Thread Starter
Michael
(@michael8888)
I agree. What makes things worse is that the PageSpeed Insights algorithm is not what Google actually uses to evaluate speed and core web vitals. We often get complaints in the Search Console and when we test with PageSpeed Insights everything appears to be alright. It is a mess. Google wants to enact internet laws and punishes you if you don’t comply, but they don’t really tell you what exactly you did wrong. Reminds me of Kafka’s novel The Trial.
For the time being, we will use the CSS combination / minification feature of W3TC and use Autoptimize for JS. That combination seems to work best at the moment.
Thanks a lot for the help!
What makes things worse is that the PageSpeed Insights algorithm is not what Google actually uses to evaluate speed and core web vitals. We often get complaints in the Search Console and when we test with PageSpeed Insights everything appears to be alright.
Yeah, … that’s the difference between “lab data” which is what a “synthetic” test measures and “field data” which is based on data gathered from Chrome users (the Chrome User Experience Report). Whereas lab data may not flag “render-blocking CSS” (or JS), the field data will still suffer from normal (as in non-deferred) CSS & JS blocking the rendering.
I generally use webpagetest.org to do more in-depth testing (selecting the 3G+ bandwidth profile and a EC2 node nearby the target audience of the site, running 5 or more runs per test).
Thread Starter
Michael
(@michael8888)
Thanks for the tip about webpagetest.org. That looks interesting.