TL;DR: it’s not actually slower
Many people use performance tests like GTmetrix, YSlow, and Pingdom that use “Fully Loaded Time” to calculate page load speed. This is a perfectly good way to measure in almost all cases, unless you have video on your website. On web pages with content that maintains a constant relationship with an external source (such as a video), this is an inherently flawed way to measure load speed. Even GTmetrix admits that this measurement has some issues. Speed tests that use this metric are not considering a page fully loaded until the video on your page has fully buffered the first bit of video.
Because these reports are incorrectly calculating the speed of a page and thus, incorrectly measuring a page’s user experience, we recommend using Google’s PageSpeed.
Google calculates the time it takes your page to load using both lab data and real-world field data (if your site has enough traffic). You can learn more about how Google measures page speed here, but this might be all you need to know: the speed score calculated using this report is what Google (the king of search) considers when ranking search results. Why use any other tool to tell you how your site is performing?
📐 What about layout shift (CLS)?
If you’re looking at Core Web Vitals, the SmartVideo player reserves space for itself before it finishes loading - so videos without explicit width and height attributes no longer cause a noticeable jump on first paint. YouTube and Vimeo replacements also keep their original dimensions instead of snapping to a default aspect ratio mid-load. The result: near-zero Cumulative Layout Shift contribution from your videos, even on fluid layouts.
Keywords: GT Metrix, latency, page speed, pagespeed, speedtest, speed test, CLS, layout shift, Core Web Vitals