Meta advertisers still talk about targeting as if it is the first lever in the system. That is the old mental model.
The newer model is messier, more AI-driven, and much more creative-sensitive. Meta's ad stack now evaluates ad content far earlier in delivery, which means creative quality and variety do more of the sorting work that advertisers used to force through audience settings.
In this post:
- What Andromeda changes in Meta's delivery logic
- Why broader targeting and stronger creative inputs now work together
- How campaign structure should change in response
- What operational demands this creates for Meta ads teams
What Andromeda Actually Changed
The clearest explanation of Andromeda is not "Meta got better targeting." The clearer explanation is "Meta moved more of the matching logic into its retrieval system."
Search Engine Land's breakdown of Andromeda and GEM describes Andromeda as the layer that determines which ads are eligible to be shown to a user before later-stage ranking decides what should appear next. That matters because it shifts the optimization game.
Under the old account playbook, advertisers tried to win through tighter audience construction and more manual segmentation. Under the newer playbook, Meta wants broader opportunity sets and stronger inputs.
| Lever | Older Meta playbook | AI-first Meta playbook |
|---|---|---|
| Audience setup | Granular segments and layered interests | Broader inputs, fewer constraints |
| Campaign structure | More fragmentation | More consolidation |
| Creative role | Important but often secondary | Primary signal in retrieval and matching |
| Optimization work | Manual refinement inside the account | Better signals fed into Meta's system |
That does not mean targeting disappeared. It means targeting is no longer the cleanest source of leverage for most teams.
Why Creative Carries More Weight Now
When the system has broader room to explore, it needs stronger clues about who should see an ad. Creative now provides more of those clues.
This is why teams keep seeing the same pattern:
- broad targeting holds up better than expected
- heavily segmented campaign trees lose efficiency
- new creative angles matter more than tiny audience tweaks
Meta has also pushed advertisers toward more automated campaign formats. eMarketer summarized Meta's reporting that Advantage+ shopping campaigns delivered about 22% higher ROAS than business-as-usual approaches. That figure is not proof that every account will see the same result. It is proof of Meta's direction: broader automation paired with stronger creative and system learning.
The practical takeaway is not "trust the machine blindly." The takeaway is "feed the machine better creative inputs and stop over-constraining the auction."
Creative Variety Matters More Than Creative Volume
One mistake teams make after hearing this is to produce more files without producing more ideas.
That is not the same thing.
Five near-identical ads do not give Meta five strong inputs. They give Meta one idea expressed five times. If the hooks, offer framing, visual treatment, and angle are all basically the same, you are not giving the system much room to learn.
If your new batch is just color swaps, crop changes, or minor copy edits, you have increased output without increasing diversity. Under Andromeda, conceptual variety matters more than cosmetic variation.
This is where many teams misread performance. They think the issue is targeting drift when the real issue is creative redundancy. The account has motion, but not enough distinct inputs for Meta's retrieval and ranking layers to explore meaningfully.
That is also why The Meta Ads Scaling Wall shows up earlier now. The bottleneck is often not strategy. It is the team's ability to generate, package, and launch enough distinct creative concepts to keep the learning system fed.
How Campaign Structure Should Change
If creative now carries more delivery weight, campaign structure should reflect that.
For many accounts, that means:
- Fewer campaigns built around narrow audience theories
- Broader targeting unless there is a strong business reason to constrain it
- More creative concepts inside the campaigns that matter
- Faster replacement of stale creative before fatigue compounds
This does not mean every account should collapse into one campaign. It means the default should move away from fragmentation for its own sake.
The higher-value question becomes: "What set of concepts do I want the system to test?" not "How many tiny audience buckets can I force this through?"
That shift also aligns with how bulk is positioned. bulk is useful when the real work is not another strategic theory but the execution layer: turning creative concepts into live ads, at speed, without drowning the team in upload and setup work.
What This Changes for Operators
The teams that adapt fastest are not just writing better ads. They are changing operating rhythm.
They brief more concepts. They ship new creative faster. They spend less time rebuilding audience trees that Meta no longer needs in the same way. They treat campaign structure as a support system for creative testing, not as the main source of performance.
That makes execution capacity a strategic issue. If your team can only get a new batch live once a week because uploads and ad creation still take hours, you are not set up for the way Meta now learns.
That is why the link between Andromeda and workflow matters. AI Upload vs. Manual Upload is not just an efficiency story anymore. It is part of the campaign strategy story. Faster execution means more distinct concepts launched, more feedback collected, and less time stuck in a stale creative set.
The old Meta advantage came from tighter manual control. The newer advantage comes from stronger creative signals, fewer artificial constraints, and a team that can keep pace with the system.
bulk handles Meta creative upload and ad creation so your team can feed AI-first campaign structures with real creative variety instead of getting stuck in manual setup work. Try bulk free →