What video should I use in the background of a Meta ad for my newsletter?

The biggest mistake almost literally every single person makes is thinking the background video just has to be recognizable to the area.

"He had videos of the local bridge everybody knows. Who cares about the bridge? I care about having fun with my family or my friends."

Why does the background matter so much?

Because it's the first thing people see. Someone scrolling has to focus to read your text, but they don't have to focus to see what's happening behind it. The order goes:

1. The background catches their eye. In a quarter of a second their brain says, "I would like to be there."
2. Now they read the text. It's about their area, and it's free.
3. If they like the text, they subscribe.

Nobody signs up because they saw people cheersing at a First Friday. The video just earns the read. If it's a hovering view of the tall buildings in your town, they may never get to the text at all.

My winning ad is a video of me walking around a little fair in my town, with text about not missing events. When I ran that same text over a generic city video, it did way worse. At one point that ad was bringing in subscribers for 18 cents. Moving backgrounds have also generally done better for me than static ones.

What should the video show?

A feeling: this looks fun, and I'm missing it. The ad is meant to stop people from scrolling, and the reason to stop is that it looks fun.

• Good: festivals, fairs, farmers markets, booths, bounce houses, live music, crowds, kids playing, people hugging.
• Weak: the resort pool, the water park, a helicopter shot over downtown, empty venues, the big statue in town.
• Best case: a recognizable place that also hosts recurring events.

Amenities are weak for a simple reason. People see the hotel pool ad all the time. Do they really need help finding the cool resort in town? They already know it. Nobody thinks, "I missed my chance to go look at the statue."

The footage also has to match the message. One client near some lakes wanted a boat going really fast. That gives off lake vibes, but the text says don't miss events, so the video has to be about events. Plenty of people use an overhead view of their town. Neither worked half as well as the fair. (Organic Instagram content is different. Do whatever you want there.)

I didn't start out knowing this. Early on I wanted to turn our water tower into a cool AI graffiti image, because a generic background doesn't register with anybody in town. I have since changed my mind.

Does the footage need to be from my town?

Not really. I've run the same nondescript downtown festival clip in about 20 locations. Bounce house, tents, people selling honey. Nobody can tell the grass is too green. I've used footage shot in Texas in Las Vegas, and you can't tell. Only the text changes from city to city.

The exception: avoid footage that's obviously wrong for the place. We rejected a desert driving clip because it would look off anywhere outside a desert town.

How should I edit it?

Quick cuts. Instead of an uncut 15-second walk through a First Friday, do three seconds of the booths with tons of people, three seconds of the band, three seconds of kids on the playground next to it. It's five videos in one, even from one event.

• Keep each clip around 3 seconds, 5 at most.
• Open with people in frame in the first two seconds.
• Film a lot and cut down. Shot horizontally? Crop it vertical. It doesn't matter.
• No fancy transitions. Cut straight from one clip to the next.

I still run the older single-clip version too, since I know it works. The new quick-cut version runs alongside it.

How do I test a new background?

Put it in the same ad set as the old one and change nothing else. I once ran three ads in one ad set where literally the only difference was the background video. Meta picked its favorite on its own, and the costs ranged from 20 to 41 cents a subscriber. The one I'd filmed myself looked much more real and moving. As usual, the more human and real, the better.

If a publisher wants to use their own footage, I launch with my proven version first for speed. It can't be whatever you want, though. The last person who sent me their own video sent stuff that just wasn't going to work. Once they have 30 to 60 seconds of a fun festival, I make a version with theirs and compare. If theirs is even similar, why not just run theirs?

Can the background be AI-generated, or should I build a library?

Real footage is best, and a library is worth it. A media pro I worked with suggested spending a day shooting local clips so new ad ideas can be built quickly from owned footage instead of stock. Almost nobody has that stockpile, and it would solve a lot of problems.

AI works for backgrounds. If I show you a 10-second drone view of a kids' baseball game, you don't know if it's real or not. What I avoid is AI people talking, pretending to be a person. For a news-focused ad running in 20 locations, where I don't have neighborhood footage for each, the background has to be generic and AI-generated. I'm also testing new concepts, like a first-person shot of a hand holding a drink while walking around a fun event. I have no idea if it'll be great, but it's a slight twist on the proven idea.

No footage at all? A few local photos animated with AI, or a fast-forward first-person night drive in 9:16 that could be anywhere, can get a generic ad launched. See using AI video in Meta ads and how to test new creative.