My site scores 100 in PageSpeed Insights on mobile, and Lighthouse still lists three render-blocking requests on every page. On one run each row said 474 ms. The estimated saving for fixing all three was 50 ms. On the next run it was 0.
That gap is what this post is about. Most guides tell you to inline critical CSS and add defer to your scripts. That advice is fine, but it skips the part that decides whether the work is worth doing: what the insight actually measures. I read the insight's source in Chrome DevTools, ran it six times against my own Next.js 16 site, and compared the fixes I have shipped with the one I tried and rolled back.
The short version: the "Duration" column is how long each file took to download in a simulation, not what you would save. The number that matters is the Est savings line. In a Next.js App Router app the blocking requests are almost always your stylesheet and the small CSS files next/font creates; the JavaScript is already async .…