How does Facebook’s algorithm optimize ads with limited data?

When data is limited, Facebook optimizes by expanding distribution, relying on high-level signals, strong creative input, and learning from industry data and the Meta ecosystem.

Facebook optimizes advertising by using machine learning to identify behavior and allocate budgets to the groups most likely to convert. This process essentially involves finding people with similar characteristics to those who have already interacted with the target audience, thus showing them ads more frequently. Whether the budget is large or small, the system operates on the same mechanism; however, with a low budget, machine learning will need more time to accumulate data to achieve consistent effectiveness.

I notice that CBO campaigns prioritize budget optimization on CPC, not over leads or sales.