The 80% rule was based on lab rats who were kept slightly hungry prior to experiments. The idea was to measure after what percentage of correct attempts is it OK to move to the next level of difficulty. It was deemed to be 80%.
However, data on hungry lab rats have to be taken to the real world with caution.
Kay Laurence, at The Training Thoughtfully conference, talked about this rule and microshaping. What we need to aim to do is set things up so the animal gets the behaviour, or first step, or weight shift etc, correct 95% of the time.
As soon as we start to reinforce big steps that are an OK approximation of what we intended we are training behaviours we will later have to extinguish. The extinction process can be stressful if not done carefully.