The Problems with Google PageSpeed Insights

Google provides two tools for testing the performance of your website from a desktop or mobile device. PageSpeed Insights offers desktop and mobile recommendations and is the oldest and most-used while TestMySite is a recent addition that only tests the mobile version of your site.

There are a few serious misconceptions about these tools which frequently cause website owners and even developers to focus on the wrong factors regarding the performance of their sites. Worse, you can spend a lot of money trying to get a good “grade” on these sites without getting any real benefit.

The Problems with Google PageSpeed Insights
One problem with the PageSpeed Insights tool is that, until recently, it didn’t even provide the actual speed of your website in terms of how many seconds it takes to load a page. It has begun to provide this data for some websites but the bigger problem is that the Optimization grade it gives is unrelated to the speed that your site loads. The grade you get is entirely based on how many of Google’s recommended optimizations your site does and not based on it’s actual load time. THIS IS INSANE! Why?

The primary reasons that people optimize the speed of their site are:
1. To provide a good user experience
2. To improve SEO because your ranking in Google search is affected by how fast your site loads

This is CRITICAL to understand. Your ranking in Google is in NO WAY related to the grade that you get in the PageSpeed Insights Tool. It is ONLY related to the actual load time of your site. You could have a site (and many people do) that loads in a couple of seconds in your browser, which is very fast, and yet that site could get a “Poor” grade from PageSpeed Insights. How does this happen?

The optimization suggestions that PageSpeed Insights makes are almost always beneficial tweaks to make to your website. It’s just that, in many instances, they don’t all make a big difference in the load time of your site and sometimes they will actually cause problems. For example a typical recommendation is to “Minify JavaScript.” This simply means that you should use a program that removes all the white space from any JavaScript used on your website. White spaces actually increase the size of a file so removing them can make JavaScript files smaller and therefore download faster. However, the difference this makes is often measured in milliseconds so there is no discernible improvement in the load time of your page but it CAN break the functionality of your site because minifying JavaScript can sometimes lead to it not functioning properly and/or causing errors.

Another example is that Google almost always recommends that you Optimize images and it will downgrade you if you have a lot of images that they believe could be optimized further. The problem is, the more you optimize your photos, the more they degrade in quality. So who is to say what level of optimization you should do? Certainly there are sites that are using images many times larger than they need to be and those sites could be sped up greatly by shrinking their images. However, some sites have optimized their images to the greatest degree that they believe is appropriate for the quality they want to display. Also, more and more sites are providing high resolution images for people with retina displays. Who is Google to say that an image should be shrunk further if it will hurt visitor’s ability to see details in a product, such as a shirt, that you are selling?

So PageSpeed Insights is interpretive in the grades that it bestows. But more problematic is that those grades are unrelated to the actual speed of your website, which is really what counts when it comes to your visitor’s experience of your website as well as your search ranking. Therefore, while these recommendations can sometimes be helpful, don’t freak out if your site gets a poor grade in PageSpeed Insights and don’t assume that this ranking will necessarily affect your SEO. Instead test the ACTUAL load speed of your site and make adjustments according to that (if needed). In the next post, we’ll cover the problems with TestMySite and talk about how to measure the REAL speed of your site.