It’s not mandatory to do anything at all. If you do then it might be a bit dull though.
Composer is really very simple. Each separate visually different part of your game is a scene - title page, high score table, game for example. Different levels are the same ‘scene’. You switch from scene to scene using gotoScene() - triggered by various things that might happen in a game, a time period expiring, a button being tapped. A bit like a rotating stage in a play (I have my own version of Composer, Corona’s doesn’t fit my style, it’s about 400 lines of code and about 1/3 of those are comments - there’s not a huge amount in it )
Whenever you go to a new scene Corona tells you wants going on - that’s what the ‘events’ are, they are messages from Corona “I’m about to switch to this scene” (will) “I’ve just switched to this scene” (did). You can ignore them if you want - though usually you have to do something. Apart from this all Corona really does is automate the transition bits for you.
Composer is actually a thing called a “Finite State Machine” which is dedicated to scene changes. If you have a look for FSM on google there are plenty of articles (it is used in gaming for many things that have a ‘state’) that will range from the very simple to the horribly complicated. It is a worthwhile ‘design pattern’ to know outside its use in Composer. It is especially useful in AI.
(As an example, in the game ‘PacMan’ the ghosts have two ‘states’ - wandering around and running away).
They are often done doing diagrams showing the states (scenes) and how you move between them (this is one I found mimicking Ant behaviour). The ant has four states, which it shifts between as things happen. It may be worthwhile drawing a similar sort of diagram for your game before you start - instead of ‘Ant states’ you would have ‘game scenes’, and the arrows, which represent some sort of action or event in the game are scene transfers.