First thing first - Is it a big deal? For brands running ads on Meta platforms, this update has direct implications. So… yes! Stay with us, here is a breakdown:
Meta has quietly rolled out Andromeda, its most significant upgrade to the ad-delivery engine in years. At its core, this is a change from “which audience do we target?” to “which ad should this person see, right now?”
What Changed
Creatives now the main lever: It’s no longer just about who you target but what you show. The quality, diversity and volume of creatives you supply now matter significantly more. It doesn't mean you can deliver bad creatives - it’s not all volume. But volume is key for success.
New retrieval engine: The “retrieval” stage (where Meta narrows down which ads are eligible to show) has been completely rebuilt. Andromeda can now scan many more ad-candidates in milliseconds, thanks to a new deep-neural-network architecture and upgraded hardware (e.g., NVIDIA Grace Hopper Superchip).
Less emphasis on micro-targeting: Historically, you’d build finely segmented audiences by interests, demographics, behaviours. Under Andromeda, broader targeting + letting the algorithm find the right people works better.
Simplified campaign structure: Fewer campaigns/ad-sets, more reliance on automation. The system favours fewer, larger buckets rather than ultra-splintered ones.
Data & signal-quality matter more: Since the algorithm is doing more heavy lifting, the input signals it receives, conversion events, quality tracking, etc , become even more important. Weak signals limit its effectiveness.
Creative fatigue hits faster: Because the system is trying more variations and matching more precisely, the insides of its creative learning loop move faster. Which means if you keep showing the same ad to the same people, performance may degrade more quickly.
What This Means for Brands Advertising on Meta
For brands running ads on Meta platforms, this update has direct implications:
Here is what you need to do:
Focus on building a diverse creative library: different angles, formats (video, carousel, static), messages, personas. Because Andromeda is optimising which creative works for which person. In other words - create more ads, but make sure they are all good ads!
Adopt broader targeting or “set it and forget it” style audience designs. Let Meta’s system find the best matches rather than manually carving tiny segments.
Simplify campaign structure: fewer campaigns, fewer ad-sets, more creatives. Use Meta’s placements and automation features (e.g., Advantage+).
Bug fix your data infrastructure: Ensure clean conversion tracking, high-quality signal, full funnel visibility. Without this, the new engine can’t show you its full benefit.
Monitor for creative fatigue and refresh more often.
What to watch out for:
If you rely heavily on precise targeting (interest stacks, micro-audiences), that strategy may be less effective now. You might see rising CPAs or declining performance if you keep doing what you used to.
If your creative is weak (few variations, minor tweaks only) or your offer / funnel steps are weak, Andromeda won’t magically fix that. The system amplifies good setups and exposes weak ones.
Because the algorithm now optimises more aggressively, your measurement strategy should shift: monitor not only superficial metrics (CPM, CPC) but true business outcomes, new customer acquisition, quality of conversions, cost of acquiring net-new rather than repeat buyers.
Key Takeaway for Upspring’s Clients
The game has changed a bit. Brands that adapt will benefit. Brands that keep running old-playbook tactics may see declining returns.
Shift budget and effort from targeting tactics → creative generation & data signal. Don’t worry, Upspring got you covered. Everything you need can be found in your dashboards.
Encourage creative teams to stay connected to performance metrics. Not just aesthetics. When creatives understand what drives CTR, ROAS, and attention curves, they can design with data in mind and feed Andromeda the variety it needs to learn faster. Again, we got you!
Emphasize “telling different stories to different people” rather than “finding the perfect audience”. See what’s working and scale.
Use the update as an opportunity to simplify campaign architecture and make ad operations more scalable.
Written by Yishai Dvash, Co-founder and CMO
001
What is Creative Intelligence
002
What is Upspring.ai’s partnership with Meta?
003
How does Creative Intelligence improve workflow efficiency?
004
How is Upspring different from creative management tools like Motion or Foreplay?
005
How does Upspring.ai help agencies win more clients and scale creative performance?
006
How can brands use Upspring.ai to scale what works?
