
What counts as a new creative concept on Meta, and what's just a variation
Key takeaways
At Fifth Dimension, a concept is the message an ad gives the customer. A variation delivers that same message in a different format, to a different persona or through a different motivator. An iteration is a change we make after launch, once the data shows what to fix. Meta says it groups ads that look or feel alike and treats them as variations of the same creative, so new concepts are what add real creative diversity. Variations still matter, because the way a message is delivered decides how clearly it lands.
The words decide what you think you're testing
Since Meta's Andromeda updates, a lot of advice tells brands to make more concepts and fewer variations. Very little of it explains what those words mean.
Ask three agencies what a concept is and you may get three answers. Some call a new hook a new concept. Some call a new creator a new concept. Others use variation and iteration as if they were the same thing. These terms might be interchangeable at other agencies, and that's fine. You just need to know which one you're making, because each answers a different question in your creative testing. This is how we use the three terms at Fifth Dimension.
A concept is the message
A concept, for us, is the high-level idea of what the ad tells the customer. It's interchangeable with the messaging angle, or ad angle. If you can say what the ad tells people in one sentence, without describing the format or the person in it, that sentence is the concept.
Say a skincare brand has a clinical result for its moisturizer. A bold headline that says how much the cream repairs the skin barrier, and how fast, is the concept. It's the message, and it's the first variable we decide on.
We plan in concepts per month or concepts per week, because the number of concepts shows how diverse your messaging is. If you run forty ads built on two messages, buyers still only get two reasons to care.
A variation is the same message in a new form
A variation keeps the concept and swaps one of the other variables. The message stays the same, and only the way it reaches people changes.
Take the barrier repair concept. We could:
- put the claim in a static ad with a bold headline,
- have a UGC creator say it in their own words, in a few different ways,
- build a carousel around it that leads with more detail about the study.
Each one is a variation, because each delivers the same claim in a different format. Format is one of three secondary variables we mix and match. The other two are the persona, meaning the kind of person the ad speaks to, and the motivator, meaning the reason that person cares. Mixing them gives us several variations out of one concept. If you're deciding who should say it on camera, our guide to AI UGC vs real creators compares the two.
| Variable we change | Example | What stays the same |
|---|---|---|
| Format | A static with a bold headline, a UGC video or a carousel about the study | The barrier repair claim |
| Persona | A creator speaking to people with sensitive skin, or one speaking to people with very dry skin | The barrier repair claim |
| Motivator | Calm skin before a big event, or finally finding a cream that doesn't sting | The barrier repair claim |
Variations still matter, and one launch shows why
Variations aren't filler, because the way you deliver a message decides how clearly people get it.
When we launched a new product for a brand we work with, we wrote one ad with a bold headline. Then we used that same headline in front of different images. One image had a striking red background and showed the products neatly. The message was crystal clear, and in the first few days of the launch Meta put 80% of the spend behind that one creative.
The headline was the same in every version, so the image made the difference. If we had made only one version, we might never have found the one that made the message land. Meta's own guidance on creative variety suggests a similar move: keep the featured product and try new backgrounds around it.
An iteration comes after launch
Variations and iterations get mixed up a lot. For us, an iteration is what comes after an ad has launched. We look at the data, then decide what needs to change: the body, the hook, the headline or the creator.
A variation is planned before launch, from the concept. An iteration is a decision made after launch, from the results. A weak opening points to the hook. People who watch but don't act point to the body or the headline. Iterations need to come back fast, which is one reason I now edit many of our ads with Claude, as I explain in my article on agentic video editing. Every AI-made version still goes through the eight checks we run before AI ads go live.
Meta uses the word differently. Its December 2025 guide describes iteration as “manual A/B testing to refine assets, formats, and messages,” and its example is two ads with identical visuals but different calls to action. That covers both our variations and our iterations. We split them because the timing changes the question you're asking.
| What to compare | Concept | Variation | Iteration |
|---|---|---|---|
| What it is | The message the ad gives the customer | The same message in a different format, persona or motivator | A change to a live ad, based on its results |
| When it happens | First, before anything is made | Before launch | After launch |
| What changes | The message itself | How the message is delivered | The body, hook, headline or creator |
| The question it answers | Which message makes people want to buy? | Which way of delivering this message works best? | What should we fix or build on, now that we have data? |
| Example | A clinical claim about barrier repair | The claim as a static, a UGC video and a carousel | A new hook on the UGC video after a weak start |
What Meta says about ads that look alike
You may have heard that Meta now treats near-identical ads as one ad. Meta's own wording is a little more careful, and it's worth reading.
In its December 2025 guide to creative diversification, Meta says its ads system groups together ads that share key visual and thematic attributes. When ads look or feel alike, it sees them as variations of the same creative, and it shares learnings and delivery optimizations at the creative level. The same guide says iteration can inform new concepts but should not replace true creative diversification.
The pool those ads compete in keeps growing. On 28 October 2026, Meta's engineering team described a new version of Andromeda, the system that picks which ads get considered for each person. It now scores every eligible ad across Facebook and Instagram and surfaces the top candidates at over one billion ads per second, which Meta calls a 100x lift on its 2024 system. The same post says generative AI tools have helped advertisers create hundreds of millions of new ads. Our Meta Andromeda entry explains the system in more detail.
In our view, ten small edits of one idea don't give Meta ten ideas. New concepts give it new messages to match with new people, and that's what we mean by creative diversity. Variations then help it find the best way to deliver each message, as they did in that launch.
How we plan creative testing around concepts
We count concepts, not ads, because that shows how many different messages go out each month.
- Research
We study customer reviews, competitors' ads and buyer conversations to find the different reasons people buy and what stops them.
- Concepts
Each reason becomes its own concept, written as one clear message.
- Variations
We deliver each concept in a few formats, to a few personas or through a few motivators.
- Launch
The ads go live, and the results show which messages and which versions people respond to.
- Iterations
We change the hook, body, headline or creator on the live ads, and what we learn shapes the next round of concepts.
How many concepts you need depends on your budget, your products and what you've already tested. With a smaller budget, we may build several concepts for one type of buyer rather than spread the budget thin across many. We also label every ad with its concept, persona and hypothesis, so we can look back at a result and see what it was meant to test.
Check your own account
- Write the message of each live ad in one sentence.
- Group the ads by message. The number of groups is your real number of concepts.
- Look at how each concept is delivered. If every version is the same format with a new background, there's room for new formats, personas and motivators.
- Separate the changes you planned before launch from the ones the data asked for, so you know what each result is telling you.
If most of your ads share two or three messages, the next step is new concepts. More versions of the same message won't add much. If your ads get clicks but no sales, check the site and tracking before you blame the creative, as our guide to Facebook ads getting clicks but no sales explains.
Where Fifth Dimension fits
This is the work behind our Creative Strategy service. It covers the research, concepts, briefs and production, with media buying if you want it. If you have your own media buyer, they can keep running the account while we handle the creative.
I'm personally involved in every account, leading the day-to-day strategy and major parts of the execution, with no account manager in between. You can see more of our work in the ad portfolio and case studies, or book a call to talk through your account.
Putting it into practice
- Count your concepts, not your ads, and write each one as a single message.
- Make a few variations of every concept so you find the best way to deliver it.
- Save iterations for after launch, and let the data tell you which part to change.
Book a call

