I spent years checking mold halves before they were put together, to make sure they would not crash when put in the press.
Each mold was different as far as the building and modifying, so we would use the finished part as our acceptance criteria.
Sometimes the molds were built to be way different than the part, so when the part would cool it would come into specification.
(inward/outward bow, flatness, shrink, etc.)
The only way to know if the build was successful would be to run samples.
We relied a lot on the experience of the engineers and toolmakers, definitely not an exact science.
Hope it helps,
Ed