Speed test with Test My Site
Google has a TestMySite website testing tool as part of their Think With Google project, to help emphasize the importance of having a fast loading website and help webmasters optimize their websites for speed. The tool that Google uses is based on the WebPageTest.org project, which is a service that scans web pages and performs a large number of tests to measure the overall performance of the page.
The test that WebPageTest performs is more comprehensive than the PageSpeed Insights test that Google also promotes. Yes, Google is using two different tools for testing and comparing the speed of websites with.
The main item that Google reports in the WebPageTest is how fast, in seconds, your website loads and then compares that to an internal study they’ve done to understand how much potential traffic you might be losing from users clicking back on their browsers before your page loads.
Difference between the two tests
WebPageTest.org is an open source project that is owned and maintained by a few folks from Google but is not an official tool used by Google. WebPageTest aims to be a comprehensive test to look at many different aspects of how quickly a full web page loads by real-life browsers. They look at every single request that a browser makes (such as image downloads) and analyze how it can be improved. They also allow you to test with slower internet speeds (such as 3G, DSL, etc..), different devices (mobile phones, tablets, etc..) or different physical locations (Australia, Europe, US, etc..).
The main output of the web page test is a speed index score (lower the better) that rates how quickly your website displays visual content, for the testing conditions. While the test is comprehensive, it is a very complex tool that gives a more realistic view of how your website performs in the wild.
PageSpeed Insights is a much more focused set of tests and will most likely be the tool that Google will use in the future to rank website speed for its Search Rankings algorithm. The primary goal of PageSpeed score is to test how Above the Fold content performs and loads for visitors. In other words, this test is only focused on the first view a user will see once they visit your page, especially from a Google Search Results Page (SERP).
Unlike WebPageTest, PageSpeed insights does not offer many different scenarios to test with (location, device, internet speed), but instead analyzes websites based on a set of best practices, such as moving scripts to the footer, placing critical rendering CSS inline in the HTML, etc.. To learn more about this and what Duda does for PageSpeed, read here. They then offer a limited set of recommendations to help you make your websites load & perform better.
What does this mean for my Duda website?
The first thing to remember is that, like all tests, the results are subjective and opinionated to the creator of the test. No test online is 100% comprehensive. Some warnings are important, some are minimal and it’s important you take stock of what actually is being recommended and make the decision for yourself if it’s worth fixing.
Duda websites are automatically optimized for the PageSpeed test. This means we’ve done much of the heavy lifting needed to score and rank well on this test. But, even the work that Duda has done has limitations.
Testing Google Load Time
Duda wants to provide you with the best performing websites. In order to give our customers the best sites, we tested the load time for 10 Duda sites based on different templates using Google’s TestMySite feature and calculated the change in load time depending on whether a site is secure or non-secure (which affects whether the site is HTTP or HTTPS). In 2015, Google began an HTTPS movement which is part of their long-term plan to increase web security and encourage web builders to secure their sites.
After testing the sites and compiling the data, we discovered that non-secured, non-www sites had the slowest load time at an average of 9.5 seconds. Some of these sites also returned with an error or timed out during the test. When sites lack a subdomain, they are redirected to the appropriate URL with a subdomain, which drastically increases load time. Secured sites had a significant better load time, averaging 5.5 seconds. Non-secured sites with a subdomain had the best results, experiencing a load time of 4.5 seconds. The data is displayed in the chart below:
Why do non-secured sites load faster than secured sites?
Even though Google encourages secure sites, enabling SSL slightly increases load time. This is because SSL negotiation adds time when the site is loading in order to create the SSL connection. If you choose to force SSL on your users, it further increases the load time of sites because it redirects the non-secured site (http://www.example.com) to the SSL secured site (https://www.example.com). Whenever a redirect is added to a site, the load time also increases. With non-secured sites, there is no need to load the SSL connection for the site, so the load time is only determined by the content on the site.
Is it important to disable SSL to improve load times?
Even though disabling SSL increases the load time of your site, it is still important to have SSL enabled. Not only does SSL increase your site’s ranking and SEO standards on Google’s search engine, but it also protects users that submit information in sites; for example, when users enter payment information to purchase a product on your site. Users entering sensitive information on non-secure sites will be notified and be recommended to avoid entering in information. Sites that have SSL enabled are not destined to have high load times; load times can also be affected by other common issues.
Common Issues and Workarounds
For the WebPageTest, Duda websites do well, but here are some common issues and how to fix them:
- Page Size too big: While the report does not state this directly, if your website takes over 6s to load, your page is probably too big. The number one cause of this is having too many images.
Online today, images usually take up about 70% of the overall downloaded content of a website, this, of course, becomes a larger percentage if you have a very image heavy page. Duda recommends limiting the number of full-sized images used on a single page to 5, and if you must add more, try and place them lower on the page to not impact the first-load experience. A full sized image means one that takes up 50% of the width of the screen or more (icons and thumbnails are probably fine and will have a smaller impact).
Here are a few other items that might increase the weight or size of your page:
- Videos, either an embed, video background or one you’re hosting can drastically increase the page size.
- Image sliders: Using an image slider multiplies the number of pages on a site.
- A few other items to look at removing are videos, image sliders, and image galleries from the page. These often add a lot of unnecessary weight to the page that TestMySite would markdown.
- Multiple landing page redirects: Try re-running the test with the www version of your website with https:// in front of the URL. For example, if you tested with dudaexample.com, put in: https://www.dudaexample.com in place first (HTTPS, assuming you have an SSL cert installed, which you should).
- Fonts: You should use two fonts on your website, maximum. Each font you select requires additional code to be loaded from the Google font library and slows down how fast your website loads overall.