This report lets you analyze the speed at which site pages are loaded. A browser loads pages in several stages. The loading speed may decrease at any one of the stages.
The report makes it possible to analyze:
Time is shown in seconds. Since the loading speed may depend on the browser, we recommend using browser segmentation.
DNS (DNS request processing) — The time spent processing requests to the DNS server when loading a page. Only values with non-zero times are counted.
This parameter is affected by the speed of the DNS servers that handle requests to the site's domain.
Time to establish connection — How long the browser waited for a connection to the HTTP server when getting page content. Only values with non-zero times are counted.
This parameter is affected by network latency between a user and the site, as well as by how heavily loaded the site is with incoming requests.
Server response — How long it took to send a request and get a response with page contents from the HTTP server to the browser. Only values with non-zero times are counted.
This parameter is affected by how quickly the response is generated by the web server, the page size, and the speed of data transmission between a user and the web server.
HTML load and parsing time — The time it takes for the browser to process the page contents after downloading it from the server and before rendering.
This parameter is affected by the size and complexity of the HTML page (complexity includes the number of tags and how many nesting levels there are).
This is the value that users subjectively perceive as the site's loading speed. The time to rendering usually takes less than two seconds.
Time to DOM loading — The time from starting to go to the page until the page has completely loaded with all its components (images, CSS, scripts, and so on). This is the value that users subjectively perceive as the page's “quality”.
Time to rendering — The time from starting to go to the page until the page starts being rendered in the browser.
When optimizing loading time, pay attention to the parameter that is most time-consuming and optimize it first.
In the report, data is interpreted using a quantile. By default, the 50% quantile is selected. This means that 50% of the time, page loading time is less than shown in the report. The 90% quantile means that download time will not exceed the time shown in 90% of cases.
For example, if we take ten random loadings of a page, on average nine of them will load faster than the time shown in the report, and one will take longer.
You can set the desired quantile yourself. The higher it is, the more likely that the user can load the page in the time specified in the report.