Yes, what you're noticing is a real and increasingly common behavior with Meta Ads — even when you manually select an ad creative, Meta’s delivery algorithm may start prioritizing another version if it's in the same ad set or campaign.
This usually happens due to:
1. Implicit Creative Optimization (even without Advantage+ fully enabled)
Meta uses internal “creative performance prediction” models.
If you’ve added multiple creatives (even manually), the system may favor the one performing better, even if your intent was to test them equally.
This happens even more aggressively in Engagement or Conversion campaigns.
2. Duplicate Post IDs or Similar Content
If you reused post IDs, or Meta recognizes similar creative structures, it may roll over learnings from past high-performers and prioritize them.
This results in Meta subtly pushing that version — even when others are selected manually.
3. Campaign Learning Phase Behavior
In early phases (especially <50 conversions), Meta tests delivery across creatives, even without you enabling A/B testing.
The best-performing ad gets the budget — which means your manual preference gets sidelined if it's not converting quickly.
What You Can Do to Control This:
| Goal | Tactic |
|---|
| Run creatives equally | Use split test feature (A/B test) or duplicate ad sets with only one creative each |
| Prevent override | Use 1 ad per ad set (Meta can’t override what doesn’t exist) |
| Avoid overlap | Don’t group creatives with very similar angles or offers — Meta will consolidate delivery |
| Force control | Run separate campaigns with budget caps if needed to manually enforce distribution |
Meta is increasingly designed to favor performance-based delivery, even when it means disobeying manual structure. The algorithm’s goal is ROAS, not fairness in testing . So unless you explicitly use split testing or isolate creatives, it’ll always push what converts best (from its perspective).