State of UGC Ads: Q4 2026

Show it, don't say it. We tracked 3,941 video ads from 193 top DTC brands and measured one thing: which ads were still running after 90 days. Ads that open on a demonstration were 1.6x as likely to survive as the average ad. Ads built around AI characters almost never did.

Brands kill losing ads fast. An ad that's still live after three months is almost always paying for itself. So survival is the closest public read we have on what works, without anyone's private dashboard.

On average, 26.7% of ads made it to 90 days. Here's how individual creative choices moved that number.

The five takeaways:

1. Lead with proof, not pain. Demo hooks survive at 43.2%. Pain-point callouts, the most common hook in the dataset, come in below average.
2. Say less, show more. Light-claim and comparison ads outlast clinical and "supports" language.
3. Real footage wins; AI is the multiplier. AI voiceover over real b-roll performs well. AI characters on screen do not.
4. Urgency beats reassurance. Time-bound offers outlast money-back guarantees roughly 3 to 1.
5. Health is harder. Health and wellness ads burn out faster than average, and they play by different rules.

1. Hooks: the most popular opener is one of the weakest

Pain-point callouts ("Tired of...?", "If you struggle with...") are the default hook in DTC. They appeared in 1,453 ads, more than any other. They survived at 20.2%, well under the 26.7% average.

The hooks that lasted showed something in the first second instead of describing a problem:

What it looks like. Rypt's grip ad runs no voiceover at all: a creator lifts dumbbells with and without the grips, set to music. It has been live for 591 days. Grazly presses blotting paper onto its beef jerky to prove the fat is natural, and that ad has run 327 days.

Neither ad says what problem it solves. The demo makes the problem obvious.

Why confessions flop. "I have to admit something..." hooks feel native on TikTok but rarely sustain on paid. The setup takes too long and the payoff is usually a soft product mention.

2. Claims: the more you explain, the faster it dies

The ads that lasted longest made one of two moves. They either compared themselves directly to the alternative, or they made almost no claims at all and let the experience speak.

Comparison works because it does the thinking for the viewer. Lettuce Grow's founder holds bagged supermarket lettuce next to a head he just harvested and asks which one you'd rather eat. It has been live for 397 days. Lucent Globe's laundry sheet ad opens on what goes wrong with liquid detergent and has run for 328.

Clinical language underperforms. "Clinically studied" and "supports healthy..." read as compliance copy. Viewers scroll past it, and the ads get cut. That doesn't mean skip the science. It means put the proof on screen (a result, a side-by-side, a credentialed person) instead of in the script.

3. Who's on camera: founders and real footage beat the talking head

The classic UGC format, a creator talking to camera, made up 64% of ads. It survived at 24.5%, slightly below average. Two alternatives did much better: founders on camera (35.0%) and voiceover laid over product footage (35.0%).

Founders win when they demonstrate, not when they pitch. Lucent Globe's founder opens with "I need to publicly apologize to anyone who's ever bought these dishwashing sheets," holding a dirty plate. It ran 389 days. Energybits and Mid-Day Squares both run founder explainers past 290 days.

Voiceover over footage is the quiet workhorse. Grazly's beef slab ad is just close-ups of the product pulled apart while a voice lists two ingredients. It has been live for 401 days. Noom's longest runner pairs a voiceover with an on-screen weight quiz and has run 558 days.

AI: multiplier yes, mascot no. This is the clearest split in the data:

• AI voiceover over real b-roll: 38.9% survive to 90 days.
• AI cartoon or animated characters: 7.3%, the lowest of any pattern we measured.

Viewers accept AI that makes real footage go further. They reject AI that replaces the human. Our read: keep real creators and real product on screen, and use AI to multiply hooks, voiceovers and b-roll around them.

4. Format and finish: the details that extend an ad's life

Unboxing and first use is the longest-lived format we tracked (45.4%). It's a demo with built-in curiosity. Lucent Globe pulls clean laundry from the machine, says "I can't believe how amazing my clothes smell," then opens four new scents. Cove drops a stack of shorts on a warehouse floor and tries a pair on. Both ran 290+ days.

Stack the hooks, then pay off. Ads that open with several quick hooks before delivering the answer survived at 35.3%, the best of any story structure. Problem-to-solution, the most taught structure in UGC, came in at 22.4%. Rorra's water filter ad is a clean example: a fitness creator rattles off three reasons before showing the unit, and it has run 482 days.

Captions: less is more. Ads with no burned-in captions survived at 36.0%. Auto-generated AI captions, now the default in most editing tools, came in at 22.6%. A bold bottom-third caption (29.4%) is the safe middle ground.

Music: let it carry the ad. Music-only ads with no voiceover survived at 32.5%. Trending sounds came in at 28.3%, about average.

CTAs: urgency over reassurance. A time-bound offer ("last day of the sale") survived at 33.3%. Money-back guarantees landed at 11.6%. Goli's "Today is the last day of the biggest sale of the year" ad ran 377 days, which tells you the sale ends but the ad keeps working.

5. Health and wellness spotlight: a harder category with its own rules

Health and wellness ads burn out faster than the market overall. Only 19.2% made it to 90 days, against 26.7% across all categories. Regulators, skeptical buyers and crowded feeds all shorten the runway, and some of the general rules flip.

Credentials help; citations hurt. A licensed professional on camera lifts survival above the health average. A script that cites studies drops it to the bottom. Let the expert's face carry the authority instead of the footnote.

Founders flip. Across all categories, founders are a top format. In health they're near the bottom. Buyers trust a doctor or a real customer over the person who profits from the sale.

What still works everywhere: real footage with voiceover, authority-led hooks, and experience over claims. For GLP-1 and telehealth brands, that means provider-POV and patient-journey creators, with the result shown rather than promised.

6. Five briefs to steal for next quarter

These are the formats we'd put into any DTC account's Q1 test plan based on this data.

1. The silent demo. 15 to 20 seconds, no voiceover, music only. Open on the product doing its job in the first second. One bold bottom-third caption with the outcome.
2. The side-by-side. Your product next to the thing your customer uses today, same shot, same light. Ask the viewer to choose. No claims needed.
3. The founder apology. The founder admits what the category gets wrong (or what they got wrong), then demonstrates the fix on camera.
4. The hook stack. Three rapid hooks in the first five seconds, each a different reason to care, then the reveal. Cut five hook variations from the same body.
5. The voiceover over real footage. Shoot 20 to 30 clips of real product b-roll once. Then use AI voiceover to produce many scripts over the same footage and test angles cheaply.