The cheapest time to add documentation is when you have the information in your head because you just worked on it. Nobody is demanding an essay for every method, but just write down what went through your head when you implemented weird hack #17 or found out that the API you're calling does something surprising.
The main challenge is that you would need to write the documentation for a person who might have never used this particular software before.
The information that is in your head might be nonsense for this person, and there is chance that it is not reducing the time it takes to understand in a meaningful way.
That is the problem
Not that it is true, it is not, for many reasons. It is a problem that is believed