choose a topic that is important (to the world, to your community, to you, (to the funding agency))
focus on one core topic
find a good general question that guides your research - even if you will not be able to answer the question in the end
try to understand the full stack of your research question, prototypes, experiments (e.g., when investigating touch screen latency, aim to understand latency of nervous system, input and output hardware, operating system, applications, and measuring devices - even if your research focuses on only one of these aspects)
do not take shortcuts - read the literature, investigate effects that you do not understand, don't hide limitations
PhD allows for longitudinal studies and iteration