Facebook ad creative sizes reduce to four export ratios: 9:16 at 1080x1920 for Reels and Stories, 4:5 at 1080x1350 for the feed, 1:1 at 1080x1080 for square placements, and 16:9 at 1920x1080 for in-stream and desktop video. Make three of them, 9:16, 4:5 and 1:1, and let Meta handle the rest. Each additional hand-built export adds a file to review and upload without adding a single new idea to the account.
What are the Facebook ad creative sizes?
Here is the whole set, with what each one is actually for and whether we would build it.
| Ratio | Pixels | Placements it serves | Build it? |
|---|---|---|---|
| 9:16 | 1080x1920 | Reels, Stories, vertical in-feed video, Facebook and Instagram both | Yes, and build it first |
| 4:5 | 1080x1350 | Facebook feed, Instagram feed, Explore | Yes, this is your feed workhorse |
| 1:1 | 1080x1080 | Marketplace, right column, Audience Network, Messenger inbox | Yes, if those placements are on |
| 16:9 | 1920x1080 | In-stream video, desktop video, search results | Only if in-stream is a real line in your spend |
Two notes that matter more than the pixel counts. First, 9:16 needs roughly the top and bottom seventh of the frame kept clear of anything you need read, because Meta's own interface elements sit there and profile names, CTA buttons and captions get overlaid. Second, 4:5 is the ratio Meta itself pushes for feed and the one that occupies the most vertical space on a phone without triggering a crop. If you only ever export one file, export 4:5 and accept that Reels will get a letterboxed version of it.
Which three sizes should you actually make?
Nine-tenths of Meta spend in the accounts we see lands in three places: Reels, Instagram feed and Facebook feed. That maps to 9:16 and 4:5. Add 1:1 and you have covered the long tail of Marketplace, right column and Audience Network without a fourth render.
The order matters if you are short of time:
- 9:16. Reels and Stories inventory is cheap and vertical-native. Build here first, because a 9:16 master crops down to the others acceptably, while a 16:9 master crops up to nothing usable.
- 4:5. Do not skip this and rely on Meta's automatic feed crop of your 9:16. The auto-crop takes the centre and your caption line frequently ends up half off the frame.
- 1:1. Cheap insurance for placements you probably left on in Advantage+ placements and forgot about.
We would not build 16:9 speculatively. In-stream is a different watching context with different intent, and our 2026 fatigue benchmark puts YouTube in-stream hook rate at 22 percent against 28 percent on Meta feed and Reels, which tells you how much less attention that context starts with. If in-stream is a rounding error in your account, a 16:9 export is a file you made for nobody.
Exports are not creatives, and treating them as one is the mistake
An export is the same argument in a different frame. An ad variation is a different argument: a new hook, a different opening shot, a claim ordered another way, a proof point moved to the front. Meta's delivery system learns from the second kind and is largely indifferent to the first.
This is where the "nine sizes" habit does real damage. A team with three concepts and nine exports each has 27 assets in Ads Manager and three ideas in the market. A team with nine concepts at three exports each has the same 27 assets and nine ideas. Same review burden, three times the learning. The operational finding in our benchmark report is exactly this: throughput of unique concepts predicts campaign longevity better than the quality of any single ad, and brands shipping 15 to 50 variants a month see 3 to 5 times longer campaign lifespan than teams refreshing quarterly.
| Approach | Assets to review | Distinct arguments live | What Meta learns |
|---|---|---|---|
| 3 concepts, 9 exports each | 27 | 3 | Which of 3 hooks works |
| 9 concepts, 3 exports each | 27 | 9 | Which of 9 hooks works, across more angles |
| 3 concepts, 3 exports each | 9 | 3 | Same as row one, at a third of the review cost |
Row three is the honest comparison against row one. Six extra exports per concept bought nothing.
What does each extra export actually cost?
The render is free or close to it. The review is not. Every export you add is another file where someone has to:
- watch it end to end at phone size, not scrubbed on a desktop
- check the safe zones for that specific ratio, because the crop moves your text
- confirm the caption burn-in did not clip
- name it in a way your reporting can group later
- upload it and attach it to the right ad
Call it four to six minutes of attention per asset if the person doing it is honest. Nine exports across five concepts is 45 assets, so most of a working day gone to files that carry three or four ideas between them. That day was the one you were going to spend writing new hooks.
There is a subtler cost. Reporting fragments. Nine exports of one concept spread the same concept's data across nine rows, so each row takes longer to reach a readable sample and the concept looks weaker than it is. Fewer exports means fewer, denser rows, and denser rows mean you can kill a loser sooner.
When more than three exports is the right call
Three cases, plainly:
- In-stream is a genuine placement for you. Then 16:9 is not optional, and it usually wants a different edit as well as a different frame, because you have a pre-roll skip to survive rather than a scroll to interrupt.
- You run a placement with hard text overlays. Some retail and Marketplace layouts sit differently over the frame. If your brand asset is being covered, build the ratio rather than fighting the crop.
- You are running paid social and organic from the same master. Organic Reels and a 4:5 feed post have different caption conventions, and reusing a paid export with a burnt-in CTA reads badly on an organic profile.
Outside of those, the fourth export is a habit rather than a decision.
How we build this in Genyad
Genyad is our product, so treat this as disclosure rather than neutral advice. It takes a library of video you already own, tags every clip, and writes each variation as a fresh script and shot selection rather than a re-cut of one timeline. Every variation comes out in 9:16, 4:5, 1:1 and 16:9 from a single credit, which is the point of multi-format export: the ratios stop being a decision you spend time on, so the only thing left to decide is how many different arguments you want in market. If you want to see the ratio set against a real concept, the Meta ad creative generator page shows the output shape.
What Genyad does not do is worth stating because it is directly relevant to a post about sizes. There are no static banner formats, so if you need a 1200x628 image ad or a display set, this is the wrong tool and a design suite like Canva is the right one. There is no product-URL import and no product-feed or CSV template rendering, so it will not fan a catalogue out across sizes for you. Output is 1080p on the self-serve plans, with 4K on Enterprise, and there is no watermark on any plan including the free one.
Frequently asked questions
What is the best size for a Facebook video ad?
9:16 at 1080x1920 for Reels, Stories and vertical in-feed placements, and 4:5 at 1080x1350 for the Facebook and Instagram feed. If you can only produce one file, make it 4:5, because it fills the most feed real estate without a crop and downgrades to other placements more gracefully than a horizontal master does.
Do I need to make a separate file for every Facebook placement?
No. Three exports, 9:16, 4:5 and 1:1, cover almost all Meta inventory that most accounts actually buy. Meta will crop or letterbox where a ratio is missing, and that trade is usually better than spending the review time on a fourth and fifth export.
Does Meta penalise letterboxed or cropped creative?
Not as a policy matter, but performance drops in practice because black bars shrink your visible frame in a feed that rewards screen coverage. The bigger risk is the auto-crop cutting through burnt-in text, which is a legibility problem rather than a delivery one, and it is the reason 4:5 is worth building rather than deriving.
How many creatives should I run instead of extra sizes?
Most active campaigns want 8 to 20 live variations in rotation at any time, and our benchmark report links shipping 15 to 50 variants a month to 3 to 5 times longer campaign lifespan. Getting there means cutting exports per concept, not adding them.