English Premier League Websites Analysis
A comparison of the ‘quality’ of the 20 Premier League Websites
We recently compared the ‘quality’ of the websites of the world’s top 20 universities (according to the Times Higher Education rankings) with those of the 20 football clubs of the English Premier League (PL).
At first glance, the two sets of ‘institutions’ appear to have little in common. But, both address large, diverse, geographically dispersed audiences searching for information or looking to become “involved with” the institution. And, both sets of sites facilitate audience transactions: for merchandise, season’s tickets, event or conference registrations or to pay fees.
The original blog post tested our hypothesis that football’s superior financial resources result in ‘better’ websites. Our blog post concludes this is not the case. One interpretation of the results of comparing the two sets of sites is that quality differences arise from fastidious website configuration decisions and rigorous maintenance processes rather than the size or resources available to support teams.
Why does website quality matter? Because, end user surveys and personal experience confirm a preference for fast, friendly and functional websites. Sites that don’t meet these criteria frustrate users, communicate disdain and eventually erode brand appeal: singly and collectively compelling reasons to get this right.
As has been pointed out in many places across the internet: minimising the time a consumer spends to achieve their objective on a site is key to improving the customer experience. Much as time wasting on the pitch gets a yellow card, so it should online as well.
We report on three aspects of Premier League football team website quality and the potential impacts on a visitor’s experience. While we address technical issues, these all potentially impact the experience a website delivers.
The rest of the report is about a ten minute read. If you prefer to download and read long articles off-line, you’ll find a PDF version here: https://goo.gl/dluzI0
We collected and analysed the data over a five-day period in late March 2017. We used our eQAfy service to scan the first 1,000 pages of each PL team’s website. We gathered data about server configuration, page structure, social media links, SEO set-up and broken links. We also used Qualsys SSL Labs’ test to verify HTTPS implementations, where relevant, and the Pingdom and WebPageTest services to measure page loading speed and count server requests.
PL teams constantly update their website’s content. It is likely that some of the data we report will have changed by publication date. However, we believe a current visitor’s experience of these sites will be substantially consistent with our findings.
To put the subsequent data in context we estimated each site’s size, in numbers of pages that can be accessed by users and we attempted to identify the content management systems being used to host each site. The latter choice shapes a site’s ability to provide features, such working well on mobile devices or integrating with third party applications.
Size — Number of Pages
We used Google to estimate the total number of indexed pages for each website. The overall size of a website offers a sense of the complexity of maintaining a site in good working order. Our guide describes how to estimate the total number of pages for a website.
Figure 1 shows the relative sizes of the websites. Everton’s site has an estimated 1.05 million web pages, almost 500 times larger than Watford’s site with an estimated 2,200 pages. Collectively, the sites cluster into three: four ‘large’ sites of over 250,000 pages; four ‘medium’ sites with between 25,000 and 70,000 pages and twelve ‘small’ sites with fewer than 25,000 pages.
With the exception of the four large sites, the rest are comparable to faculty or divisional sites within a university: substantial content repositories, but not unruly.
Content Management Systems
These websites support three broad roles: to inform, influence and enable interaction. To do so they need to manage different types of content, readily facilitate updates and integrate external resources and services. Their content management systems (CMSs) support these features.
Every organisation makes a choice of whether to use an open source or a proprietary content management system: there are arguments in favour of either approach. The higher education world has gone open source, while PL teams have gone proprietary.
Table 1 shows our ‘educated’ guesses for the PL website CMSs.
Football has yet to be as enthusiastic as other ‘industries’ in adopting open-source solutions: just two of the sites run Drupal. On the other hand, SiteCore, which emphasises customer experience (or personalisation) rather than simply serving out content, has made headway with three PL clubs.
The balance of the PL websites seem to run on versions of Ektron CMS (an ‘educated guess’), and we understand from a press release that a transition is underway to an Episerver platform in 2017. Episerver and Ektron merged in January 2015.
Feel free to contact us for any corrections: firstname.lastname@example.org
Until recently the web world has been slow in migrating from ‘in secure’ user website connections to secure website connections: HTTPS adoption. Starting in 2017 HTTPS adoption is set to accelerate. Google has stated that its Chrome browser will move to warning users when they connect to non-HTTPS sites.
This may sound purely technical and implementation is highly technical, but the implications for website visitors are profound. HTTPS connections ensure that no one can intercept a connection to a website or know which pages are being browsed: in other words, visitors can be assured they are on the ‘official website’ and their privacy is maintained at all times.
Only four of the 20 PL websites are HTTPS enabled at the time of writing. We tested the quality of the HTTPS implementations using SSL Labs’ tool.
1. The balance of the PL sites should move to HTTPS in 2017.
2. The existing HTTPS-enabled sites should re-run the SSL Labs test and implement the highlighted actions to move to an A+ rating.
3. All sites should invest in an Extended Validation certificate: you can see our EV certificate in the URL bar here, or view Twitter’s here. Consider an EV certificate displayed in the URL bar as brand reinforcement.
The multi-dimensional factors influencing page loading speed mean that measurement has to be done carefully and consistently. For the current exercise, we used WebPageTest (a third-party service backed by Google) and Pingdom. We simulated a desktop device and an iPhone 6 mobile device connecting to each site from remote locations (US and Japan). We ran each test three times and we report the average value.
The specific measure we used to evaluate loading speed was time to first render: a measure of the time at which content becomes visible to a user. A page may take longer to load fully, but well organised pages load visible content first and other ‘content’ later.
Figure 2 shows the relative loading speed for the home page of each site for desktop and mobile connections. We’ve sorted the graph in descending order of the time taken to load pages over a mobile connection, based on the assumption that more than 50% of site visits are from mobile devices.
1. The detailed results from both WebPageTest and Pingdom show where slow-downs occur and should be used to address specific configuration and page set-up issues.
In the previous section, we checked page loading time for mobile devices. We also ran each website’s home page through Google’s Mobile Friendly test. Google applies a number of criteria to determine if a page will display satisfactorily on a mobile device. The test checks for half a dozen mobile usability errors, the blocking of resources need to display the page and whether Flash media is need.
The home pages of all the sites passed Google’s Mobile Friendly test. In some cases, the pass was ‘qualified’ by access to resources needed to render the page fully being blocked.
There are two reasons that attention to these issues matters. First, as more visitors use mobile devices, the pages they access simply need to work. Second, mobile friendly pages and sites rank higher in Google search results than non-mobile friendly pages.
We also assessed ease of navigation by counting the total number of links across the sample pages for each website. We used this data to calculate an average number of links per page.
Links Per Page
Figure 3 shows how the average number of links per page various across the sample sites. The average value is 2.54 links per web page. University websites tend to be slightly more ‘complex’ with an average of 3.26 links per page.
Broken Links Per 100 Pages
As a further measure of ‘friendliness’ we checked the number of broken links encountered during sample page scanning. When a user clicks on a broken link, she usually encounters a ‘404’ or error message rather than the content of interest. Broken links are time consuming to find and eliminate. Some websites allow visitors to report broken links, but the most workable approach is to actively search them out.
Clearly, one site has an issue with broken links (we’d be happy to help). Otherwise, Premier League team websites had just over two (2.03) broken links per 100 web pages: which is substantially lower than the rate of seven broken links per 100 pages for the sample of university websites in our original study.
A zero-tolerance policy for broken links is visitor-friendly, prudent and achievable.
Images and SEO implications
Sports websites use large numbers of images, video and audio files. The way in which these are ‘managed’ has implications for search engine optimisation (SEO) effectiveness. We counted each page’s image files to calculate an average image intensity for each site. We plot the relative image intensity value for each PL website in Figure 5.
As images are an essential communication device for each website, we also ‘tested’ how easy it would be to find images through Google searches. And, once found we also checked for the additional information a visitor would obtain when interacting with (hovering over) an image.
If web page image has descriptive text placed in its ALT attribute Google is able to display the images in relevant search results. These images are also accessible to site visitors using screen readers. In other words, good SEO practice is to use the properties of the ALT attribute.
At the same time, if descriptive text is used in an image’s Title attribute visitors who hover a cursor over the image will be given supplementary information about the image. Many news and media sites use this technique to aid visitors understand the role of specific site images and their relationship to news stories.
Content management systems do not necessarily enforce adding ALT or Title image attributes inserted into pages or as part of articles. It is very common for one or both of these attributes to be missing — with an attendant negative impact on a visitor’s experience.
Figure 6 shows the relative number of images without associated ALT attributes for each PL website.
Figure 7 shows the relative number of images without Title attributes.
Continuing with the assessment of basic search engine optimisation we checked each sample page for Titles, Descriptions and Keywords.
A page’s Title text generally serves as the summary link text in a search engine result. The page description usually provides a page contents abstract. If site visitors find content through searches, that experience is better if all pages have unique titles and descriptions.
Our scans also look for keywords. Embedding keywords have no impact on search results, as Google has made quite clear.
Figure 8 shows the relative rate at which pages are missing Title tags for each PL website. Figure 9 shows the equivalent results for page descriptions. Figure 10 shows the rate at which pages are ‘stuffed’ with keywords on PL websites.
Most sites have close to 99.5% of their pages with titles. A small set of sites likely has minor content management system configuration issues, resulting in slightly greater numbers of pages missing titles: further investigation is required to confirm this.
The PL sites have relatively minor issues with missing page titles, but we did note that in many cases titles are repeated across pages (Google prefers unique title text for each page on a site) and titles often exceed the 60 characters generally recommended to avoid truncation in search results.
Google generally uses a web page’s description as the additional detail below the link in a search result. In the absence of description text, Google may use text from the page and may even use a snippet of text when a description is present, if the snippet is deemed more relevant.
It is good SEO practice to include page descriptions, where relevant. The sample page results suggest that missing page descriptions are widespread. We again note in passing that many pages share the same description (which can be valid) and page descriptions often exceed the recommended 160-character length.
PL websites seem enamoured with using keywords — perhaps this is a content management system option? The practice does no harm. However, evidence from other website reviews suggests the habit can be symptomatic of pages not being regularly reviewed, being spasmodically rather than periodically maintained and being slightly out of touch with current SEO practice.
We recognise that search-referred traffic may be a minor contributor to total site traffic, but it still makes sense to follow leading practice.
1. Run the Google Mobile Friendly Test on a sample of the most frequently visited pages, as determined by Google Analytics. Act on the results as appropriate.
2. Review all robots.txt files to ensure that resources needed to render pages are not being blocked.
3. Verify ALT tags on all images and add where missing. Configure CMS to enforce adding ALT tags on images.
4. Verify Title tags on all images and add where missing. Configure CMS to enforce adding Title tags on images.
5. Identify and review pages for missing Title tags and implement changes as appropriate.
6. Identify and review pages for missing Description tags and implement changes as appropriate. Revise operating practice to minimise inserting keywords on web pages.
Table 2 shows the tag management solution in place for each of the websites.
We also reviewed the number and type of third-party tags installed on each site’s home page. We waited until the home page had completely loaded, before making our tally. In common with other commercial websites the majority of the tags support advertising. We’ll address the nature and role of tags in a subsequent article.
Figure 11 shows the relative number of tags installed on each site.
1. For those sites not using a tag management solution, implement one.
2. For those sites using Google Tag Manager, ensure that the configuration reflects both the use of third-party tags and the events and triggers needed to populate the associated Google Analytics installations.
3. Read our two blog posts on the principles of managing tags, How to Maintain Tag Control & Privacy on University & College Websites and How to Stop Tags Choking University Website Performance
Like large crowds, complex, content-rich websites can become unruly. And, content management systems don’t always provide the perspective to track down issues that affect user interaction and performance. Moreover, many of the issues need to be actively sought out and fixed as users will not bother to report them or have a practical means to do so.
Regular scanning gathers the data to understand a website’s current status and provides the means to identify and resolve many user experience issues.
[All copyrights, trademarks and other rights are duly acknowledged]
If you’d like a PDF version of the article to read off-line, you’ll find one here: https://goo.gl/dluzI0