It was recently revealed that Google uses data from Chrome User Experience Report to measure page speed as a ranking factor. As usual, Google only gave us one piece of the puzzle, but it’s definitely something valuable to know and use to our advantage.
With that in mind, I decided to dive deep into the current status of mobile page speed in Ireland to find out how the most popular websites in the country are performing, where they stand against their competitors and, hopefully, open a conversation on how to serve faster and better pages for everyone.
Below are the steps taken to find the data and the methodology of this study. If you prefer, jump directly to the main results.
Before showing the findings, I believe it’s relevant to explain the methodology behind this study, both to corroborate the results and to understand the limitations of it.
The study includes three main sources of data:
- Similar Web: the source of the top 100 most popular websites in Ireland
- Google PageSpeed Insights: source from the first two speed measurements, First Contentful Paint (FCP) and DOM Content Loaded (DCL)
- WebPageTest.org: the source of the fully loaded page data, number of requests and page size. The results are the average of two tests and had a Chrome browser on a Motorola G (gen 4), over a 4G connection in Ireland
The following filters:
- The page benchmarked was always the homepage as it appears on Google mobile for Irish users. If the homepage is a subdomain, then this URL was used as a source
- Duplicate homepages were ignored (e.g. Google.com and Google.ie) and the next website on the list was considered
Since January, metrics from the Chrome User Experience Report have been available on the PageSpeed Insights report. This is a free and simple to use tool, where you can place any URL and have the information for the First Contentful Paint (FCP) and DOM Content Loaded (DLC). Here’s Google’s explanation for these metrics:
First Contentful Paint (FCP) measures when a user sees a visual response from the page. Faster times are more likely to keep users engaged. DOM Content Loaded (DCL) measures when HTML document has been loaded and parsed.
From WebPageTest.org, the third and last metric is the fully loaded page, simulated this over a 4G connection on a Moto G (Gen 4) using Chrome as a browser. This is what Google uses as a reference on the Test Your Mobile Speed, so I decided to use the same.
Another relevant point to highlight is that while the FCP and DCL are using real data from users, the fully loaded numbers are a simulated test, isolated on a 4G connection in Ireland. As real users will be using pages in all sorts of conditions (3G, 4G and from different locations), the numbers won’t perfectly match, however, the treatment is the same for all.
There are other speed moments such as Time to First Byte (TTTF) and Time to interactive, but I decided to keep it simple and focus on the metrics from the PageSpeed report.
The findings below give a simplified view of how the homepages of the most popular websites in Ireland performs in several niches. My guess is that some of the findings are likely to be similar if this test was done in another country, but nevertheless, there are learnings to take.
The pages were classified in 13 niches, based on the first level categorisation done by SimilarWeb. The top performers in each of the conclusions are the relative top half (the top 6 niches).
I found out that:
Most websites don’t score a “good” grade on PageSpeed Insights for mobile. To be considered good, a page should score at least 80 out of 100 points. There is a number of optimisations that can be done, and by experience, often it’s possible to go achieve the 80. The tool will straight away inform where the page was approved and where not, so you can start taking action.
The PageSpeed score looks at some elements such as the number of CSS/JS files requested, image weight, redirects and other items, but doesn’t actually measure the loading speed.
It shouldn’t be an obsession to score higher if this means serving a worse page or losing an important functionality. You should give more priority to metrics such as loading times, requests and size first. If you do have a need for a 100 score, check how Felix Tarcomnicu managed to score the highest on this blog post on Moz.
Only 2 out of the 13 niches have average scores over 80. Even looking at the 100 homepages individually, only 33% managed to score 80 or more. Overall, there is definitively space for improvement.
Faster First Content Load (FCL) and DOM Loaded Content (DLC) can help the following loading stages, but they don’t always do the trick alone. Looking at the top half of the list sorted based pm lower FCL, only 50% of them (3 niches) made the top in all loading stages. The conclusion is: optimise for all stages on mobile.
Google recommends up to 50 requests on a page. In this case, only 34% of the mobile pages achieved the recommended. The numbers kept growing until an impressive number of 709 requests on a single mobile homepage!
Size matters? Yes, but in this case, the smaller, the better. Sorted by page size, 4 out of the 6 best performing niches have in common lighter pages and quick fully loading time.
Google’s best practice here is a maximum page size of 500kb. Obviously, a page might have several other reasons to be large and shouldn’t be an obsession to decrease the size for the sake of it. However, based on my own experience, items like a high number of requests, heavy images and unnecessary trackers contribute to a heavier page.
Want to find out how many ad trackers are loading on a page? Use Ghostery to review which trackers are still being used. Trackers will increase your requests and consequently, the page size and loading times, in a cascade effect.
Extras: https, mobile pages and redirects
While this study had as a main goal to show the mobile speed of websites in Ireland, I also have a few more curious findings on the side.
Separate mobile websites are still a thing: 13% of the top 100 most popular websites in Ireland are still using separate mobile websites. If Google had decided to switch to the mobile-first index straight away for everyone, I don’t know what could happen to them, but it’s unlikely that it would be positive.
Homepage redirecting to a subdomain: 8% of the mobile homepages are, in fact, subdomains (excluding the separate mobile websites). I can’t imagine a good reason for a move like this from an SEO or any other perspective. These brands promote their websites as any other brand, using a normal “domain.com” or “www.domain.com” structure, but the users have to wait for another second while redirected.
Usage of www vs non-www: 66 websites are using www as their preferred version. This is purely a matter of choice. I personally prefer non-www and to be honest, only discovered the difference once I started working on SEO. It makes sense as a lot of people recognise a website with the “www” at the start.
Http vs Https: only 12 websites have not adopted https, which by now has become a web standard (one year ago, 50% of page 1 results were https; the number should be higher now). Curiously, this list includes an eCommerce that only has https in pages where customer data is collected, therefore not taking advantage of https as a ranking factor (or at least as a tie breaker as some people say).
It’s safe to say that the most popular websites in Ireland do a good job, but there is certainly room to take mobile speed a level up. We have all been through the frustration of looking at a small blank screen for a few seconds and feel like the eternity is upon us, until deciding to do something else.
The free tools Google provides to webmasters are a good way to find the gaps and start saving a few seconds here and there. Check where you can make cuts, and happy optimising!