When it comes to generative design processes, it is vital that you know your design parameters inside and out.Every good generative design project starts with a clear and precise understanding of the design problem and a clear description of the goals.
Algorithms are great at churning through thousands of design options very quickly, but they don’t perform nearly as well if they're given vague or imprecise instructions. You must be able to define your problem in a mathematical way (i.e. with some sort of number that can be used to rank outcomes).
Some good questions to ask when formulating design goals are:
What do you want to achieve?
Which features must your ideal design have?
Which features cannot appear in your ideal design?
Do you simply want to see a lot of design options?
Do you want to optimize your design for some specific characteristic?
Do you want to optimize your design for multiple competing characteristics?
What would you like to maximize? Why?
What would you like to minimize? Why?
Can your maximization or minimization question be quantified mathematically? If so, how precisely?
Being able to confidently answer at least some of the questions above is a good first step to figure out precisely which objectives your computational procedure should have.