The reason shot forty does not match shot one is almost never the model. It is that nobody wrote down what the world was supposed to look like, so every prompt re-invented it slightly. A set specification is the cure and it fits on one page.
It states one key light direction as a clock position, one colour temperature in kelvin, one named surface and background, one focal length range, one depth-of-field choice, and a list of what is never in frame. Every item is a decision that would otherwise be made forty times by whoever happens to be at the keyboard.
Written, not remembered. A specification in somebody’s head survives until the second batch runs while they are on holiday.
How long should a set specification be?
One page. If it runs longer, it has started describing shots rather than the world they sit in, and the shot list is a different document.
What is the most commonly missed item?
The exclusion list: what is never in frame. Props and background objects are where a set drifts fastest, because nothing in the brief forbids them.