Skip to Main Content
Performance

How to Get The Best from Google PageSpeed Insights

With all Google has done in the past few years around site speed and SEO, it’s no secret now that marketers need to be concerned with improving performance – to stay golden in the eyes of Google, and, more importantly, to meet user expectations. But an optimization project can seem like a steep hill to climb, and so enters Google PageSpeed Insights. Growing in popularity, this handy tool analyzes your mobile and desktop site(s) and provides optimization suggestions. It’s a great tool for getting started with web performance, especially from a multi-screen perspective. PageSpeed Insights also gives you a detailed insight into where the problematic code lives, and which technique you should use to fix it, and provides resources for learning more.

It’s a useful tool, but often users are alarmed by the results they see. Sometimes this is warranted – their sites are miserably slow. But other times it’s not as much the site as it is the tool’s particularities; like when well-performing sites still receive lower-than-expected scores or when the optimization suggestions run counter to how your website was built. Though a good starter tool, PageSpeed Insights can also often cause unnecessary panic, so we’re here to help you make the most out of it without losing your mind.

What Does PageSpeed Actually Score?

Here’s the secret: PageSpeed looks only at the front-end performance of your site based on common optimization techniques, but doesn’t take into account application performance, infrastructure, load handling, or content loaded on the browser. Essentially, the score is an accurate representation of how well your developers built the front-end code, more specifically, how quickly it renders given everything it’s working to display. So while PageSpeed provides actions you can take to optimize the front end further, the tool itself can only score your site performance so accurately without taking into account back-end components and infrastructure. Front-end optimization plays a key part in overall performance and can be the easiest to implement given that it’s at the code level, but it is by no means the only step in creating a well-performing site. Plus, your SEO might not be affected if PageSpeed tells you your site sucks – more on that below!

When a Suggestion Isn’t the Best Solution

So when exactly does Insights get it wrong? Most of the suggestions are best practices and should be implemented, but there are a few that aren’t useful or completely accurate.

Overlooked CDNs

Adding a content delivery network (CDN) can decrease your score, which will sound counterintuitive to anyone who uses a CDN to improve their site’s performance. The reason lies in the front-end-only analytics: CDN implementation can add JavaScript to site code, signifying to PageSpeed Insights that the site might be slower. It could also be that the CDN set a cookie and your code is no longer hosted on a cookie-less domain, bringing down your score. Does this mean you shouldn’t use a CDN? Absolutely not! A CDN is a valuable tool in improving back-end site speed, even if it isn’t reflected in Insights.

Pesky “above-the-fold” content

When PageSpeed Insights prompts you to eliminate render-blocking JavaScript or CSS in “above-the-fold content,” this is targeted at JavaScript or CSS that renders anything you don’t need to scroll to see. It’s a good general rule of thumb to follow, however, marketers today are increasingly designing sites (or at least homepages) that display all of the content above the fold. In cases like these, however, Google still views the page styles as applying to the entire site and not differentiated as above or below the fold. It also represents discrepancies in responsive web design. If you have a responsive site, you are most likely using media queries and other layout styles, so if you eliminate render-blocking CSS first, you’ll produce a flash of unstyled text. This type of disorienting content flash can be more disruptive to the visitor than a few additional milliseconds in render time.

Minifying HTML? Maybe not

A standard suggestion from Insights is to minify your HTML. This isn’t a bad practice to consider, but it is most likely not going to make that much of a difference. While minifying JavaScript and CSS files almost always show big performance gains and faster load times, minifying HTML probably won’t move the needle. In this case, HTML minification would mainly serve to improve your PageSpeed score. If the score is that important to you, it’s worthwhile to invest in, but keep in mind that PageSpeed score doesn’t directly impact your search engine ranking positions (SEPRs)…

PageSpeed and SEO

Google doesn’t actually use the PageSpeed tool to determine SERPs when crawling sites and analyzing performance; they use real user monitoring (RUM), which paints a more accurate picture. Does this mean that instead of using simulations, they measure actual users? browsers and bandwidths, and will not use your PageSpeed results to determine SEO rankings. So take PageSpeed Insights with a grain of salt, and approach performance from more than one angle.

Think about users over numbers

Using a single tool or idea to drive your optimization efforts is never a good idea. Figure out what works by actually testing with your users. Etsy, for instance, optimized according to assumptions about what would make the site better actually resulted in worse conversions. Read more here about what they did next!

Go beyond FEO

Today, front-end optimization is only one piece of the performance puzzle. Application sequencing for just-in-time content, firewall protection, minimized site disruptions, and optimizing dynamic content – these all play an important part in creating a fast and engaging user experience. (You can learn more about the benefits of application sequencing and prioritized content here.)

Use additional resources

PageSpeed Insights shouldn’t be your one-stop for web performance data. Use free testing tools like Websitetest.com and monitoring tools like Yottaa, Pingdom, Keynote, and Gomez to complete the picture and dig deeper into bottlenecks across devices, browsers, and locations.


Yottaa 9 Tips for Benchmarking Your Web Performance Download

Don’t let slow site performance cost you conversions.Let's Talk