

25. "POV: [you finally found the product that does what the label says]."

26. "Get ready with me while I explain why [I threw out every other serum I own]."

27. "Day [30] of [using this every morning]. Here's the before and after."

28. "Come with me to [return the $90 cream] and pick up [this instead]."

29. "The moment I realized [my skincare was the problem, not my skin]."

30. "Me, [six weeks ago], vs me [now], same lighting, no filter."

POV hooks lean hardest on the visual. The face, the setting, and the framing carry as much of the stop as the words, which makes the generation prompt matter more, not less.

From hook to generation prompt

A hook is a line plus an emotion plus a shot. To make an AI UGC clip land, you have to encode all three into the generation prompt, not just paste the line. Here is how to translate.

Take hook 1: "My skin was getting worse every month and I couldn't figure out why." The emotion is quiet frustration. The shot is a close, handheld talking head in a real-feeling room. For the talking head, Higgsfield Soul 2.0 is the pick because it locks identity across cuts. Prompt structure:

[Avatar description] looks directly at camera in [specific real location],
says "[hook line]" with a [specific emotion] expression. Vertical 9:16,
[lighting source], UGC style, handheld feel, slight natural camera shake.
No music, clean audio.

Filled in:

Young woman with light brown skin, hair in a loose bun, looks directly at
camera in a softly lit bathroom at 7am, says "My skin was getting worse every
month and I couldn't figure out why." with a tired, slightly frustrated
expression. Vertical 9:16, natural window light from the left, UGC style,
handheld feel, slight natural camera shake. No music, clean audio.

The specific emotion word ("tired, slightly frustrated") is what keeps the delivery from reading as a confident spokesperson. Match it to the hook type: agitate hooks want frustration or fatigue, curiosity hooks want a conspiratorial lean-in, contrarian hooks want a flat, matter-of-fact certainty.

For the product or b-roll shot that follows the hook, route to Seedance 2.0. It holds a real product's identity through motion, which is what a proof shot needs. Prompt structure:

[Product reference] on [surface], [specific motion], [lighting], vertical 9:16,
product stays in focus center frame, 5 seconds, authentic handheld framing.

For copy-ready Seedance structures per shot type, see Seedance 2.0 prompts for UGC ads. For high-volume, face-light formats like unboxings and lifestyle b-roll, Kling 3.0 is faster (roughly 60 seconds per clip) and cheap enough to generate 10 variants until one looks natural.

The pairing that works: hook line to Higgsfield for the talking head, proof shot to Seedance for product motion, filler b-roll to Kling for volume. Keep the same lighting words across all three prompts so the clips cut together without a color jump.

Common mistakes

Writing the hook to sound clever instead of specific. "Skincare that actually works" targets no one. "My dark spots got worse, not better" targets exactly one person and stops them.

Mismatching emotion and line. A great hook delivered by an avatar with the wrong expression is the most common reason a clean generation still flops. Set the emotion word in the prompt, then generate five variants and cut the ones where the face and the line disagree.

Opening a loop you don't close. Curiosity and contrarian hooks that overpromise train viewers to bounce. The payoff in seconds 4 to 8 has to be at least as strong as the tension in seconds 1 to 3.

Testing one hook at a time. Hooks are cheap to generate and impossible to predict. Ship 10 to 15 as a set, read hook-to-click rate at 48 hours, and let the data pick. Do not fall in love with the one you wrote first.

Skipping captions. 85% of TikTok plays are sound-off. If the hook line is not on screen in the first frame, most of your audience never reads it.

---

Pick five formulas, fill in your product and pain point, and generate a testing set today. Start from the full AI UGC workflow, route your winners into human production using the AI UGC vs real creators playbook, and clone the UGC assembly template on 8frame's workflow library. New to the format? Start with what UGC is.