Two speed tools, two verdicts, one render-blocking villain

Two speed tools, two verdicts, one render-blocking villain

A publisher chased a green PageSpeed Insights score and got it. Real revenue from organic landing pages kept sliding — down 16% over two months, despite 'fast.'

The investigation ran WebPageTest against PSI side by side. PSI gave one composite verdict; WebPageTest's filmstrip and waterfall told a story. Frame by frame, the hero content appeared at 1.2s — then a third-party consent script executed at 2.1s and blanked the screen for 1.8 seconds, re-rendering everything. PSI's aggregate score smoothed that flash into a footnote. The waterfall showed it as a wall.

Turns out PSI answers 'how fast, roughly?' WebPageTest answers 'what exactly happened, and when?' For a render-blocking, late-executing script, you need the second question.

They deferred the consent script and reserved its layout box to kill the reflow.

The mid-load blank vanished; bounce on landing pages dropped 19%, organic revenue recovered to +11% in six weeks. PSI for a number; WebPageTest for the film. When 'fast' doesn't convert, watch the filmstrip frame by frame.