Star Wars dice are extremely difficult to analyse statistically and I speak as someone who has taught statistics at a Master's degree level. My congratulations to the people who designed them for that.
You cannot treat success and advantage as though they were independent outcomes because they aren't; both dice have faces with one of each. You also can't add up all the symbols and divide by the number of sides as say that is the chance because this treats .7 successes as a meaningful result which is only reasonable in dice pools with dozens of dice: this never happens. Alternatively, .7 represents the long run average over all possible dice rolls, however, in the long run all humans are dead, what matters is if we avoid death now!
So, what does happen? (All results are /24)
Result Green Yellow
- 3 2
S 6 6 (inc. T)
2S 3 4
AS 3 6
A 6 2
2A 3 4
Upgrading gives a 5/24 chance of significant improvement. Reducing you chance of nothing by 1/24, giving 2 success instead. It upgrades single advantage 4/24 times, adding a success 3/24 and another advantage 1/24. In addition it carries that magical triumph symbol, which can change the path of the adventure.
Now additional advantages (and triumphs) are always advantageous (heh) but additional successes are more situational.
In combat each extra success do another wound (yay!); outside of combat you generally need one more success than failure and any left over are wasted, a bit like carrying water through the desert - you have to have enough but too much is just dead weight. To work this out you need to consider the whole dice pool - how many failures will you need to overcome? If it's a lot, a 4/24 chance of an extra one from an upgrade is not to be sneezed at. If success is a cake walk, sneeze away!
As to if light side points are better used for narrative benefits, that is situational on your style of play as well as the dice pool situation. A well prepared group who likes to cover all bases in planning is likely to need them less than a group that takes things on the run.
How valuable a light side point is also depends on their velocity in your particular group. This is largely dictated by the GM's attitude but players have their impact too. A group where points pass back and forth rapidly will value them less "easy come, easy go"; hoarders on both sides of the screen drive the value up. Interestingly, both play styles will tend to have one when they really, really need one: the laissez-faire because they usually have one, the misers because they usually save one.