I have been looking at the results and was confused by what is most important. When I make changes or rerun my load speeds don’t seem to improve much but then I noticed the score is better.
So what is the best way to evaluate - the first view/repeat view laod time or Page Speed score?
It’s FAR better to optimize for the actual load times than the scores. The scores are generated from rules that usually will improve performance but the impact for some of the rules can be so small that it’s not even measurable. If you want to share a link to a result we can look at the waterfall and tell you where your biggest bang will be.
At the end of the day, learning to interpret the waterfalls and optimize based on that is the best bet.