Most founders pick their ad format the same way: whichever one is easiest to make. Single image wins by default — Canva takes ten minutes, no filming, no editing. But "easy to produce" and "right for the campaign" are two different questions. Format isn't a creative preference. It's a strategic decision tied to what you're asking the campaign to do. Pick it by objective, and a mediocre creative can still perform. Pick it by comfort, and a great creative gets wasted.
What each format actually does
Single image is one static frame. Fast to make, cheap to test, and the format Meta's algorithm has the most historical data on. Recommended ratio: 4:5 for feed.
Video is motion plus sound. It holds attention longer and is the only format that works natively in Reels and Stories. Keep it under 15 seconds, with the hook in the first 3.
Carousel is 2 to 10 swipeable cards. Each card has its own image, headline, and link. It's built for showing multiple things: products, steps, angles.
That's the mechanical difference. The strategic difference is which objective each one serves.
When single image wins
1. Retargeting and bottom-funnel
People who already visited your site don't need a story. They need a reminder and a reason to act now. A clean single image with a sharp offer beats a 15-second video they've effectively already watched.
2. You're testing the message, not the creative
Single image isolates one variable: the words. If you want to know whether "free shipping" beats "30-day returns", run both as single images. Strip out motion so the copy is what's being tested.
3. Speed and budget
At ฿300/day, you can't afford a week of filming for one test. Single image lets you launch 5 angles by tonight.
When video wins
1. Cold traffic that doesn't know you
A stranger scrolling Reels won't read a paragraph. Video earns attention in motion. It stops the thumb, builds context, and creates feeling in a way a static frame can't.
2. The product needs to be seen working
Skincare texture, a gadget in use, food being made. If "showing it" sells better than "describing it", that's a video.
3. Reels and Stories placements
These placements are 9:16 and motion-native. A single image dropped into Reels looks like a mistake. If you want that inventory, you need video.
When carousel wins
1. You sell more than one thing
A 6-product carousel lets one ad merchandise your whole bestseller shelf. Meta even reorders the cards to show each person the product most likely to make them click.
2. You're explaining a process
"How it works" in 4 steps. Before and after across cards. A carousel turns a sequence into something the reader controls by swiping.
3. Warm audiences comparing options
Someone who knows you but hasn't bought is weighing choices. A carousel gives them range without forcing them onto a landing page yet.
The trap nobody talks about
"Video always wins" is the most expensive myth on this list. Founders see Meta push video, assume video means better, and burn ฿5,000 on a clumsy 30-second clip when a single image would've out-performed it at a tenth of the cost.
Video only wins when it earns the attention it costs. A boring video is worse than a sharp image: you paid more to make it and it still didn't stop the scroll. Format follows objective, not trend.
The other trap is stuffing all three formats into one ad set to "let Meta decide". Meta will, but it'll crown a winner before any of them have real data. Test formats in separate ad sets, or test them one variable at a time.
Quick reference
| Your objective | Format |
|---|---|
| Retargeting / closing warm buyers | Single image |
| Cold traffic, brand new audience | Video |
| Selling a catalog or multiple SKUs | Carousel |
| Testing copy or offer head-to-head | Single image |
| Reels and Stories placements | Video |
| Explaining a multi-step process | Carousel |
What to do next
Before your next campaign, write the objective down first (one sentence), then pick the format. Not the other way around. When you generate a blueprint, the Creative section already suggests a format based on your objective and audience temperature. If it says video and you were about to make another single image, that gap is worth a second look.
