Make them catch the bug. Leave the reader enthusiastic to learn more. That's the #1 thing that would make me want to buy it for others, being confident what they get from it won't be a flash in the pan.
If your book makes less money for you than it's bibliography makes for Amazon, I'd call it a success.
I'm shocked and very intrigued that no-one seems to agree with this one. Maybe I need to do some updating.
What am I missing? Why isn't it a huge deal whether the readers are motivated to keep improving and getting stronger after they've finished the book?
Followup to: The Most Frequently Useful Thing
What's the number one thing that goes into a book on rationality, which would make you buy a copy of that book for a friend? We can, of course, talk about all the ways that the rationality of the Distant World At Large needs to be improved. But in this case - I think the more useful data might be the Near question, "With respect to the people I actually know, what do I want to see in that book, so that I can give the book to them to explain it?"
(And again, please think of your own answer-component before reading others' comments.)