Every time a platform changes how it ranks ads, the advice cycle runs the same way. Somebody posts a number. Everybody repeats the number. Six months later nobody can source it.
So let us stay on solid ground. Meta has been rebuilding the machine learning that decides which ads get considered and shown, under the name Andromeda. The design goal is a system that can scan a much larger pool of candidate ads and pick the one most likely to suit a specific person in a specific moment.
You do not need a leaked percentage to work out what that implies for creative. If the system is better at choosing between options, the value of giving it options goes up. And options only count as options if they are actually different.
What actually changed, in plain terms
Older ad systems had to narrow the field early. There were simply too many ads to evaluate properly, so a lot of candidates were filtered out before anything sophisticated happened to them.
An overhaul like Andromeda pushes that ceiling up. More candidates survive to the stage where the system can make a fine-grained judgement about who should see what. The matching gets sharper.
Two consequences follow for anyone buying ads:
- Your creative library became your targeting. Manual audience carving matters less than it used to. The set of ads you supply is now the main lever you control over who sees what.
- Near-duplicates got less useful. When the system can tell people apart more precisely, being handed twelve versions of one ad does not help it serve any of them better.
That is the whole strategic argument, and it does not depend on any specific statistic being true.
The two responses
Since the shift, supplement and DTC brands have split into two camps.
The volume camp. Ship dozens of creatives a week. Hire UGC creators in bulk, cut every clip into many variants, flood the account and let the system sort it out. The logic is reasonable on its face: more candidates, better odds.
The diversity camp. Ship fewer creatives, but make each one a genuinely distinct reason to buy. Different argument, different world, different emotional register. The logic is that the system is choosing between concepts, so concepts are what you should supply.
The volume camp is louder because volume is measurable and easy to sell. It is also the camp that quietly runs into the wall we described in why every supplement ad looks the same: fatigue on steroids, because a large batch of near-identical ads burns the same audience faster than a small one.
Why hook-swapping counts as one creative
Here is the test we use. Show two ads to someone who already skipped the first one. If they recognize the second within half a second, those are not two creatives.
By that test, most "creative volume" is not volume at all:
| What brands ship | What it really is | Counts as |
|---|---|---|
| 12 hook lines over the same footage | An opening test | One creative |
| Same ad, different caption | A copy test | One creative |
| Same creator, same room, new script | A message test in one world | Roughly one creative |
| Same argument, new creator | A casting test | Close to one creative |
| New format and new argument | A genuinely new concept | A real second creative |
None of the top four rows are wasted work. Hook tests are how you improve a concept, and you should run them. The mistake is counting them as portfolio diversity. They are depth inside one concept. Depth is not width, and the ranking system is choosing on width.
What "diverse" actually means
We hold creatives against four axes. Two ads that differ on none of them are the same ad. Two that differ on two or more are genuinely separate bets.
- Reason to buy. Mechanism, price, speed, safety, identity, social proof, avoiding a specific mistake. A different reason reaches a different person, not just a different mood.
- Format and visual world. Live UGC, claymation, 3D character, 2D explainer, dramatized process. This is the axis that decides whether the viewer recognizes and skips.
- Emotional register. Funny, warm, alarming, clinical, matter of fact. The same argument lands very differently depending on the tone carrying it.
- Audience being addressed. The person who already knows the category versus the one who has never considered it. Those are two ads, always.
Run your current live ads through that grid honestly. Most supplement accounts we look at have between one and three real concepts, spread across thirty file names.
The hidden cost of the volume approach
Beyond the wasted production budget, mass volume of near-identical creative does three things to an account.
It accelerates fatigue. One idea shown under many names still exhausts the same people, only faster. You reach the cold remainder sooner and it looks like the platform got worse.
It hides what is working. With forty variants of one concept live, wins and losses average out into noise. You cannot tell whether the mechanism angle beat the price angle, because you never ran a clean version of either.
It sets a treadmill speed you cannot sustain. A weekly quota of live-action assets means a permanent shoot schedule. The moment the pace slips, the account has nothing in reserve, because nothing durable was built - only inventory that got consumed.
Why an animation pipeline produces diversity cheaper
The reason live-action volume collapses into sameness is not laziness. It is economics. Every genuinely new concept needs a new setting, new talent, new logistics. So teams do the rational thing and squeeze more variants out of the footage they already have.
Animation inverts that. The expensive part is building the world once. After that:
- A new argument is a new scene. The mechanism ad and the comparison ad share a pipeline, not a shoot day.
- A new register is a lighting and pacing change. The same character can be warm in one ad and blunt in the next.
- A new market is a new voice track. No recasting, no travel, no accent problem. We produce ads in several languages this way.
- A new format is a design decision. A 2D comparison and a 3D character explainer can come out of the same team in the same month.
That is the whole reason we build animated ads for this category. Not because animation is prettier, but because it is the only way we have found to produce genuinely distinct concepts at a pace that does not require a permanent film crew. The formats it makes possible are laid out in 7 supplement Facebook ad examples, and the persuasion logic behind them in show the mechanism, not the claim.
How to build a diverse slate
A practical starting shape for a supplement brand, rather than a quota:
- Keep your control. Whatever is working now stays live and untouched. Everything else is measured against it.
- Add one mechanism concept. The ad that answers "why would this work" with a picture instead of a sentence.
- Add one identity or frustration concept. A character story about the daily consequence, aimed at the person who has not named the category yet.
- Add one comparison concept. Fast, flat, readable in three seconds. Your product against the alternative or against doing nothing.
- Then go deep. Once a concept proves out, run hook variants inside it. That is where hook-swapping earns its place.
- Retire on the signal, not the calendar. Kill a concept when its hook rate slips and frequency climbs, not because it turned four weeks old.
Four real concepts with depth beats forty variants of one, both for the ranking system and for the person scrolling. And unlike a shoot quota, it leaves you with something at the end of the quarter: a set of worlds you own and can keep building on. That is the same argument we make about content operations generally in what a done-for-you content agency actually is.
If you want to see what a distinct concept looks like in this category, the full set is on our animated ads portfolio.