Improving site speed is crucially important, Craig at Distilled recently wrote a phenomenal post site speed for dummies which really breaks down WHY improving site speed is a high ROI activity and there was a great Youmoz post recently published which explains HOW to improve page load speed: Optimizing Page Speed - Actionable Tips For SEOs and Web Developers.

So understandably when Google Analytics released a new feature that lets you track your pages load speed right within their interface I got crazy excited. This is huge news and I think is another big win for Google pushing people to create faster websites. It was so exciting I really couldn't contain myself when I heard the news - here's my tweet stream:

SEOmoz Data

As soon as I heard the news I grabbed Casey and got him to install the required snippet of code on the SEOmoz site so we can start tracking our load speed. Thanks to Casey moving on this so quickly we have results from about a day and a half of tracking. Here's the data page by page:

And here's the summary stats for the same time period:

As you can see, our load speed is varying quite wildly but I think this is largely down to the small sample size. That said, we clearly have some work to do. When you add this tracking code it actually makes another call to the GA servers so it slows your page down a tiny bit and therefore it only fires on a sample of you traffic. Explained here:

Be aware that enabling Site Speed tracking for your site results in an extra request made to the Google Analytics servers, separate from the page tracking GIF request made for pageviews. To minimize impact on overall site latency, this request is sent only on a sampling of pageviews to your site. For this reason, you might not always see the site speed GIF request when attempting to debug this feature in Firebug or other tools. This is normal and expected.

Why this is important

Here's just a few reasons this is big news and some of the things I'm really looking forward to playing around with. You can now:

  1. Track load speed by page type easily and efficiently
  2. Measure the impact of load speed on conversion rates
  3. Calculate load speed for pages behind a login
  4. Segment load speed by user-type (i.e. custom variables) such as registered and non-registered users
  5. Analyze load speed by geographic region, browser, etc

Supporting Resources

Has anyone else been able to install the code? What are you seeing? I'd love for you to share screenshots in the comments below!


Thinking of attending a conference anytime soon? You should definitely check out the Distilled Pro Seminar in Boston 16th/17th May and the SEOmoz Mozcon in Seattle July 27th - 29th.