- The Performers
- Posts
- I Tested Meta’s Andromeda Update. Here’s What Happened
I Tested Meta’s Andromeda Update. Here’s What Happened
How to Survive the Andromeda (Testing and Scaling Roadmap)
What's up, Marketers! This is Aazar.
This newsletter is about leveling up your paid growth marketing skills by analyzing the best brands' paid strategy, tactics, positioning, and value props.
This newsletter is divided into:
Sharing what I've learned
Sometimes sharing some other performance marketers’ lessons with you
And I analyze & compare the best ads on the internet (this issue)
I've analyzed over 40,000 ads from 11 nine and ten-figure D2C brands.
Here's what I discovered:
The brands generating $2.8 billion in revenue aren't doing what you think.
They're using emotion over hooks. They're launching 2,000+ ads. And they all have one thing in common: a clear villain.
I spent 40+ hours breaking down what's actually working:
→ The exact ad formats Dr. Squatch, AG1, and Oura Ring use to dominate
→ Why comparison ads still crush (and how to use them without naming competitors)
→ The psychology behind ugly ads that convert better than polished ones
→ 9 key takeaways you can apply to your brand today
The shocking part? Most of these brands are running the same proven formats for months because they work.
I've turned this into a complete video breakdown with real ad examples, swipe files, and frameworks you can steal immediately.
Plus, I'm sharing a full Excel analysis in the description so you can study at your own pace.
This newsletter is sponsored by Insense.
And most brands don’t even realize it yet.
Andromeda is quietly downgrading ads that look too similar, even if they perform well. The result?
Ad fatigue hits faster
Winning ads lose momentum overnight
CPAs climb, despite strong strategies
The fix isn’t bigger budgets or endless testing.
It’s creative diversity.
1. The Meta shift that’s reshaping UGC
Meta Andromeda now rewards variety: different creators, hooks, and visual styles.
If your UGC ads share the same vibe or feature the same faces, the algorithm flags them as duplicates before they even hit the auction.
That’s why leading DTC brands are shifting to diverse, modular UGC systems that scale content and authenticity.
Different creators, styles, and storylines — all Andromeda-approved.
2. The new winners: modular UGC at scale
Instead of spending more, top brands are producing smarter.
They build modular content systems where one shoot = dozens of unique hooks, angles, and personas.
But the real edge? More faces.
To meet Andromeda’s demand for diversity, brands now need greater creative volume, sourced from more creators than ever before.
3. Here’s the shortcut
Insense built the fastest way to scale modular UGC that’s Andromeda-ready:
Request both raw footage & edited videos using an interactive brief that helps you specify hooks, CTAs, and styles for maximum diversity
Match with 70K+ vetted creators across 35+ countries - ensuring a mix of faces, backgrounds, and styles that resonate globally
Manage everything end-to-end: payments, lifetime content rights, and even FB Ads Manager integration for Partnership Ads
No more scrambling for “just one more variation.”
This is a systematic approach to creative production for achieving diverse UGC at scale.
4. Two ways to roll
Self-Serve Platform: You control briefs, creators, and campaigns.
Managed Services: Their expert team handles everything from concept to delivery.
Choose your setup. Scale your creative diversity.
If your Meta ads have felt weird recently… It’s not just you.
CPM is going up. Winners are dying faster. New ads are struggling to get out of learning.
Same budget. Same funnel. But different results.
Everyone's talking about Andromeda right now. Every group chat. Every Slack channel. Every advertiser is trying to figure out what the hell just happened.
So I did what I always do: I tested it.
Ran campaigns. Watched what broke. Watched what worked. And honestly? I'm still figuring parts of this out 😅.
But here's what I know so far, and it's changing how we're running ads completely.
Meta flipped a big switch called Andromeda, a new AI engine that changes how your ads are selected and delivered.
The rollout started earlier this year, but most accounts are only feeling the impact now or in a couple of months.
And the shift is bigger than most people realize (But the good creative strategists will still keep winning, I’ll tell you why in a sec).
Here's the simple version:
Meta now compares your ads visually before performance even matters
If your ads look similar, Meta treats them as the (almost) same ad
It picks one winner from the batch and pours all the spend into it
That winner burns out faster than before
Everything else gets ignored
That's the surface level. But the way this actually plays out in your account is where it gets interesting.
And this is exactly why your ad performance now feels unstable.
In this edition, I'll show you:
What Andromeda actually changed
Real account examples
How testing campaign works now
How scaling campaign works now
Let's dive in.
What Actually Changed With Andromeda
Let's keep this very simple.
Meta didn't just "update" delivery. It changed how its brain works.
Meta used to judge ads by performance. Now it judges ads by how visually different they look.
If 8 of your ads look like cousins (same colors, same layout, same energy), Meta groups them together.
Then it chooses one from that group… and pushes all your spend into it.
That's why:
Your tests feel incomplete
Your variations don't get spend
Your winners burn faster
Scaling feels unstable
Why Similar Ads Are Dying Instantly
We tested multiple creatives for a brand.
Different text. Different lines. Different hooks.
But visually?
Same yellow palette
Same ingredients
Same top-down angle
Same cup
Same layout
Meta saw them as the same.
So it picked one winner. Everything else got ignored.
Even when we added tiny changes (slightly new layout or slightly new testimonial), Meta still collapsed them into one bucket.
This is why your ads don't "get spend." They never even get a chance.

How Testing Has Changed Under Andromeda
Most advertisers are still testing the old way:
1 concept
4–5 small variations
Slight background tweaks
Under Andromeda, this does not work.
The new way to test:
1. Pick one concept
Example concepts:
Shame/embarrassment
Quitting a bad alternative
Effortless transformation
Made for moms
Price comparison
Social proof
Status identity
You can test multiple later, but one concept at a time keeps your data clean and your learnings clear.
2. Create 5–8 visually different variations of that concept
Not remixes. Not color tweaks. Not tiny edits.
Visually different.
Different:
Faces
Colors
Hooks
Ad formats
Angles
Visual archetype
3. Meta will still pick 1–2 winners
This is normal under Andromeda (I’m still learning).
4. Move that winner into scaling
This is where the real game begins. As you can see below, all ads look different in one of our accounts’ winning/scaling campaigns.

Here’s the truth about testing
Out of my 12 accounts, 9 are still using the old method of testing one ad with different variations, and they are still finding successful winning ads. So, it's not too bad. Only 3 of my accounts are seeing a variety of ads being used. If you're using the old testing method, it still works. However, this new update might become important in 3-6 months, so we should consider changing our testing process. That's all I wanted to share about testing.
Scaling Under Andromeda (How to Stop Your Winners From Dying)
Here's the truth: scaling is where fatigue hits the fastest now. Andromeda is ACTUALLY about having diversity in your scaling campaign imho.
Meta picks the one ad it loves most… then overserves it. That ad dies in days or weeks.
But this is fixable.
Here's the new scaling strategy:
1. Your scaling campaign needs VARIETY, not volume
Stop adding 5 similar ads from the same shoot. Meta collapses them.
Your scaling ads need to look like they belong to different worlds:
A UGC talking-head
A clean studio shot
A lifestyle POV moment
A meme-style graphic
A bold product-only visual
A lo-fi creator selfie
A comparison shot
A testimonial moment
2. Scaling works like a playlist
One track on repeat gets boring fast. Same with ads.
You need 5–8 visually different ads that can rotate naturally.
(The above ad is still in testing, so the ROAS is high.)
3. Never move a testing winner into scaling alone
If you do this:
Meta pushes it
Frequency jumps
Fatigue hits
ROAS dies
Instead: Add the winner + 5–7 ads that look nothing like it.
This protects it.
TBH: I think I was doing it anyway and all winners looked different from each other.
4. Scaling ads must talk about different things
Hit different angles:
Taste
Convenience
Identity
Social proof
Pain relief
Price
Routine
Emotional benefit
Different angles = better distribution.
5. Weekly rotation matters
Winners might die faster.
So essentially, you need to test more ads with variety so you keep pumping more winners in your scaling campaign.
Why You Now Need More Creators
One creator = one visual world. Andromeda needs multiple worlds.
Your ads now need:
Different faces
Different voices
Different shooting styles
Different rooms
Different lighting
Different first 3 seconds
This is the only way Meta treats creatives as truly distinct.
The strategy is clear. The execution is the hard part.
You need systems to:
Source and manage multiple creators
Produce visually different variations of the same concept
Test with structure, not chaos
Build a rotating library of fresh visuals
Keep fatigue at bay week after week
If your creative process can't keep up with Andromeda's demands, your performance will plateau no matter how good your media buying is.
(This is the exact problem we solve for brands, scaling creative production without the usual trade-offs in quality or speed.)
One more important thing we’re seeing:
Brands are placing all their TOFU, MOFU, and BOFU ads in one ad set and allowing Meta to optimize them. Interestingly, they don’t turn off any ads, even though two of them are spending. Hence, we might move in this direction too.
Final Takeaways (TL;DR)
Andromeda groups similar-looking ads together
If your creatives look alike, only one gets spend
Variation matters more than ever
Testing = 1 concept × 5–8 visually different ads
Scaling = 1 winner + 5–7 ads that look NOTHING like it
Weekly rotation keeps winners alive
One creator is no longer enough
More creators = more visual variety = better results
Visual diversity is the #1 lever
All of this means you need multiple creative strategists or agencies to help you out with this. You need DIVERSE, unique ideas. That’s where we can help you as an agency. Book a call if you want diverse, unique, and thoughtful ads that win consistently.
That’s it for this week.
Thanks to our partners who support this newsletter.
Tools worth checking out:
Atria: You're only as good an advertiser as your swipe file. Atria helps save good ads and analyze them in-depth. But the best part? Their AI helps me create concepts and scripts within seconds. Check it out for free. Most importantly, they now have built-in ad analytics to make more winning ads.
They also provide now weekly iteration ideas so you move fast and push more content.
Happy Growing with Paid Social,
Aazar Shad
Since this newsletter is free, I do it to follow my curiosity. But I’d love it if you could leave some feedback so I know if I am helping you or not.
What did you think of this newsletter? I appreciate your feedback! |
Three ways I can help you, whenever you are ready:
Work with me to get you growth from paid marketing. Book a call here. I’m open to more clients now.
Level up your paid marketing by joining my community, where we share the latest tactics and get nuanced paid marketing questions answered here (we are now 90+, but there is a waitlist to join).
If you’re looking to level up your creative ad strategy, check out our bootcamp recordings and resources on-demand, only for $197. Prices are going up by 30% soon. Simply pay here, I’ll send you a follow-up email immediately.
Promote yourself to 15,000+ highly qualified paid marketers by sponsoring this newsletter. Advertise here.


Reply