We watched many presentations about speed at Chrome Dev Summit this year . Therefore, it is important to test your site and increase the speed of your site by doing experiments in this area. However, when you do not measure correctly or when you use different tools and get different results for the same pages, things get complicated. We talked about known testing tools in Vol.1 article, but these tools show you the results on a single page URL.
Unfortunately, there is currently no tool that you can use to scan your entire site and directly see which pages your speed score is low. In this article, we will focus more on measuring on multiple pages, comparing with competitors, and tracking. Although optimization is a process applied to the entire site, we realized that there were important business owner database pages that were overlooked in many sites where we conducted speed tests and that it was necessary to identify and optimize them. You can also copy and use the competitor comparison dashboard example we prepared under the Chrome UX report in the second section . Let's start with the Google Analytics tool we use the most.
1. Google Analytics Site Speed Tab
We have all looked at the site speed tab in Google Analytics. It provides us with an important metric, especially the average page load time. Before we delve deeper into the subject, it would be appropriate to make this warning: Don't be fooled!
Google Analytics is a bit ruthless in the site speed tab. The average load time it gives is a value created by sampling, and this sampling value is usually very low.
For example, let's say you have 1000 page views on a page. Google Analytics calculates the average load time of that page from only 4-5 samples and gives the average value. Google also states on the Support.google website that this value is set to a maximum of 1% by default.
I can almost hear you asking, "How can I see my sampling count?" To do this, go to the page timings section under site speed. Click on the icon at the top of the table display options.
And under explorer select technical view.
You will now see the page sample count as well as the average page load time for each of your pages.
As you can see in the example above, it only takes 2 samples from a page with 702 page views and gives the result. We cannot be sure if the loading time of that page can really be 0.64s. Also, you should only collect data from the first pageview, if it includes users who entered faster after the first entry due to cache, the result will mislead you again.
So What Should We Do?
We do not recommend using Analytics data for speed. If you really want to collect and evaluate this data, you should work with Analytics experts to ensure that all adjustments are made properly and the data collected is accurate. The sampling rate is a rate that can be increased. However, increasing this rate alone will not be enough for accurate data.
2. Chrome UX Report
At the 2017 Chrome Web Summit conference, Google announced its Chrome UX report, which is a real-time sample of Chrome user experiences. The Pagespeed Insights tool now provides this data in Field Data.
