Do UGC talking-head videos work for newsletter Meta ads?

We haven't gotten UGC talking-head ads working better than our standard ad. One local media person I talked to was doing it, and it was pretty good, but it was worse than our fun-event video, so they paused it.

Most people doing this just keep making variations of whatever is cheapest. I could try something more creative, but if it costs me 40% more, I'm going back to what works.

That's for event-driven newsletters. For news brands, talking heads are often the best tool available.

Why does UGC beat a produced video?

"You need it to be not overproduced because you want it to look like UGC, not an ad."

A real person in a casual setting looks native to the feed. Big logos and designed graphics don't work no matter how good they look, because people can tell it's an ad. With UGC, the exact wording matters less than being a real, engaging human. People like to follow people.

How big should the person be on screen?

Smaller than you think. On one presenter ad I reviewed, she was full screen for the first couple of seconds, and it felt too long. Put the person in the corner with something interesting behind them, like local event footage.

Too small has its own problem. On another version the presenter was really tiny, and I suggested making her about 30% of the screen. I plan to use the same pattern for my own ads: me in the corner, something entertaining behind me.

How do you film one?

1. Have someone from the newsroom sit in front of a camera, the way you'd sit on a video call.
2. Have them say something simple and repeat that it's free: "Our newsletter is the best in the city. It's free. We cover this. You really got to check it out."
3. Film 5 to 10 minutes so there's plenty to cut from.
4. Put them in a corner on a green screen, with compelling local video behind them and fixed text on top.

How do you edit a long clip down?

On one 40-second creator video, the beginning was close to perfect. My edit:

• Keep the hook and the specific benefit from the first 10 to 12 seconds.
• Cut the rambling middle about event details.
• Keep the call to action: "If you're local, just sign up, it's free."
• Keep a short closer: "You'll be glad you did."

Those three pieces together should be no more than 18 seconds. Listen for the sections in a simple editor; you don't need AI to find timestamps.

Can an AI person do it instead?

Yes, with limits. A lot of operators like me use AI avatar tools right now, and they work. The version that's been working for me in local is the green-screen concept with an AI avatar really small in the corner. You can't see them well enough to notice the mouth isn't quite right.

I still don't love it, because people can still tell. It works better than you'd think. My hypothesis is that a real local person speaking with real energy and real emotion will work even better than our crappy AI ones. For the face of a brand, I want a real person. For my own brands I record myself.

You may not need AI at all. For a seniors newsletter, the obvious play is to hire an older person on Fiverr for about $40 to record the ad.

Should the ad run from a person's page instead of the brand's?

It's a test I'm setting up. Create a new brand profile in your Meta business account named after a person, with their face as the profile picture. Run the exact same ad from it. Nothing else changes.

Pair it with a short green-screen video, maybe 15 seconds, of the owner saying: I live here, I run this newsletter, I'm excited to bring you giveaways, deals and the best local info. It looks more natural and less like a company ad. Almost everyone I know who does this says it lowers their cost. I didn't make this up, but I haven't proven it on my own brands yet.

What's the catch?

Someone has to make the video. Newsrooms have to actually film it, and the background needs interesting local b-roll, which I can't supply for cities I don't live in. One lighter option I chose for a recent batch: a static image ad with a small video of me talking in the corner, instead of a full video.