Winning creative from a test does not work when scaled

Many tests win on a small pocket of audience. When scaled, the system expands to less responsive users and performance normalizes.

This issue is common among Meta advertisers and usually indicates a structural or optimization problem that can be resolved with the right approach. Understanding the mechanism behind the problem is the first step to fixing it.

This problem typically stems from one or more of the following:

• Check audience size and overlap between test and scale setup
• Check if the test was under-delivering (small sample)
• Scale by keeping structure stable and adding new creative angles

Follow these steps to identify the root cause:

1. Check audience size and overlap between test and scale setup
2. Check if the test was under-delivering (small sample)
3. Scale by keeping structure stable and adding new creative angles

Once you've identified the cause, apply these fixes:

1. Start by identifying the root cause using the diagnostic checklist above
2. Make one change at a time and wait 24-48 hours to see results
3. Document what you changed and the outcome for future reference
4. If the issue persists, consider reaching out to Meta support with specific details
5. Use the quick fix page for immediate action items