It’s akin to getting your driving license. You study/practice for the test…you pass the test…but your driving history (e.g., accidents, tickets, subsequent insurance rates, etc.) can be used to determine the effectiveness of your training (and learning from experience).
Great example.
It's imperative to define (and IMO document - makes people more serious about "define") what the goal(s) of each training effort is. In specific.
Back to the driving example: What is the goal of the driving lessons?
If it's "to pass the assessment and get the licence" - if the new driver passes the assessment, the lessons have achieved their goal (hopefully; at least it's reasonable to conclude that they contributed), so it's a fair statement that they were effective. End of story.
If it's, however, to teach the new driver to be a "good" (or safe, or anything else) driver, and they passed the assessment, it still doesn't mean they've learned how to be a good driver (as we all know, sadly!).
The measurement has to be aligned with the goal the training is aimed to achieve. The more accurately the goal is pinpointed, the easier it will be to frame the right (effective, haha) way to measure success, and the more likely the measurement will be relevant. But being pedantic is hard work and not much fun, right?