This latest release brings to you improved consistency in report timings.

---

How we improved consistency

When a test is launched on GTmetrix, it requires hardware resources (namely CPU resources) to launch a browser, load the page, capture data, and generate a report.

Current GTmetrix test servers all use matching or similar hardware specifications (CPU, memory, etc.) to ensure consistent test results. However, due to varying test server loads, this still can result in some inconsistent page load timings. To improve the page load timing consistency, we’ve added more capacity to handle peak loads and normalized CPU resources during periods of low loads.

With our recent move to Lighthouse, test consistency with regards to timings is even more important, as key metrics like Web Vitals all rely on measurements of time in seconds or milliseconds.

Note: This update is a work in progress, and will be continuously monitored and tweaked as test data comes in.

 

---

 

How does this affect your GTmetrix report?

This change should not affect your Structure Score (or your PageSpeed/YSlow scores on Legacy reports), as it is focused on front-end structural recommendations (i.e. how well your page is built for speed).

However, as for timings, the general answer is – it depends. Generally, timings and Fully Loaded Time will likely increase if your past tests were run during a non-peak load period. If you have set up Alerts that are based on timings, you may need to adjust the timing thresholds.

Other things to note, as with all GTmetrix reports in the past, is that the test server performance is just one factor in page load performance – the other major components include:

• DNS resolution
  The time to resolve a domain name will vary depending on if the domain name has been cached and the performance of your DNS servers (as well as third-party hosts’ DNS servers).
   
• Network conditions
  Due to the nature of the Internet, there may be variations in network conditions (e.g. packet loss, different packet routing) causing fluctuations in timings.
   
• Remote server performance
  The load of the servers serving the resources and the performance of your application may cause variations on connection and waiting times.
   
• Changes in the page
  If your page content changes between page loads (e.g. loads a different advertisement), then this can have a significant impact on the page load timings.
   
• Browser performance
  Due to the variable nature in all of the above and how the browser loads and executes code on the page, the browser will also introduce additional variability into the results.

Taking all the above into account however, changes (if any) should be minimal and you shouldn’t see a drastic swing in numbers.

Note that in the real world, visitors have all sorts of devices varying in performance. A lower-end phone may fully load your page in 2 seconds while a high-end desktop may do it in 0.8 seconds. GTmetrix serves as a more consistent benchmark.

 

---

 

Summary

Consistency in tests is very important to us at GTmetrix, and we’re always striving to deliver accurate performance data you can rely on.

Let us know if you run across any bugs or issues and feel free to tell us your thoughts on Twitter (@gtmetrix) or contact us.

 

---

 

Other updates and changes

Some general changes and housekeeping items:

• Added Section CDN for auto-detection in GTmetrix Use a Content Delivery Network (CDN) audit

 

---