This is for real, although we should acknowledge that some successes would be much harder to see than failures. (Some should be much more obvious, though... The FBI loudly celebrates terrorist attacks it prevents, and yet all the self-described wins turn out to be pretty lame.)

SIGINT sometimes does really well. Enigma is the obvious example, but I'm sure the recently-revealed wiretaps on international conferences produced some politically valuable stuff.

There are even occasional wins on analysis. The German tank problem was a major statistical breakthrough, and Abraham Wald pioneered operations research doing air force design analysis. But... both of those were unconventional, marking breaks from normal intelligence work.

Moreover, almost all the wins I've outlined here came during wartime, a setting where a very clear 'they' is undeniably out to hide secrets and attack you. Things tend to get way less impressive during peacetime.

I can't find it at the moment, but John le Carre tells a brilliant story about this. He was an MI6 man in the cold war, before he was a writer. And the beginning of Tinker, Tailor, a mission to Hungary to meet a defecting general, was inspired by a real experience. But in real life, there was no general - le Carre's superior had made him up. His boss led him on a dangerous covert mission across the Iron Curtain for no reason, simply because real Cold War spy work was so aimless and boring.