Most Meta ads fail before they're ever made.
They fail at the brief. They fail because the person sitting down to create has no framework to operate inside, so they default to instinct, copy what they saw last week, and produce creative that has no specific reason to exist.
The result is what we call junk-food creative. It fills the account. It eats budget. It looks like work. But it does nothing nutritious for the algorithm, because nothing about it was designed to win, it was designed to ship.
If the research from Part 1 of this series is the raw material, ad creation is where it gets turned into something Meta can actually use. And there's a process for that. Four layers, in order, every time.
Skip a layer and the ad gets weaker. Reorder them and the ad gets confused. Hit all four in sequence and the ads start carrying themselves.
This is the framework we use to brief every creative that goes into a YMP-managed account.
Why Creativity Without Constraints Isn't Creativity
Before the framework, the most important idea in this entire post: constraints don't kill creativity. They make it possible.
If you sit down at a desk and someone says "write a book," you'll stare at the blank page for an hour. Where do you start? What's it about? Who's it for? What genre? How long? You have infinite options, so you have no traction.
If the same person says "write a 60,000-word horror novel about a monster in a remote town", suddenly your brain is working. You know the boundaries. And inside those boundaries, your creative freedom actually expands, because every decision you make is anchored.
That's what the framework below does for ad creation.
Without it, you'll get lost in possibility, especially if you're newer to paid media. You'll think you're being creative when actually you're just generating ideas you'll never execute, because you have no system for filtering them. With it, you always know what decision you're making and why.
Four layers. Angle, format, trend, and structural test. Every brief runs through all four.
Layer One: The Angle
The angle answers a single question: why would this customer, specifically, buy from us, specifically, right now?
Not "why is our product good." Not "what are our features." Why would this person, in the emotional state they're in today, make the decision to convert?
This is where the customer research from Part 1 stops being a document and starts being an ad. Every angle should trace back to one of four sources:
1. The post-purchase survey. If you've been running one, you already have angles in your customers' own words. The reasons they actually bought, written by people who actually bought. There is no better creative input than this. Most agencies have never read theirs. Most brands don't run one.
2. The ideal customer profile. The ICP gives you the emotional state. The angle gives that emotional state a voice. If your customer's dominant feeling is frustration with how long they've been hiding a problem, the angle isn't "our product works fast." The angle is "you don't have to keep hiding this."
3. Competitor review mining. Read the reviews on your competitors' sites. Read the comments on their ads. The places where their customers are frustrated are angle opportunities for you. The places where their customers are delighted are angle threats, features you need to either match or reframe.
4. The ads library. If a competitor's been running a specific testimonial ad for six months, it's not because they like it. It's because it's working. The angle inside that ad is a market-validated angle you should be considering.
Most advertisers think they already know why people buy from them. They're often half right, but they've identified the second most popular reason, not the dominant one. The research isn't extra work. It's the only way to find out which angle is actually the strongest.
A good angle is so specific that the customer reads it and thinks: this was written for me. A weak angle could describe any product in the category.
Format is the simplest layer. It's also where most creative briefs go wrong, because people pick the format before they've picked the angle, which is exactly backwards.
The format choices for Meta in 2026:
- Video (single-shot, edited, or short-form native)
- Static image (single creative, no animation)
- Carousel (multi-card swipe-through)
- Catalogue / DPA (dynamic product ads for ecommerce)
But format alone isn't a brief. Format only becomes useful when you cross it with an execution style. That's where it unlocks.
Take video. "Video" is meaningless. UGC-style video is a brief. Founder-led video is a brief. Testimonial-led video is a brief. Product-demo video is a brief. Same format, four completely different ads.
This is where the framework becomes a pick-and-mix. You're not picking a format. You're picking a combination:
- Video + UGC
- Video + founder
- Video + testimonial
- Image + Us vs Them
- Image + press validation
- Carousel + testimonial
- Carousel + step-by-step mechanism
- Catalogue + offer overlay
Every combination is a different ad. The moment you stop seeing format as one decision and start seeing it as two, format plus execution style, the number of creatives you can brief from a single angle multiplies.
This is also where the 5 Levels of Creative Diversity framework comes in. The accounts that scale are the ones running diverse formats across diverse executions, all anchored to a small set of strong angles. The accounts that plateau are the ones running 12 variations of the same video.
Layer Three: The Trend
This is the layer most advertisers skip. It's also the one that increasingly separates ads that win from ads that just exist.
Meta is a content platform. The ads that perform best are the ones that don't feel like ads, the ones that borrow the visual codes, the editing pace, the audio choices, the on-screen text styles of what's working organically right now.
Trends aren't trends because they're cool. They're trends because they're earning attention. And attention is the only resource Meta is selling.
The question to ask at this layer is: is there a current organic format or convention that could credibly carry our angle and offer?
The keyword is credibly. A clumsy trend-jack, using a trending sound or format that has nothing to do with your product, is worse than no trend at all. It signals desperation. It tells the viewer "we don't actually have anything to say, so we're hoping the algorithm carries us."
But a well-mapped trend, where the organic format and your angle align naturally, is one of the highest-leverage creative decisions you can make. You get the conversion mechanism of a direct-response ad with the engagement profile of organic content. That's the unlock.
For context on why this matters: 3.2 billion people use the Facebook family of apps every day, and 3.5 billion reels are shared between them. The platform is short-form video-native now. Ads that act like organic content belong on the platform. Ads that act like ads are increasingly being trained out of the feed.
This doesn't mean every ad should be a reel or a trend-jack. It means every ad brief should explicitly answer the question: is there a trend lens here that strengthens this angle, or does this angle work better as a clean direct-response play? Both are valid. Skipping the question is not.
Layer Four: The Structural Test, Does This Belong In A Conversion Campaign At All?
Before any ad goes into a conversion campaign, it has to pass one final test.
If the ad has no clear call-to-action, or if it's overtly brand- or lifestyle-driven, it doesn't belong in a conversion campaign.
Put it in an awareness campaign. That's what awareness campaigns are for. Don't ask Meta's conversion algorithm to optimise for purchases off the back of an ad that doesn't ask for one, it'll burn budget trying to find people who match a pattern the ad itself doesn't support.
This sounds obvious. It is not.
We've audited hundreds of accounts where conversion campaigns are full of self-congratulatory brand films, glossy lifestyle reels, and "about us" content with no CTA, no offer mention, and no mechanism for the algorithm to associate the ad with a buying action. The campaigns then underperform, and the operator concludes their conversion strategy is broken.
The strategy isn't broken. The casting is wrong. The right ad is in the wrong campaign.
A conversion ad needs three things at minimum:
- An angle that connects to a buying decision (not "we're great", why would someone act today)
- A clear offer or product mention (Meta needs to know what's being sold)
- A direct call-to-action (the viewer needs to know what to do next)
If any of those are missing, put it in awareness, where the success metric is reach and engagement, not purchases. Don't mix the two.
The Final Filter: Would You Convert On This?
Once a creative is briefed, built, and ready to ship, there's one last test before it goes live.
Look at the ad. Strip away your biases, the fact that you wrote it, that you know the product, that you've seen the brand a thousand times. Imagine you're seeing this ad cold, in a feed, between a friend's holiday photos and a meme.
- Would you stop?
- Would you watch the full thing?
- Would you click?
- Would you buy?
If the answer to any of those is no, the ad isn't ready. It might be close. It might be 80% there. But "close" gets killed in the feed, because every ad you're competing against is also close, and the ones that win are the ones that get all the way there.
This is the question Ryan asks constantly inside our team, and it's the question that separates creative that ships from creative that performs. Most ads fail this test. The ones that pass it tend to win.
A Note On Volume
If you've followed the framework so far, you might be thinking: this is a lot of work to brief one ad.
It is. And that's the point.
But here's what most advertisers get wrong about volume: they assume the goal is more ads. The goal is more good ads. Five briefed-properly creatives in a conversion campaign will outperform fifty briefed-lazily ones, every single time. The algorithm consolidates spend onto its winners, and if you don't have winners, having more losers doesn't help.
The trade-off is real. We'd rather you take longer to produce three or four genuinely thought-through ads in a creative cycle than rush twelve thin ones into the account.
That said, once you've internalised the framework, the work compounds. The research from Part 1 gives you angles for months. The format-plus-execution combinations multiply each angle. The trend layer keeps the creative current. The structural test stops you wasting budget on the wrong ad in the wrong campaign.
After six months of operating this way, briefing a new ad takes minutes, not hours, because every decision is anchored, and most of the thinking has already been done.
What This Looks Like In Practice
Pulling the four layers together, a real brief looks like this:
Angle: Customer is a 35-year-old founder of a service business, frustrated that her last agency optimised toward leads that never closed. Dominant emotion: distrust. Buying trigger: the agency she's currently with just lost her £8k in misspent budget last month.
Format + execution: Founder-led video. 45 seconds. Talking to camera. No B-roll until the proof points.
Trend layer: Use the "what they don't tell you about [X]" hook structure currently performing organically in the B2B coach space. Maps cleanly to a problem-aware audience.
Structural test: Conversion-ready. Clear angle, clear offer (free audit), clear CTA (book a call).
That's a brief that a creator can shoot, an editor can cut, and a media buyer can put behind real spend with conviction.
That's the framework, applied.
