I make sure I understand the problem and the desired result.

I consider if I can break the problem down into simpler problems.

I want to have a map in my mind, or in notes on paper, so I’m writing code according to a plan, rather than writing code in hopes a plan or solution will happen by accident.

Measure twice, cut once.

I review issues and read articles of interest until we do our "stand up" which is basically first thing in the morning.

There are too many interruptions before that to spend the effort required to establish a mental model of the problem domain and potential solutions prior to that.