Campaign Structure and Creatives That Scale on Meta Ads
Learn how to structure campaigns, how many creatives to run per ad set, and the small tweaks that give you big leverage on scale.
What changed in Meta Ads campaign structure
The campaign structure that scales on Meta Ads today isn't stacking 20 proven creatives inside one ad set and letting it rip. What works now is the opposite: fewer creatives per ad set, the same body copy with different hooks spread out, and letting the platform deliver across audience slices. If you keep running the old model, you're watching delivery stall without understanding why.
The logic is simple. Meta's algorithm got a lot smarter at reading who converts. When you throw too many creatives in one place, it gets confused deciding who to deliver to. When you feed it the same body with different hooks, it already knows which audience bites and just changes the front door.
How many creatives per ad set?
Today the number that works is 4 creatives per ad set, max. It's not a rule from Meta's playbook, it's what delivery shows in practice when you run volume.
The old standard was the classic structure test: 1x1x2, 1x3x1, with every proven creative in the same ad set. You'd end up with a pile of ads fighting for delivery inside their own campaign. Looked like scale. In practice, it cannibalized.
The mindset shift also applies to your launch volume. People used to start with 50 campaigns in a hard push right off the bat. Today you start with 3. You test, read what caught, and only then open up. Scale comes after the read, not before.
Same body, different hooks: why does this deliver better?
This is the game changer. Instead of completely different creatives per ad set, you run the same ad body and only change the hook in each ad set.
Here's how it works: Meta already identified which audience converts with that body. By changing only the hook, you don't force the algorithm to relearn everything from scratch. It keeps the conversion read and tests which front door opens best for each slice of people. Delivery gets cleaner and your cost per acquisition responds.
Less brand new creative, more intelligence about what already performs.
Low-effort variations that give huge leverage
When you hit on a winning creative, the biggest mistake is running off to test new ones. The play is to make variations of the winner before anything new.
The variations that pay off most and cost almost nothing to produce:
- Different soundtrack on the same video
- Re-editing the first 3 seconds, where the hook either holds or loses the person
- Changing the hook recipe while keeping the body
- Swapping the expert who appears in the creative
- Tweaking the caption, including the color
Caption color sounds like a silly detail. It isn't. A creative that was scaling with black caption on a white background jumped from $10k to $15k per day just by switching to blue text. Same creative. One color.
The cycle is this: hit on a creative, make variations of it first, and once it takes off again you go back to testing new stuff. Before you have the next winner, you squeeze everything out of the current one.
When you run this cycle across several accounts at once, the bottleneck stops being creative and becomes setup. Launching 4 hook variations in 3 ad sets, replicated across 5 BMs, by hand, is a late night lost in Ads Manager. That's exactly where DirectAds' standardized naming and configuration across accounts removes the friction: every variation launches consistent, with no naming errors or swapped targeting.
Is it worth following the platform's recommendation?
It is. Meta sends recommendations and a big chunk of the market ignores them out of stubbornness. Mistake.
There was a case of a direct recommendation: add more variations of this creative so it performs, because CPA is rising. I did. It delivered a lot more. It's not magic, it's the platform reading a signal you don't see in the dashboard.
The idea that Meta is against the advertiser makes no sense. Their revenue comes from ads running. If the algorithm suggests something, it's because it spotted a pattern in the data. Why row against it?
Automated campaigns perform well consistently today. The media buyer decides the strategy, the offer, and the read. The platform handles delivery. Respecting that split saves you a pointless fight with the system.
Image or video: which cuts cost more?
Depends on what you need: a scale ceiling or a low cost of entry. And the best move is not to choose.
Abroad, video traffic got expensive. A lot of operators moved to image, which is much cheaper, pulls more people into the VSL, and lets the VSL do the converting. The image works as the hook and the text above it does the qualifying. Cheap to produce, easy to test.
Video scales much more. The ceiling is sky high, it can handle heavy spend and lasts longer at scale. The problem is the cost and how hard it is to nail.
Image has a lower ceiling and dies faster when you force scale. On the flip side, you can make 10 variations in a minute. When video isn't coming together, image is what saves the operation, each one holding around $10k a day while you find the next winning video.
Run both. Video for the ceiling, image for speed and survival.
Takeaways
- Cap it at 4 creatives per ad set and start scaling with a few campaigns, not 50 out of the gate.
- Run the same body with different hooks per ad set: the algorithm already knows who converts and just changes the front door.
- When you hit on a creative, exhaust the low-effort variations (soundtrack, first-3-seconds hook, caption color) before testing anything new.
- Follow the platform's recommendation when it asks for more variations. The signal comes from data you don't see.
- Keep video and image running together: video for the scale ceiling, image to cheapen entry and save the operation on bad days.
Frequently asked questions
How many creatives should I put per ad set on Meta Ads?
4 max. More than that makes ads fight for delivery inside their own ad set and confuses the algorithm when deciding who to deliver to.
Why use the same body with different hooks?
Because Meta already identified which audience converts with that body. By changing only the hook, it doesn't relearn from scratch and tests which hook opens best for each audience slice. Delivery gets more efficient.
Is it worth following Meta Ads recommendations?
Yes. The recommendations come from reading data the dashboard doesn't show. When Meta asks for more variations of a creative because CPA is rising, acting on it usually increases delivery.
Does image scale as much as video?
No. Image has a lower ceiling and dies faster at scale, but it's much cheaper and faster to produce. Video has a sky-high ceiling. The ideal is to run both in parallel.
