If you have Google Analytics or any other 3rd party JavaScript installed on your website, you should take eight minutes to read this post about tags and how they can negatively impact website performance.
Outcome Measurement
For eighty percent of TERMINALFOUR’s Global Higher Education Survey respondents the primary objective of their web strategy is student recruitment. While our last post pointed out that actual university and college websites are organised to achieve objectives other than student recruitment, universities and colleges are definitely in the marketing game. And, as we’ll see later in this post most of them are playing tag, as well.
In last time’s post we also noted that although many higher education institutions admitted to running without a published web/digital marketing plan, but using web analytics to measure some aspects of the activity and behaviour on their websites.
In this post we discuss our findings from reviewing the tools that university and college websites have in place to track and analyse visitor behaviour. The analysis described below applies only to the main home page for each university’s website.
Consistent with our self-imposed mandate to publish posts based on field research, we returned to our sample of UK university websites. The group’s slightly larger than when we looked at CMS usage, as we’ve included almost every recognised institution awarding degrees in the UK (159). We’ll use data for the US and Canada in subsequent posts.
Google Analytics
Forty-four percent of TERMINALFOUR’s Global Higher Education Survey respondents stated that they used site visits to measure the outcome of their web/digital marketing plan. The good news is that 99% of UK university websites have some variant of Google Analytics installed. The slightly disconcerting news is that some sites have three different versions (Universal Analytics, Google Analytics and DoubleClick) running, some have two different versions installed (usually UA and GA) and 20% run only the ‘obsolete’ GA version. A couple of sites don’t have Google Analytics installed on the home page, but installed on all pages linking to the home page (providing a slight page loading performance boost). And, even the one site without Google Analytics installed, has an alternative solution in place.
Overall, UK university websites are in a position to collect pageview, session and hits data. Our experience from past Google Analytics consulting exercises, is that most sites have it installed, a smaller group has it appropriately configured to collect 'clean' data and an even smaller group has it set up to enable dashboards and generate actionable reports.
Google Tag Manager
The second most popular outcome measure referenced in TERMINALFOUR’s survey was form submission (31%). Google Analytics can readily record “events” generated by forms submitted from a web page. A flexible way of enabling this is through Google Tag Manager (GTM), as this tool allows non-technical staff to configure, test and add code snippets (tags) that allow Google Analytics to collect and report the associated activity.
Sixty percent of the 159 UK university websites we reviewed have installed Google Tag Manager. These sites can manage how Google Analytics is installed, allow non-technical staff readily configure event recording to measure marketing campaign effectiveness and install other code to track or analyse activity. For the 40% of sites that have yet to install GTM, Michael Fienen’s presentation at the HighEdWeb16 offers an overview of GTM and its relevance to higher education. We also note that over 90% of sites with six or more tags being loaded use Google Tag Manager: confirming that the managers of these sites already appear to recognise the complexity of their tag usage.
Of course it is not quite as simple as described in the previous paragraph. Some sites have installed GTM, but are not using it to manage all of their tags. And, many of the sites running multiple versions of Google Analytics have GTM installed, but are not benefiting as much as they should.
Playing Tag Whether You Know It Or Not
Given the presence of multiple versions of Google Analytics and less-than-universal implementation of GTM, we developed a working hypothesis that many site managers may be unaware of the number of tags running on their home pages and the implications. To discover if "tag proliferation" might be an issue, we wrote a simple script to identify and record the snippets of 3rd party JavaScript code (tags) present on UK university website home pages. Our script identifies 95 different tags and across 159 university websites we found examples of 69 of these.
Tag Categories
To begin our analysis, we used Ghostery.com’s tag categorisation scheme to develop an overview of the type of tags installed on university websites. Ghostery uses eight tag categories:
- Adult Advertising
- Advertising
- Audio/Video Player
- Comments
- Customer Functionality
- Essential
- Site Analytics
- Social Media
We recommend installing Ghostery’s browser extension to show which tags or trackers are being loaded on a web page. Ghostery’s extension is more sophisticated than the script we wrote, as Ghostery examines the requests made to other servers while a page loads. Our script simply scans the page source to identify JavaScript elements. As a result, our script identified one site as having 14 different tags, while Ghostery reveals 17 different tags being loading.
As the two graphs summarising our findings about tags installed on UK university websites show, tag proliferation is a material issue. The first graph shows the categories into which the 69 different tags we identified across 159 websites can be allocated (using Ghostery's categories). As graph 1 shows, most tags exist to support advertising, closely followed by tags for site analytics.
Graph 2 shows the relative occurrence of tags in each category. We found a total of 894 tags installed across the 159 websites, and just under 40% of the total number belong to the advertising category, with almost 30% being installed to support site analytics.
Graph 1: A total of 69 different tags were found across 159 UK university websites. The graph shows the tag population allocated to Ghostery's eight categories. The majority of the tags are associated with providing advertising functionality, followed by site analytics. The Essential category covers tags such as Google Tag Manager, Google Analytics, ClickTale and 3rd party JavaScript used to add fonts.
Graph 2: A total of 894 tags were implemented across the 159 UK university websites reviewed. The 'average' site has 5.6 tags. The graph shows most tags are implemented to support advertising functionality, and site analytics.
One of the first questions prompted by the graphs is the reason for the predominance of advertising tags on university websites? A partial answer may lie in Google Tag Manager and Google Analytics often being configured to enable DoubleClick and Google Remarketing to obtain consumer demographics data for use in Google Analytics. In our experience, relatively few sites use the demographic data. If you’re unable to clearly state why you need this data for marketing or communications decisions, it would likely make sense to turn off these options and avoid loading the tags.
What Are All These Tags Doing?
To take our analysis from cataloguing to understanding the potential issues we obtain help from Mollie Panzer at Ghostery. Ghostery clients (Disclosure: we are neither affiliated with nor a Ghostery client) have access to a range of services, but one tool in particular – Trackermap – offers a visual representation of tag activity when a web page is loaded.
We selected three different universities to be run through Trackermap and we have obscured the names to:
- Site 1 - we had identified as loading 14 tags on the home page;
- Site 2 - we had identified as loading five tags (the median number of tags for sites in our review) on the home page;
- Site 3 - we had identified as loading one tag on the home page.
The relevant Trackermap results are shown below.
Map 1: The site we identified as having the largest number of tags installed
The map shows all the server requests made in loading the home page. Seventeen different tags ‘fire’. Importantly, the map shows us that although Google Tag Manager is installed, most of the tags load directly, thus negating the benefits of Google Tag Manager.
Map 2: An example of a typical UK university website loading five tags on the home page
The map again identifies all server requests while loading this site's home page. Five different tags ‘fire’, with one of these, Crazy Egg, operating directly off the page. We note that Crazy Egg is one of fifty or so standard tags that can be implemented directly in Google Tag Manager. It would also be interesting to know what incremental understanding of how well the home page is meeting its objectives is being derived from deploying three site analytics tools rather than one: Crazy Egg, Google Analytics and HotJar?
Map 3: The lone site not using Google Analytics and with a single tag installed
The Trackermap again identifies all server requests made while loading the home page. This site has three associated domains, which we have obscured in addition to the root domain. In this instance the site uses a web analytics service other than Google Analytics and operates satisfactorily without a tag management tool.
To answer the question posed by this section's heading, what the tags are doing is related to the number of tags installed on a site. For sites with relatively low numbers of tags (one to four tags), the tags generally exist to enhance data collection and site analytics. As, the number of tags increases (more than five tags), the incremental tags are directed at advertising and social media interaction.
Are There Any Issues With Having All These Tags Installed?
Performance
A reasonable question to ask at this point is, does the number of tags being loaded impact site performance? The unequivocal answer is yes. The more tags being fired, the slower a page loads. We ran each of the Trackermap sites through WebPageTest to measure page loading speeds for desktop and mobile device visitors to our three test sites. Our test results are summarised as follows:
Table 1: Page Load Speeds for The Three Sites Shown in The Trackermaps. TTFB measures the Time To First Byte (how long before the server first responds to a request). As would be expected the number of requests made to different servers in loading the home page increases with the number of tags being loaded as does the total time taken to load the page.
The results show the average of three runs over a broadband connection for a desktop user and a 3G connection for a mobile device user.
Site 1 (14 tags) is by far the slowest and mobile users in particular may lose patience waiting to access relevant content. Site 1 also reported a small number of 404 responses that further detracted from site performance. Sites 2 and 3 performed well for both desktop and mobile users – lending support for the hypothesis that fewer tags will result in a better visitor experience.
Privacy
Given that tags have two main roles, supporting site activity analysis and tracking visitors, we also need to ask if there are privacy issues raised by installing tags? The simple answer is yes. The more nuanced answer is, maybe. Tags can operate in ways that are not necessarily obvious to website administrators, particularly advertising tags where session data may be passed to other entities that participate in advertising networks. Furthermore, our initial analysis shows that there are some discrepancies between published website cookie policies and the cookies actually being loaded on a visitor’s browser.
The privacy issues are complex and we’ll discuss them in our next post on 14 November 2016.
Recommendations
If you haven’t already installed Google Tag Manager, do so and work through the training tutorials to learn how to use it appropriately.
Our analysis indicates that many UK university website load tags on their home page without being aware of the performance and potential privacy issues being raised. Any site, even one with only Google Analytics installed will benefit from using tag management software, because it performs two important roles:
- It provides an inventory of the tags being used and their loading can be optimised;
- Adding JavaScript directly to a site requires ‘technical’ knowledge. Tag management software allows non-technical staff to define, implement and test tags and to do so quickly and thus readily support digital marketing campaigns.
Download and install the Ghostery Browser extension and use it to identify and record all the tags actually installed on a page.
After cataloguing the tags, identify who ‘owns’ them, to understand why they are present. If you can’t identify an owner, then you likely don’t know what objective a tag meets, at which point uninstall it. And, for the tags that can be associated with an ‘owner’ determine what outcome the tag is designed to meet:
- Visitor tracking – this may already be fully covered by Google Analytics;
- Event tracking – this is likely better implemented through Google Tag Manager;
- Re-marketing – be clear that you understand how you want this to work, following up on your own AdWords campaign makes sense, handing off user data to an ad network may not.
Move any tags that have survived ‘identification’ into Google Tag Manager, unless you can determine there is a compelling reason that they can’t be managed through this mechanism.
