How accurate is WebpageTest for identifying performance issues? WebpageTest is claiming a 6-second page-load time for MediaWiki, but Google PageSpeed Insights says that is “good” with a 90% performance rating. https://developers.google.com/speed/pagespeed/insights/
Here is a pretty huge difference: https://sandbox.nicolesharp.net/wiki/project:sandbox/gallery
WebpageTest says it takes up to 105 seconds (!!!) for the first-view loadtime. But Google PageSpeed Insights says that is “medium” performance with a score of 78%, and actually “good” mobile performance, with a mobile score of 84%.
Looking at the WebpageTest results, it would seem like there are serious performance problems that would make it difficult for people to access the website. But Google PageSpeed Insights says that the website has “medium” to “good” performance, with average performance scores above 75%.
Here is the exact same page hosted by Wikimedia: https://www.mediawiki.org/wiki/user:Nicole_Sharp/sandbox/gallery
WebpageTest says that the page on MediaWiki.org takes a mere 1 second to load, but Google PageSpeed Insights gives a nearly identical score (only a 1% difference) between the page hosted on MediaWiki.org versus NicoleSharp.net.
So if I only go by the Google PageSpeed Insights score, it would seem that my website performs similarly to Wikimedia’s MediaWiki.org. But WebpageTest provides a dramatically different conclusion that is much more worrisome. What might be causing the difference? Should I trust Google over WebpageTest?
How do we test the performance of web page when the URL is same for every other page in an application
Any two speed testing services are going to give you different results. They have different methods of testing, are ran on different networks, etc. WPT allows you to simulate multiple devices and throttle network bandwidth, while GPSI just gives you a desktop and mobile result.
That being said, I have noticed some inconsistent results within the public WPT.org. If you love WPT but want it more consistent, try setting up a private instance, or use a premium monitoring service like SpeedCurve or MachMetrics.
Important thing is to pick one tool and check it over time, that’s the only way to ensure you’re tackling your page speed from both an offensive and defensive approach.
These tools have a slightly different way of looking at things which is why you’re seeing different results. At the end of the day these tools are just guidelines. The real goal is to make the site as fast as possible for the customer using the site.
Time to first byte differences could be how they are measuring or it could be load on the server at different times. Im not sure in the differences in the SSL negotiation. Google should be talking that into account with their calculations as that would impact time to first byte.
A good way to see time to first byte is to use the browsers developer tools, and look at the waiting time on the page itself. That is the time it is taking Miva to process all its data and return information back to the browser
kadammanali987 PageSpeed Insights are not the same as the internal scrores of Google’s Core Vitals!
Webpagetest speed delivers accurate results when you use the settings of Google:
Chrome Emulated Nexus 5 - 3G Fast Device (1.6 Mbps / 768 Kbps 150 ms RTT), etc
How to setup on Webpagetest:
Classifying Good, Needs Improvement, Carmon Poor for Core Web Vitals: Classifying Good, Needs Improvement, Carmon Poor https://developers.google.com/speed/docs/insights/v5/about#categories.
P.S.; I test the site in your signature. You have a long way to go: https://www.webpagetest.org/result/200719_AH_f134ceffa623e3ed87fbf94f39e86591/
Core advise? Think Mobile First + Use X-DNS.