What’s it like to manage a website today? Complex. How about keeping up with the latest optimization techniques and tools? Daunting. Now imagine your optimization efforts are graded out of 100 and you’re given separate scores that are 20 points apart. Head. Spinning.
That’s how you might feel after using the three Google PageSpeed Insights tools; Web version, Chrome extension, and Firefox extension, to score your site’s performance and getting varied results. Google PageSpeed Insights is one of the most popular tools to track your optimization efforts, however, using two or all three tools could confuse more than help.
Testing the Tools
We dug into the scores to help you read and understand Google PageSpeed results to gain a clearer sense of your site’s current performance and areas for optimization. We tested all three tools for Alexa’s top 10 websites and here are the results:
|URL||Insights (Web) Score||Chrome Score||Firefox Score|
Why am I getting an “A” and a “C” on the same test?
As you see, there’s a large disparity (up to 20%), in the scores for each site across the tools. One of the key reasons is the tools use different formulas or rules for scoring. Based on Google Developer’s FAQ section, both PageSpeed Insights Web version and the Chrome extension use impact-based (weighted) scoring to determine the score. On the other hand, the Firefox extension uses an average of all of the individual rule scores. Adding further discrepancy, the Web version has the most up-to-date rules, while the browser extensions are lagging on updates.
CAUTION: A High PageSpeed Score Does Not Mean Faster Load time
A common pitfall for PageSpeed newbies is to think that a higher Google PageSpeed score means a faster page load. We previously covered research on top Internet Retailers that showed PageSpeed is not correlated to actual end user speed. Pages with a low PageSpeed Score may load faster than pages with a PageSpeed score of 100.
For this article, we decided to repeat the research on the browser extension tests. The table below shows the time for Document Complete for the two browsers using a 100Mbps of fiber-based corporate connection, hence the low numbers. As you can see, there are no correlations when you compare the time for Document Complete with different websites.
|URL||Chrome Score||Document Complete (Chrome)||Firefox Score||Document Complete (Firefox)|
|Google.com||97||0.48 s||96||0.48 s|
|Facebook.com||100||0.60 s||96||0.61 s|
|Youtube.com||77||1.04 s||85||1.15 s|
|Yahoo.com||79||1.23 s||92||1.45 s|
|Baidu.com||100||2.41 s||97||1.19 s|
|Wikipedia.org||89||0.51 s||75||0.28 s|
|Qq.com||86||18.73 s||91||6.26 s|
|Linkedin.com||93||0.84 s||94||0.53 s|
|Live.com||89||0.87 s||91||0.64 s|
|Twitter.com||100||0.41 s||98||0.47 s|
Rather than measure and benchmark page speed, PageSpeed Score grades how well your webpage is utilizing web performance optimization techniques. The score is useful for anyone handling Front-End Development. Since the score doesn’t correlate with page load time, it’s essential for an organization to look at other factors that impact speed.
Four Tips to Using Google PageSpeed Scores:
- Stick to the Web version only – Not only do we suggest using just one of the tools for consistent scoring to benchmark progress, but Google suggests using the Web version of Insights while they update the extensions.**
- Rely on rules, not scores, when using different tools – PageSpeed Scores are meant to offer actionable suggestions to optimize your site’s front-end architecture. Focus on optimization, not the 2 or 3 digit number.
- Don’t be discouraged by a low score – Your score isn’t not necessarily correlated with speed, but it does highlight areas for improvement. Implement the suggested optimizations and you’ll likely improve performance no matter how fast or slow your site is.
- Don’t get carried away with a high score – There are countless factors that impact the speed to load a web page and there’s always room for improvement.
** UPDATED 1/23/2014 – Ilya Grigorik (Developer Advocate at Google) has confirmed the browser extensions are way behind on updates and recommends the use of the Web version of PageSpeed Insights, which is more up-to-date.