On this page
- The steps
- Real supplements video ads
- Why a hybrid ad works better than all-AI
- What long-running supplement ads open with
- Set up the hook in Kling 3.0
- Keep the presenter consistent across clips
- Cut to real demo footage and finish in an editor
- Test three hooks and stay honest
- Prompts to copy
- 15-second UGC ad plan
- Mistakes to avoid
- Questions
Real supplements video ads to model yours on
9,890 video ads from 1,180 advertisers in the Meta Ad Library, read October 5, 2026. The median one ran 25 days, and an ad still paid for after months is the closest public sign that it works. Live ads first.
- Ran under a day
- Ran 2 days
- Ran 5 days
- Ran 3 days
- Ran under a day
- Ran 1 day
- Ran 1 day
+9,878 more supplements video ads
Hooks, landing pages and run length for each
- Ran 7 days
- Ran 9 days
- Ran 7 days
- Ran under a day
Every creative, with its copy, format and run length
The openers that ran longest
Watch this video about 12 herbs that cleanse the whole female system; help shrink fibroids and cysts and ease…
FibroBlend · running 2,063 daysBoost your wellness journey with Dermapro's exclusive essentials!
Dermapro Lebanon · running 739 days🔥 25,000+ Customers Swear by Intellectual Power!
Intellectual Power · running 620 daysCreatine is a game changer mentally and physically...
Braven Grant CEO · ran 563 days
Why a hybrid ad works better than all-AI
Kling 3.0 is strongest at the part of UGC that is hardest to film: a believable person looking into a lens and talking. It is weakest where an ad cannot afford errors, such as your exact packaging, label text and how the product really works. So split the job. Generate the selfie hook in Kling. Film or photograph everything that has to be accurate.
A 15-second ad has four beats. The first is a 0 to 3 second hook that names the problem. Next comes a 3 to 6 second beat that shows the product and what it does. Then a 6 to 12 second close-up of the product in use. Finally a 12 to 15 second product shot with one call to action. Only the first beat needs a generated presenter. The rest can be your own phone footage.
This structure also makes iteration fast. Because the hook is its own clip, you can regenerate a weak opening without redoing the whole ad. Keep the style of UGC: casual delivery, ordinary lighting, quick cuts and a slightly imperfect frame. Do not present an AI actor as a real customer giving a testimonial.
What long-running supplement ads open with
In the supplements category of the Meta Ad Library there are 9890 video ads from 1154 advertisers, and the median run is 27 days. The openers below ran far longer. Braven Grant CEO ran "Creatine is a game changer mentally and physically..." for 563 days. Blackstone Labs ran "Remember, 99% of us are dealing with something." as the start of its opener for 432 days.
MuscleMax ran "Boost Your Stamina & Performance" for 502 days with a Shop now button. Mizu Lab ran "Transform your plain water into your favorite boba drink flavors" for 334 days, also with Shop now. The pattern is one plain sentence, one idea, said the way a person talks. Braven Grant CEO and Blackstone Labs used no button at all, so the opener carried the ad.
Translate that into your Kling hook: one spoken line, one idea, short enough for a 4 to 5 second take. Pick a statement the presenter can say in one breath, like the Braven Grant CEO line, or a transformation line like the Mizu Lab one. Skip long lists of claims. The Sprites ads library shows real ads from a category, and the supplements video ads page is a good place to see more openers.
Want supplements video ads like these for your product?
The supplements video ads that kept running, read from the ad libraries. Every change waits for your approval.
Set up the hook in Kling 3.0
Open Kling, choose VIDEO 3.0, then Image to Video. Upload a vertical reference image of your presenter: chest-up, face clearly visible, even light, framed like a selfie. The reference sets the look of the person, the room and the framing. Choose an image that already feels like phone footage rather than a studio portrait.
Start with a single 4 to 5 second shot. Turn on Native Audio if the presenter needs to speak, then put the exact spoken line in quotation marks in the prompt. Describe the camera the way a phone would capture it: front-facing, arm's length, slight handheld movement, one continuous take, no zoom. Say what to leave out: no music, captions or on-screen text, because you will add those in your editor.
Keep the line short enough to fit the clip. One sentence of about 8 to 12 words fits a 4 to 5 second take with a small pause. Generate a few versions and judge each on three things. Does the mouth match the words? Does the face stay stable? Does the delivery sound casual rather than scripted? Regenerate only the hook until one passes.
- Use a vertical 9:16 reference image, not a cropped landscape one
- Pick everyday settings: kitchen table, hallway, car seat
- Write one spoken line per clip
- Add one small physical beat, like an eyebrow raise or a pause
- Ask for quiet room tone and no music
Keep the presenter consistent across clips
If you generate more than one clip with the same presenter, reuse the same reference image or bind a character Element. That keeps the face, hair and clothing steady between the hook and any follow-up reaction shot. Using a new reference for each clip is the fastest way to make an ad look stitched together.
Kling 3.0 also supports Custom Multi-Shot for videos up to 15 seconds. It suits a sequence such as selfie, hands close-up, selfie again, where the presenter stays the same. The trade-off is control: when one shot goes wrong, you regenerate the whole sequence. For testing, separate clips are safer. Move to multi-shot once a hook has proven itself.
Whichever route you pick, write each shot as a camera setup plus one action. Name the framing, such as chest-up selfie or hand close-up, the one thing that happens, and the sound. Avoid stacking three actions into one shot. Single-action shots are easier for the model to follow and easier for you to cut.
Cut to real demo footage and finish in an editor
After the hook, cut straight to a real phone clip of the product doing its job. Film it vertically in ordinary light: a hand opening the tub, a scoop into a shaker, the drink poured. Real footage shows your true packaging, label text and how the product works. Those are the details AI is most likely to get wrong.
If you do generate product footage, upload your product photos as references and inspect the exported clip frame by frame. Look for warped logos, changing details and inaccurate interactions. Kling's reference tools help with consistency, but you still have to review the output. Any label text that looks off is a reason to swap in real footage.
Add captions, the call to action and the end card in your editor, where you control spelling, placement and timing. Check the Meta ad specs and TikTok ad specs for safe areas before you export. Put one call to action in the last 3 seconds. In this category the most common buttons are Shop now, Learn more, See details and Order now, so match your on-screen line to the button.
Test three hooks and stay honest
Run your first test by keeping the demo and call to action identical and swapping only the opening. Make three hooks: a problem question, a product-in-action shot, and a direct benefit line. Use the same audience, budget and placement. When only the first 3 seconds change, the result tells you which opener earned the attention.
Keep the winner and write three new variations of it. Change one thing at a time: the line, the setting, or the delivery. The long-running ads above show that a simple opener can hold for hundreds of days once it works, so refine rather than restart. Save the reference image and prompt of every winner so you can recreate the look.
Stay honest. Keep the UGC style, but do not present an AI actor as a real customer giving a testimonial. For supplements, keep generated lines to routine, taste and convenience, and avoid health or cure claims, which platforms review closely. See UGC ads for more formats that keep the creator feel without faking a customer.
Prompts to copy
Paste each one as written, then change the product, the setting and the light to match your own footage.
Hook: problem question selfie
Vertical 9:16 front-facing smartphone selfie video. Match the presenter in the reference image. They sit at an everyday kitchen table in soft window light, framed chest-up at arm's length. Slight handheld movement, one continuous take, no zoom. The presenter looks into the lens and says in a casual, slightly amused tone: "Still forgetting your creatine every single morning?" A small eyebrow raise on "every single morning," then a brief pause. Quiet room tone. No music, captions, or on-screen text.
Hook: product in action
Vertical 9:16 handheld phone-style video. Use the reference image of the product tub. Kitchen counter, soft morning light. A hand twists off the lid, drops a scoop of powder into a shaker bottle, then shakes it. Slight handheld sway, one continuous take, quick natural movements. Do not invent any label text or logos. Sound: lid click, scoop, shaker rattle. No speech, music, captions, or on-screen text.
Hook: direct benefit selfie
Vertical 9:16 selfie video. Match the presenter in the reference image, standing in a bright hallway and holding a plain shaker bottle at chest height. Handheld, one continuous take, no zoom. They look into the lens and say in a relaxed, direct voice: "One scoop, ten seconds, done before my coffee." A small nod on "done." Natural room tone. No music, captions, or on-screen text.
Demo: close-up pour
Vertical 9:16 close-up with a smartphone look. Use the reference image of the shaker bottle on a wooden table. A hand picks it up, twists the cap open, and pours the drink into a glass of ice. Shallow depth of field, soft side light, slight handheld movement, one continuous take. Sound: liquid pouring, ice clinking. No speech, music, captions, logos, or on-screen text.
Custom Multi-Shot: 10-second sequence
Vertical 9:16 UGC sequence, about 10 seconds, same presenter as the reference image in every shot. Shot 1, 0 to 4 seconds: chest-up selfie at a kitchen table, presenter says "Okay, here's my morning routine." Shot 2, 4 to 7 seconds: handheld close-up of their hands mixing a drink in a shaker. Shot 3, 7 to 10 seconds: selfie again, presenter smiles and shrugs. Natural room tone. No music, captions, or text.
End: product shot for call to action
Vertical 9:16 product shot using the reference image of the product tub. It sits on a clean kitchen counter in soft daylight. The camera makes a slow, gentle push-in with slight handheld drift, then holds steady for the last second. Keep the top third of the frame empty so captions can be added later. Do not invent label text or logos. Soft room tone only. No speech, music, or on-screen text.
15-second UGC ad plan
| Time | Shot | Source | Kling setting |
|---|---|---|---|
| 0 to 3 sec | Selfie hook that names the problem | Generated in Kling | VIDEO 3.0, Image to Video, single 4 to 5 second shot, Native Audio on |
| 3 to 6 sec | Product and what it does | Real phone footage, or generated from product photos | Product photo as reference if generated; review for warped labels |
| 6 to 12 sec | Close-up of the product in use | Real phone footage preferred | Optional hand close-up clip with no speech |
| 12 to 15 sec | Product shot with one call to action | Real footage or photo, text added in editor | No text in Kling; add captions in your editor |
| Hook test | Problem question, product in action, direct benefit | Three separate Kling clips | Same reference image or character Element for each |
Mistakes to avoid
- Generating the whole ad in one Kling prompt, so a weak hook forces you to redo everything.
- Letting the AI reproduce your packaging and label text instead of using real footage.
- Using a polished studio portrait as the reference image, which kills the phone-selfie feel.
- Asking Kling to render captions or on-screen text instead of adding them in your editor.
- Changing the demo and the opener at the same time, so you cannot tell what won.
- Presenting an AI actor as a real customer giving a testimonial or making health claims.
Rather have the videos made for you?
Director by Sprites is a full creative team for your startup. Every Monday you get finished, on-brand video ads, reviewed by a human. You never write a prompt.
Apply to Director by Sprites →Questions
Can Kling 3.0 make a UGC style video with the presenter talking?
Yes. Use VIDEO 3.0 with Image to Video, upload a vertical image of your presenter, and enable Native Audio. Put the exact spoken line in quotation marks in your prompt. Start with a single 4 to 5 second shot.
How long can a Kling 3.0 video be?
Kling supports Custom Multi-Shot for videos up to 15 seconds. For UGC ads, separate 4 to 5 second clips are often easier to iterate on, because you can regenerate just the hook.
How do I keep the same presenter across several Kling clips?
Reuse the same reference image for each clip, or bind a character Element. Avoid swapping in a new reference image per shot, since that changes the face and clothing.
Should I generate the product footage in Kling too?
It is safer to film the product on your phone. If you do generate it, use your product photos as references and check the export for warped logos, changing details and inaccurate interactions before publishing.
What should the first three seconds of a UGC ad say?
One plain sentence about one idea. The longest-running supplement openers in the Meta Ad Library, such as Braven Grant CEO at 563 days and MuscleMax at 502 days, are short, direct statements. Test a problem question, a product-in-action shot and a direct benefit.
Is it okay to use an AI presenter in an ad?
Keep the casual UGC style, but do not present an AI actor as a real customer giving a testimonial. For supplements, avoid health or cure claims in the spoken line and stick to routine, taste or convenience.