Do the 20% that delivers 80% of the result. Measure. Adjust. Do it again.
Do not tackle 100% of what must be done in some random order, or, even worse, spend far too much time documenting 125% of what could be done because you have overthought and over engineered some made up problems that you imagine might need to be solved.
Deliver the 20% that has 80% of the results and don’t bother measuring and learning and never come back to improve upon what you have done and deliver the rest.
Both the 125% list and 20% hit and run are useless. You might as well not even bother. You are wasting time and money.
Deliver in small batch increments doing the highest impact to effort ratio work first. Measure the results. Adjust. And for damn sure repeat.