27 UGC Ad Hook Formulas That Stop the Scroll (With Examples)
A swipe file of 27 UGC hook formulas in 6 families — templates, filled examples across skincare, gadgets, supplements, home, and pet, plus how to test them.

On this page
A video ad is not 30 seconds long. It is 1.5 seconds long, followed by 28.5 seconds that only exist for the people who survived the first 1.5. That makes hook rate — the share of impressions that watch past the 3-second mark — the leading metric for the whole creative. CPA, ROAS, and CTR are all downstream: if nobody stays past second two, nothing else in the video happened.
So hooks are the highest-leverage words in performance marketing, and the one part of an ad you should never write once. This is a swipe file: 27 formulas in 6 families, each with a fill-in template, examples across skincare, gadgets, supplements, home, and pet products, and a note on when it works. One rule first: a hook is a spoken line and a first frame — the pairing section covers the visual half.
Family 1: Problem and agitation
The oldest structure in direct response, compressed to one line: name the pain in the viewer's own words, not the product's category language.
1. Name the pain, precisely. Template: "Your [body part/possession] isn't [symptom] because of [assumed cause]. It's because of [real cause]."
- Skincare: "Your skin isn't breaking out because it's dirty. It's breaking out because you're stripping it twice a day." Works when: the audience misdiagnoses their own problem; the correction is the curiosity.
2. The relatable ritual. Template: "If you [oddly specific behavior], this is for you."
- Supplement: "If you've been awake since 3am doing math about tomorrow — this is for you."
- Pet: "If you lint-roll your outfit in the car every morning, sit down." Works when: the behavior is specific enough to feel personally called out.
3. Twist the knife. Template: "[Pain] is annoying. [Pain's consequence] is the part nobody warns you about."
- Gadget: "A dead phone is annoying. A dead phone when the boarding pass is on it — that's the part nobody warns you about." Works when: the surface problem is tolerable but the downstream cost is not.
4. The quiet confession. Template: "I didn't realize how bad my [problem] was until [moment]."
- Skincare: "I didn't realize how bad my dark circles were until I saw myself on a work call." Works when: you want problem-awareness without a hard sell; it plays as a story, not an ad.
5. Permission to give up. Template: "Stop trying to fix [problem] with [common failed solution]."
- Supplement: "Stop trying to fix 3pm crashes with a fourth coffee." Works when: the market is exhausted from failed attempts — validate the fatigue first.
Family 2: Curiosity and pattern interrupt
These buy the second sentence by opening a loop the viewer has to close. Expect the highest raw hook rates in a batch — and the biggest gap between watchers and buyers, so judge them on CPA.
6. The forbidden angle. Template: "I probably shouldn't say this about [industry], but…"
- Skincare: "I probably shouldn't say this as someone who worked at a skincare counter, but…" Works when: an insider frame is plausible; the viewer stays to hear the secret.
7. The unfinished sentence. Template: "Nobody tells you that [category] actually…" (finish it after the cut)
- Supplement: "Nobody tells you that most magnesium actually…" Works when: paired with a hard visual cut at the exact moment the sentence breaks.
8. The weird object. Template: "This looks like [mundane thing]. It is not [mundane thing]."
- Gadget: "This looks like a lipstick. It is absolutely not a lipstick." Works when: the product has an unusual form factor — the reveal sells.
9. The countdown open. Template: "Three things I wish I knew before buying [category]. Number two saved me [money/time]."
- Home: "Three things I wish I knew before buying an air purifier. Number two saved me $200." Works when: listicle structure keeps mid-video retention up, not just the hook.
10. The interrupted action. Template: "Wait — before you [common action], watch this."
- Pet: "Wait — before you buy another bag of that kibble, watch this." Works when: you can name the exact action your buyer is about to take this week.
11. The oddly specific number. Template: "[Precise number] [units] later, here's my honest take."
- Skincare: "Day 47 of using this on one half of my face only. Here's the half." Works when: round numbers read as marketing; odd ones read as diary.
Family 3: Social proof and POV
The viewer trusts other viewers more than you. These hooks borrow that trust — the product is pre-validated, or the viewer is inside the scene.
12. "TikTok made me buy it." Template: "Okay, [platform] made me buy the [product]. I owe [platform] an apology."
- Home: "Okay, TikTok made me buy the sunset lamp. I owe TikTok an apology." Works when: the product plausibly has organic buzz; buying reads as catching up, not risking.
13. The reluctant convert. Template: "I fully expected to return this."
- Gadget: "I bought this to make fun of it in a video. I'm on my second one." Works when: the audience is skeptical too — leading with doubt earns the praise that follows.
14. Straight POV. Template: "POV: [specific moment your buyer lives every week, now improved]."
- Pet: "POV: it's 7am, it's raining, and your dog's paws are somehow already clean." Works when: the scene is instantly recognizable; also the cheapest format — pure voiceover over B-roll.
15. The overheard recommendation. Template: "My [trusted person] wouldn't shut up about this, so."
- Skincare: "My esthetician mentioned this twice, unprompted. That's never happened." Works when: the recommender has natural authority for the category — esthetician, groomer, trainer.
16. The volume flex. Template: "There's a reason [large number] people keep [buying/repurchasing] this."
- Home: "There's a reason this thing is sold out four months a year." Works when: you have a real proof point — keep the claim honest and generic, never a fabricated stat.
17. The comment-section hook. Template: "You kept asking about [thing in previous video], so here's the full story."
- Gadget: "Everyone in the comments asked what that thing on my desk was. Fine. Here." Works when: it borrows the feel of an ongoing conversation — reads native, not ad.
Family 4: Price and value anchor
Money hooks: the viewer does the math before you finish the sentence, and the math is the pitch.
18. The expensive twin. Template: "[Expensive alternative] costs [big number]. This does the same thing for [small number]."
- Skincare: "The clinic version of this is $300 a session. This was $34. Once." Works when: the comparison is category-level, not brand-level — naming brands invites takedowns.
19. Cost per use. Template: "This works out to [tiny number] per [day/use]."
- Supplement: "It's 40 cents a day to not feel like a zombie at 3pm. I did the math so you don't have to." Works when: the sticker price is a hurdle but the daily cost is trivial.
20. The replacement audit. Template: "This replaced [list of things] on my [counter/desk/shelf]."
- Home: "This one thing replaced three gadgets and an entire drawer." Works when: decluttering is part of the value — you are selling space back.
21. The regret-of-waiting. Template: "I waited [time] to buy this because of the price. That was the expensive part."
- Pet: "I put off buying this for a year because $60 for a dog brush felt insane. The year was the mistake." Works when: your buyer has been on the fence for weeks — this talks directly to the fence.
Family 5: Contrarian and myth-bust
Disagreement is a pattern interrupt with authority built in: attack a belief the viewer holds, and they stay — to argue, or to be relieved someone finally said it.
22. The category heresy. Template: "Unpopular opinion: [thing everyone in the category does] is a waste of money."
- Skincare: "Unpopular opinion: your 10-step routine is why your skin is angry." Works when: part of the audience already secretly agrees — you are saying the quiet part.
23. The myth flip. Template: "Everything you've heard about [category] is [timeframe] out of date."
- Supplement: "Most of what you know about protein timing is from 2009." Works when: the category genuinely moved on; follow immediately with the one updated fact.
24. "You're doing it backwards." Template: "You're not supposed to [common usage]. You're supposed to [correct usage]."
- Home: "You're not supposed to spray it on the glass. You're supposed to spray the cloth. Changed everything." Works when: a small technique correction produces a visible before/after in the same video.
25. The anti-pitch. Template: "This is not for everyone. If you [disqualifier], skip this video."
- Gadget: "If your desk setup is already perfect, genuinely, keep scrolling." Works when: you want fewer, better viewers — expect a lower hook rate and a better CPA.
Family 6: Urgency and FOMO
Deadline energy without fake countdowns — urgency built from seasons, scarcity you can stand behind, and the social cost of being late.
26. The seasonal clock. Template: "You have about [timeframe] before [seasonal moment]. Here's the cheat code."
- Skincare: "You have about six weeks before wedding season. This is the six-week product." Works when: the deadline is real and external — the calendar is the salesperson.
27. The "everyone but you" frame. Template: "Half of [group] already switched to this. The other half is still [old behavior]."
- Pet: "Half the dog park has one of these now. The other half is still untangling leashes." Works when: the product is visibly spreading; keep "half" figurative and the tone playful.
Pairing hooks with first frames
The spoken hook and first frame get judged together in the same second — and sound-off scrollers only get the frame. Working pairings:
| Hook family | First frame that fits | Avoid |
|---|---|---|
| Problem/agitation | The problem mid-happening (shedding couch, 3am ceiling stare) | The product — reads as an ad instantly |
| Curiosity/pattern interrupt | An ambiguous, mid-motion shot raising the same question | Anything answering the question early |
| Social proof/POV | Face-to-camera, mid-sentence, imperfect framing | Studio polish — kills the "real person" read |
| Price anchor | The price as burned-in text, or the expensive alternative | Text unreadable at thumbnail size |
| Contrarian/myth-bust | Face-to-camera, stopping gesture, headline caption | Soft B-roll — disagreement needs a human |
| Urgency/FOMO | Seasonal context (sun, snow, event prep) or the empty shelf | Fake countdown graphics |
Two production notes: burn the hook line into the first frame as a caption — for muted viewers the caption is the hook — and cut within the first second; a static opening frame reads as a photo and gets scrolled.
One winner, five variants: a worked example
When a hook wins, do not move on — mine it. Say this pet-category hook is your winner: "If you lint-roll your outfit in the car every single morning, sit down." Five variants, each changing exactly one variable:
- Same structure, new specific: "If there's a lint roller in your car door, your desk drawer, and your gym bag — sit down."
- Same pain, new family (POV): "POV: you wore black around your golden retriever and it's fine."
- Same pain, new family (price anchor): "I've spent more on lint rollers this year than on this. This ended it."
- Same line, new messenger: identical script, different avatar — the messenger routinely moves results more than wording, and swapping is trivially cheap with AI UGC pipelines.
- Same line, new first frame: identical audio over the covered couch instead of the face-to-camera.
Ship all five against the original. Usually one or two beat the parent — and because each changed one variable, you know exactly what carried the win.
Testing hooks systematically
A swipe file without a testing loop is a bookmark folder. The working process:
- Write 6–10 hooks per product, from at least 3 families. Rewrites within one family test wording; three families test the angle — the bigger lever.
- Keep the body constant. Same demo, same offer, same CTA — only the first 2 seconds change, or you cannot attribute the result. With the Prizmad generator that's a hook-line swap, not a full re-render; script-first tools like Arcads duplicate the script with a new opening line.
- Judge on hook rate first, CPA last. As a working rule of thumb, 25–30%+ of impressions surviving 3 seconds is competitive; under about 20%, fix the first frame and line before anything else. Then let purchases pick the winner — curiosity hooks can win hook rate and lose CPA.
- Feed winners back in. Every winning hook becomes a template in your own swipe file, plus five variants for the next batch.
The full batching structure — budgets, kill rules, naming conventions — is in the TikTok creative testing guide; the commodity-product version of the loop is in the dropshipping video ads guide.
FAQ
What are AI UGC ads?
AI UGC ads (AI-generated user-generated content ads) are video advertisements that use AI avatars, synthetic voiceovers, and automated editing to replicate the authentic, creator-style content traditionally filmed by real influencers or users. Prizmad generates AI UGC video ads from any product URL in minutes — at a fraction of the cost of hiring human UGC creators.
How long does it take to generate a video?
Most videos are ready within 5 minutes. Complex ads with custom avatars may take up to 10 minutes. You'll receive a notification when your video is ready.
Do I need video editing experience?
Not at all. AI handles everything — from script writing to final editing. Just paste a product link, choose your preferences, and let the AI do the rest.
Do I need a product link to get started?
No! A link is optional. Paste a URL from Shopify, Amazon, WooCommerce, or any product page and everything gets extracted automatically. Or skip the link and describe your product manually.
Can I edit the AI-generated script?
Yes! After AI generates a script using proven ad hook formulas, you can review and edit every line before rendering. Full control over the final messaging.
What platforms can I publish to?
Export videos ready for TikTok, Instagram Reels, Facebook Ads, YouTube Shorts, Shopify, and Amazon. All standard formats are included: 9:16 (vertical), 1:1 (square feed), and 16:9 (landscape) — all in Full HD 1080p.
How does AI UGC compare to hiring human UGC creators?
Traditional UGC creators charge $500–$2,000 per video and take 1–3 weeks to deliver. AI UGC ads from Prizmad cost approximately $3–6 per video (depending on your plan), generate in about 5 minutes, and can be produced in unlimited variations for A/B testing — in 15 languages from a single product link.
Do I own commercial rights to my AI-generated ads?
Yes. You own full commercial rights to every video ad you generate with Prizmad. Use them on any advertising platform, in any market, for as long as you want — no additional licensing fees.
Is it legal to run AI-generated video ads on TikTok, Meta, and Google?
Yes. AI-generated video ads are permitted on major ad platforms including Meta (Facebook and Instagram), TikTok, Google, and YouTube, provided they comply with each platform's advertising policies and any applicable AI-content disclosure requirements. You own full commercial rights to every video ad generated with Prizmad and can run them on any advertising platform without additional licensing fees.
How do tokens work?
Each video generation costs tokens based on complexity. The Starter plan includes 80 tokens (approximately 16–20 video ads), and the Pro plan includes 350 tokens (approximately 70–87 video ads). Tokens refresh each billing cycle and don't roll over.
Can I cancel my subscription?
Yes, you can cancel anytime from your account settings. You'll keep access until the end of your billing period. No questions asked.
Have your six hooks written? Paste your product URL into the Prizmad generator — it renders the base ad and swaps hook lines and avatars without full re-renders; the Launch plan at $19/mo covers a full first batch.