I’ve been testing a few pages with a PageSpeed checker and the scores swing between runs. A page can score 72 one minute and 81 the next without any changes.
Still, the suggestions about render-blocking resources and image sizing were useful. I fixed a couple of obvious issues and the numbers improved. I try to focus on the recommendations rather than the exact score.
Do you chase a specific score, or focus on real user experience? I’m leaning toward the latter, but clients often want a green number. How do you explain the variability to people who only see a single figure?
Dana Paxson's Writings - Forum
https://danapaxsonstudio.com/forum/viewtopic.php?showtopic=127