The short answer
Talking-head ads, one person speaking straight to the camera, carry the highest median estimated spend of any video type on our board: $613K across 51 winners, nearly three times the board median. They also run the longest (median 162 days). Their weak spot is the top tier: only 7.8% get there, because most run as a single version for months.
The numbers behind talking-head ads
Ad Radar tags the video type of each winning Meta ad. The UGC talking head type covers ads where one person speaks to camera for most of the video. On October 9, 2026, the board held 51 of them from 38 brands, out of 2,443 live ads with $50k+ in estimated spend.
$613K
median estimated spend (highest of any video type)
162 days
median time live (longest of any video type)
18 of 51
live for more than a year
7.8%
reach the top tier (about 10% board-wide)
| Video type | Ads | Median est. spend | Median days live | Top-tier share |
|---|---|---|---|---|
| UGC talking head | 51 | $613K | 162 | 7.8% |
| Interview clip | 23 | $406K | 137 | 13.0% |
| Story narration | 108 | $398K | 100 | 14.8% |
| Expert authority | 72 | $348K | 135 | 12.5% |
| Demo of the mechanism | 131 | $259K | 158 | 10.7% |
| Voiceover listicle | 95 | $286K | 114 | 6.3% |
The talking head is the steady earner. When it works, it works for a long time and absorbs a lot of budget. A third of the 51 ads have passed a year live, against 14% of the whole board. Seventeen are above $1M in estimated spend.
So why the low top-tier share? Ad Radar's winner score weighs signals like variants and how fast engagement is growing, not only total spend. Most talking-head winners run as one version for months, with slow, steady engagement. They look like workhorses, not breakouts. For how variants factor in, see ad variants as a scaling signal.
Who uses talking heads
Pets lead by a wide margin. 20 of the 51 are pet brands, followed by beauty (10) and health (9). Pet owners trust other pet owners, and a person talking about their dog's problem in their own living room is about as credible as an ad gets.
Mostly problem-aware audiences. 27 of the 51 (53%) target people who know their problem but haven't found the fix. That is the highest problem-aware share of any video type we track. Only 9 target unaware viewers. A talking head can describe a problem the viewer has. It struggles to convince someone they have one.
Mostly real UGC. 31 of the 51 are shot as UGC selfie videos. Ten use footage with a voiceover alongside the speaker, and five combine the talking head with a product demo. See UGC ads examples for the production side of the selfie format.
The setups that work
The confession
The speaker admits something: a mistake, a lie, a habit they're embarrassed by. It is the rarest opener in the group and one of the most striking.
I need to publicly apologize to anyone who's bought Rosabella Beetroot before, because I lied.
- Est. spend
- $1.1M
- Days live
- 213
- Format
- Video, 27s
A 27-second talking head aimed at people who already know the product. The apology frame makes the viewer stay to find out what the lie was, and the answer is the offer.
The "nobody told me" discovery
The speaker found out something useful by accident and is sharing it. Variants of "here is my secret" and "nobody told me" recur across the group.
Nobody told me my mason jars could do this outside of canning season.
- Est. spend
- $945K
- Days live
- 217
- Format
- Video, 100s
The speaker frames the product as a discovery about something the viewer already owns. It turns a device pitch into a tip worth hearing.
The "why I stopped" switch
The speaker explains why they quit the usual solution. It is a talking-head version of the wrong-question angle, and it gives the viewer permission to stop doing what isn't working.
Why I stopped buying cat diffusers to calm my cat.
- Est. spend
- $556K
- Days live
- 543
- Format
- Video, 54s
Live for 18 months. The hook names the product category the viewer probably already bought, then the speaker explains what they switched to.
The personal observation
The speaker noticed something in their own life that sets up the product's reason to exist.
I moved to Japan at 26 and quickly discovered that these dudes rarely go bald.
- Est. spend
- $167K
- Days live
- 85
- Format
- Video, 97s
A first-person observation that opens a root-cause story. It sounds like something a friend would tell you, not a brand.
How talking-head ads open
Problem/Benefit Opener is the most common hook (16 of 51), followed by Question Opener (9), Warning/Shock Opener (7) and Offer-First Opener (6). The mechanism under 32 of the 51 is a problem-to-solution bridge. Story/case study is the leading angle (23).
The ads that open on a problem have the longest runs: a median of 282 days for the 16 Problem/Benefit openers. Problems don't go out of date.
For the first seconds specifically, see the first 3 seconds of a video ad and our library of problem/benefit hooks.
Lengths and landing pages
Median length is 72 seconds. Only three of the 49 with a known length are under 30 seconds, and ten run two minutes or more. Most sit between 45 seconds and 2 minutes.
Traffic splits almost evenly between dedicated landing pages (18) and product pages (17). The talking head does a lot of the presell, so a product page is often enough.
A talking-head script template
Use this for a 60 to 90 second talking head. Write the beats, give them to the speaker, and let them say it their way.
| Beat | Time | Example direction |
|---|---|---|
| Open | 0 to 5s | A confession, a "nobody told me", or "why I stopped..." in one sentence |
| Problem | 5 to 20s | What it was like before. Specific: the room, the habit, the cost |
| Turn | 20 to 35s | How they found the product. A friend, a groomer, a mistake |
| Show | 35 to 55s | The product in their hand, in use, in the real setting |
| Result | 55 to 70s | What changed, in their words. No medical or exaggerated claims |
| Offer | 70 to 90s | Price, deal or where to get it, said casually |
Setup checklist:
- Film at eye level, phone in hand or on a stand. Natural light from a window.
- Film where the problem lives. The kitchen, the bathroom, next to the pet.
- One take per beat. Joining a few natural takes is better than reading a full script.
- Burn in captions. Most people watch muted.
- Plan variants before you need them. Talking heads tend to run as one version for months. Record three alternative openers on the same day so you can test new hooks without a reshoot.
For briefing the people who'll be on camera, see how to brief UGC creators. And to read the transcripts of the 51 talking-head winners and see which setups your niche is using, filter the Ad Radar board by video type.
Figures marked as estimated spend come from Ad Radar's model of engagement on public Meta Ad Library ads. They are estimates, labeled as such, and are best used to rank ads against each other.