Scientists make predictions, and predicting the future is far more valuable than explaining the past.
Ask a physicist what will happen if you fire a projectile like this in that direction, and she'll know. Ask a chemist what happens if you mix x and y, and you'll get the right answer. Even quantum mechanics mechanics can give you probabilities that work out in the long run.
Analysts who come up with plausible explanations for what just happened don't help us as much, because it's not always easy to turn those explanations into useful action.
Take the layout of Craigslist. Just about any competent online designer would have predicted that it would fail. Too clunky, undesigned, too many links, not slick or trustworthy… Or consider a new r&b artist, or a brand new beverage.
After the fact, it's so easy to say, "of course it worked…" and then make up a reason for whatever it is that just succeeded.
The practice, then, is to start making predictions. In writing. You don't have to share them in public, but the habit will push you to understand your instincts and to sharpen your ability to see what works (and what doesn't) without the easy out of having to explain what already happened.
Look at startups or political campaigns or new products or ad campaigns… plenty of places to practice your predicting skills.
I predict you'll learn two things:
- It's really difficult to make predictions, because success often appears to be random
- Based on #1, it's probably smart for you to initiate more projects that aren't guaranteed winners, because most winners aren't guaranteed.
And a bonus… the more you practice your predictions, the better you'll get at discerning where the science is.