CPA needs an observed outcome
CPA equals spend divided by observed outcomes. When outcome count is blank, outcome evidence is required. When it is zero, CPA remains unavailable rather than becoming infinity in the normal metric row.
A hypothetical boundary
Hypothetical example: a team supplies a USD 30 target CPA and leaves the optional maximum blank. The calculator shows its product default of USD 42, or target CPA × 1.4. This is a visible heuristic, not an industry benchmark.
At USD 20 spend with no outcome, the bounded state remains collect evidence. At USD 42 with no outcome, it becomes inspect before scaling. Neither state sends an instruction to Meta.
Small samples stay small
One or two efficient outcomes are labeled as a limited sample. The result can protect the current budget, but it cannot call the ad a repeatable winner.