More feeling please when testing the website speed

Jan Hornung Last updated 21.10.2020
6 Min.
Website speed
Last updated 21.10.2020

The loading time of a website is important. No question about it. For the user experience and the conversion. But for the visitor, it's not the actual load time that matters, but the perceived website speed. Because a user experiences a website without knowing the metrics. Therefore, the perceived loading time is an important value for optimizing the conversion and bounce rate. We show you how to measure it reliably.

Everyone knows it: You have discovered an exciting new product or found the link to a supposedly exciting blog article, click on the link and at first nothing happens. The site remains white for a few seconds and then the page components build up agonizingly slowly. The purchase or the registration for a newsletter is rather unlikely on such a site .

The colleagues from Kissmetrix have already dealt with the connection between page load time and the conversion rate in shops some time ago. The result: Every second counts. Because the average visitor is impatient. A loading time that is one second longer can result in a seven percent lower conversion rate. With every second that your site takes longer to load, you lose sales.

An antidote can be the Above the Fold optimization. This means optimizing the loading time of the area of site that is visible without scrolling. This measure reduces the perceived loading time, which in turn can lead to better conversion rates.

But how can you properly measure perceived web page speed? Sure, you can just load the site in an incognito tab and try to ... well, feel the difference. Or you can rely on a tool that gives you the perceived load time as a tangible and comparable value.

GTmetrix is such a tool. It allows you to measure the, in my opinion, most important load time value of your site and makes it comparable. It is very easy to use and provides valuable measurement data.

This article is part of a series on load time measurement:

  1. In Part 1, we show why the type of data a test provides is so important
  2. In part 2 I show why the popular PageSpeed Insights test is absolutely unsuitable for performance comparisons.
  3. Part 3 shows you which data you need to make a good performance comparison - e.g. after a hoster change.

GTmetrix: Lots of optimization recommendations, website speed history and video feature.

The operation of GTmetrix is very simple: You enter the URL of the site to be tested and the tool spits out the result after a few moments. The disadvantage:

  • GTmetrix offers only one server location, Vancouver
  • GTmetrix tests only with Firefox

This means that the measurement results of GTmetrix are always valid for a visitor from Vancouver who accesses yours site via a desktop computer and Firefox. As long as you work exclusively with GTmetrix for your performance measurement, this does not affect the results. But if you want to know how fast it site loads for European visitors, you have to use another tool.

GTmetrix gives you four large data packets:

  1. The overview with the potential analyses and important key figures
  2. The waterfall diagram
  3. The (chargeable) video function
  4. The time series analysis

The special feature of the overview: it provides the results of two potential analyses at the same time

In this overview, all important key data of your site are displayed. In addition to the potential analyses from Google (1) and Yahoo (2), you will find the Page Load Time (3), i.e. the loading time of your site , the size of your site (4) and the number of HTTP requests under Page Details. (5)

By the way, the load time and the HTTP requests are particularly suitable for the performance comparison of two websites or website versions.

Measure website speed correctly with GTmetrix: An overview
Under "Test Server Region" and "Using" is clearly visible: In the free version GTmetrix uses a server in Canada and the Firefox browser to determine the website speed. For European sites this test setup means longer loading times.

This initial overview also includes the detailed results of the two potential analyses PageSpeed Insights and YSlow. These show exactly how well the site has been optimized so far and where there is still room for improvement. The fact that GTmetrix uses two different sets of rules (one from Google and one from Yahoo) is commendable. Because this offers you as a user more orientation. If an optimization measure is suggested by both tests, it is most likely an important optimization step that is worth working on.

The perceived website speed is important. The overview of the potential analysis of GTmetrix shows you where there is potential for improvement.
This is what potential analysis looks like at GTmetrix. Nothing new, if you have already seen a Google PageSpeed Insights result or the results of YSlow. Via a dropdown you can show detailed information about the individual steps.

In its help section, GTmetrix offers detailed descriptions and interpretation aids. This makes it very easy, especially for beginners, to familiarize themselves with the complex result structures.

The waterfall diagram is the heart of the analysis

Just like Pingdom, GTmetrix only reveals the really interesting data in the so-called waterfall diagram. This breaks down all loaded data packets (1) in chronological order (2) and with their loading time (3). In addition, the type of request is displayed (4), whether the respective request is OK (5) and whether the server issues an error (6).

Detail view of a waterfall chart from GTmetrix for website speed analysis.
The waterfall diagram shows in detail which factors have which share in the website speed.

This detailed information now allows you to make reasonable performance comparisons between two sites . This becomes relevant, for example, after a change of hoster or an onpage optimization. It is important that the measurements always take place from servers in Canada and via the Firefox browser. If you want to test other parameters, you either have to buy GTmetrix or use another tool, for example webpagetest.org.

The detailed view of the loading time provides an even more accurate picture of the individual data packets and their contribution to the website speed. In the example: the first contact with our server. 5 milliseconds are needed for the DNS lookup, 167 milliseconds are needed for the client to connect to the web server, after 172 milliseconds the first data is downloaded. The long waiting time of more than a second is partly due to the geographical distance of the test server from the actual server location. The test therefore provides the data for a customer from Canada.

Details from a waterfall diagram to analyze website speed
Detailed breakdown of the browser's first request to the web server.

Key benefits of GTmetrix: Time series analysis and video feature for perceived load time.

However, compared to the tool I introduced a few days ago, Pingdom, GTmetrix offers two distinct advantages:

  1. a free time series analysis
  2. a paid video analysis

1) The time series analysis

Comparing the loading time of your site over several days or weeks can be very revealing. For example, if you are optimizing your site step by step over several weeks. The history function provides this time series information (in the free version) for three dimensions: The load time of site , the number of HTTP requests and the size of site and the scores of the potential analyses YSlow and PageSpeed Insights.

Time series comparison of website speed at GTmetrix
Not shown in the picture: the time series analysis of the potential analyses. However, these are not relevant for the website speed at first.

2) The video analysis - the only real indication of the perceived page load time unfortunately chargeable

Now we finally come to the feature that makes GTmetrix so valuable. And also better than Pingdom, for example: the video function. In such a video, the page structure is shown. Logical. But this simple trick allows to get a clear differentiation between the measured load time and the felt load time. Unfortunately, the tool is only accessible after registering with GTmetrix (and probably only after purchasing the service). This doesn't have to be, because webpagetest.org offers this video function for free, for example.

Conclusion: A real professional measurement needs information about the perceived charging time

For the analysis of the loading time and especially the loading time experience of the users, the video function is elementary. Because it allows comparisons on a level abstracted from the bare data. We can only recommend tests that allow video analysis. Because in some circumstances, the measured loading time is quite high, but the perceived loading time is absolutely justifiable. The value allows you not only to better understand how customers or users perceive your offer, but also whether a page optimization with a view to performance is worthwhile at all. Because under certain circumstances it is worth investing more in conversion optimization than in the loading time of the site .

All in all, GTmetrix is another good tool to measure the loading time of site . It offers a bit more information than Pingdom, since the free version of GTmetrix includes two potential analyses. The really interesting features are hidden behind the paywall, but are basically there, so that you can understand the performance of your site very well with GTmetrix and also monitor its development.

But for us GTmetrix is not the number one of the free tools. Because it can be even better: webpagetest.org is for us the non-plus-ultra of the free measurement tools. However, the tool also makes the greatest demands on the user.

Have you had any experience with other measurement tools that might provide even better data?

RAIDBOXER from the very start and our Head of Support. He loves talking about PageSpeed and website performance at BarCamps and WordCamps. The best way to bribe him is with an espresso – or a Bavarian pretzel.

Comments on this article

Post a comment

Your email address will not be published. Required fields are marked with *.