Performance optimisation of YouTube’s iframe for embedding video

I created this case study to show how you can optimise the performance of embedded video on your website.

Always be testing

I recently completed an upgrade (complete overhaul) of the front-end of a website. While testing it’s performance, the page load time and download size alarmed me. The website had an embedded YouTube video in the hero on the landing page (as they so often do). I had used the default code that YouTube provides for embedding a video with an inline <iframe>. Looking at the requested files, it was clear to see how these 3rd party scripts had a negative impact.

These are the services I use for performance testing:

My initial thought was to display a placeholder image with an external link to the video on YouTube. But the ability to watch the video inline on the website is likely to be a better user experience. The website used to load the video in a modal (ugh, more bloat). This got me thinking: include the <iframe> when the user wants to play the video.

Measuring performance

After a little bit of digging around online, I came across the solution on SitePoint. Other examples are available on Tuts+ and Labnol.

Needless to say, it was not the first time someone had thought of this. Why I felt compelled to reproduce it, instead of copying it and moving on, was the desire to measure the change.

It is also easier to quantify the differences by eliminating all other factors. So I set out to create a case study, detailing the different methods for embedding a YouTube video.

1. The inline <iframe>

The first method embeds the video with an inline <iframe>. YouTube provides this code for you and makes embedding videos a simple affair. As mentioned, the downside to this is that you force the download of all the necessary scripts on page load.

2. Using JavaScript

In the second method, JavaScript inserts an <iframe> into the page when you click to play the video. A few lines of CSS provide styling to create a play “button” via a pseudo element. YouTube allows you to access a cover image for any video and the figure element uses this image.

After clicking play, the <iframe> gets inserted and replaces the figure. Only then are the 3rd party scripts that YouTube requires, downloaded.

The results

In these two basic examples, the load time for the JavaScript method is under 1 second. It downloads about 60 KB for a full page load. In comparison, the inline <iframe> almost takes 3 seconds and over half a megabyte to download.

Another major benefit is that using JavaScript results in 10 fewer network requests!

The code is all available on GitHub.


Could my code be better? I have no doubt. But it does a good job of illustrating an alternative way of dealing with an inline <iframe>.

Further reading on web performance