Higher education digital marketing relies on third-party technologies
Effective higher education digital marketing and communications rely on an intricate set of technologies. A foundation comprising internal and third-party applications, sometimes collectively called martech.
As marketing strategies evolve and tactics change, digital marketing technologies get layered on top of each other potentially creating fragile Jenga-like structures or Frankenstacks. Often neither the original decision makers nor their documentation is available to clarify past choices.
Many institutions find it tough to answer the fundamental question: are our third-party tags helping or hindering our digital marketing?
Satisfactorily answering that question begins with two reverse engineering projects. The first one builds an accurate list of all of an institution’s web properties. The second examines each website on that list to determine the third-party tags actually in use.
Rather than build a list of sites at a single institution we used 170 Canadian university, polytechnic and college home pages. Then we examined each home page to see which tags are used and for what purpose [see footnote for a caveat].
This article dissects the scope and scale of third-party tags used in university and college digital marketing, exposes some of the common issues that arise and examines the privacy issues marketing and communications teams need to understand and resolve.
What are third-party tags and how are they used?
By contrast the WordPress Simple Share Buttons Adder social media plug-in (a first-party tag) runs the application ssba.js whose code is stored locally at https://www.exampleu.ca/wp-content/plugins/simple-share-button-adder/js/.
Third-party tags perform six main functions on higher education websites:
- Advertising Support - Provides data collection, behavioural analysis or retargeting for advertising or advertising-related services
- Audio/Video Player - Enables websites to publish, distribute and optimise video and audio content
- Customer Interaction - chat, email messaging, customer support and other interaction tools
- Essential - tag managers, privacy notices and technologies critical to website functionality
- Site Analytics - collects and analyses site usage and performance data
- Social Media - integrates social media site features
Four of these functions directly address digital marketing execution – advertising, audio/video player, customer interaction and social media. The fifth category – site analytics – measures execution.
We uncovered 868 third-party tags in use across 172 website home pages or an average of five tags per site. Tag use is unevenly distributed. Sixty percent of the sites install four or more tags, while 15% use just a single tag.
Our research shows that a majority of higher education websites deploy multiple tags, as well as using Google Analytics (see Table 1). This finding suggests that tag implementations are sufficiently complex that they need a management mechanism, which we address below.
How much tagging is going on and to what ends?
We sorted the 868 tags in rank order to identify the ten most frequently used tags across all websites:
|Tag Name||Category||Number of sites with tag||Percentage of all sites|
|Google Analytics||Site Analytics||165||95.9%|
|Google Tag Manager||Essential||92||53.5%|
|Facebook Connect||Social Media||80||46.5%|
|Facebook Custom Audience||Advertising||63||36.6%|
|Google AdWords Conversion||Advertising||34||19.8%|
|Twitter Syndication||Social Media||28||16.3%|
|Twitter Button||Social Media||27||15.7%|
|Google Ajax Search API||Essential||23||13.4%|
Table 1: Top Ten Tags for Canadian Higher Education Institutions
Table 1 reflects the good news that Canadian higher education institutions use Google Analytics to measure website visitor activity and behaviour. They can use that data to inform and optimise their digital marketing. However, our work with higher education websites also shows that sites install Google Analytics, but do not configure their installations to deliver actionable data.
The less-good news is that only half of all sites use Google Tag Manager (GTM). This discovery is less-good news, because, in addition to its support for measurement customisation, GTM is a “self-documentation” system. Implementing all tags through GTM means tag details are always up-to-date – handy when marketing tactics or staff change.
Table 1 also highlights how widely advertising and social media tags are installed to support digital marketing. As the chart in Figure 1 confirms, about one third of all tags support advertising. This proportion of tags dedicated to advertising seems inconsistent with the relatively few university or college AdWords, Twitter or LinkedIn ads we actually see.
The last major group, social media tags, typically support signing on to applications from Facebook or Twitter accounts. It is not entirely obvious why this functionality is relevant on university and college home pages. Feel free to enlighten us.
Overall, 55% of installed tags are directly related to digital marketing support. The balance collect data or exist to make sites work more effectively.
Returning to advertising tags - one explanation for the seeming ‘overabundance’ of advertising tags may be confusion about account settings in Google Analytics. Many institutions will have configured their accounts to share data with other Google products and services under Google’s Data Sharing Settings, which, in turn, means activating some of Google’s advertising capabilities.
A further explanation for the number of advertising tags is that the Add-this and Share-this plug-ins for social media sharing are actually classified as advertising tools – adding to the high overall proportion of advertising tags.
How many tags do higher education websites actually use?
About 16% of sites (n=27) use a single tag, mainly Google Analytics. Three sites do not record site activity, so their single tag loads Adobe fonts. Another 16% of sites (n=27) use two tags, principally Google Analytics and Google Tag Manager.
Sites using four tags typically use a Google Analytics, Facebook Connect, Facebook Custom Audience and DoubleClick combination, but not Google Tag Manager.
As the number of tags in use increases there seems to be no discernible pattern to tag deployment.
An alternative (more comprehensive) approach is to examine all instructions passed to external servers. On this count, the outlier site eventually installs 34 tags on complete page loading. All of the additional tags support advertising by piggy backing on the advertising tags the institution originally installed.
The marketing impacts of not managing tag use
Poor user experience
Research shows that website users prefer web pages that load quickly, whether on mobile or desktop. Google reports that 53% of mobile site visitors leave pages taking more than three seconds to load. Others may stay on the page, but find their experience frustrating.
The 22-tag website takes just over 14 seconds before desktop users can start interacting with it and just over 24 seconds before it works for a mobile user on a 3G network: yikes! Choosing a four-tag site, at random, we measured load times of 4.8 and 14 seconds respectively. Better, but still with room for further improvement.
Checking which tags are being used and why, can rationalise tag use and boost the attendant user experience: less is more.
Wasting time collecting data that will never be actioned
Large numbers of tags may theoretically be collecting important data, but without a management plan that data may languish unanalysed and ultimately never contribute to a marketing or communications decision.
About one in eight of the sites we examined had more than ten tags being loaded. Most of these tags collecting data around Google, Facebook and Twitter advertising programs. If this data is not being used to optimise campaigns, there may be no value in continuing to collect this data.
Running tags that are obsolete or have been deprecated
Without oversight or regular monitoring of which tags are supposed to be supporting marketing efforts, sites may run tags that no longer function, run more than once (invalidating their counts of on-page events), are out of date or even present security risks.
We noted a number of sites operating with multiple, conflicting Facebook pixels (usually present as the fbevents.js code). We noted on other sites set of tags violating their site’s Content Security Policies and not being loaded.
Running unauthorised tags or piggy backing tags
Piggy backs or tag re-directions often arise when tag vendors partner with re-targeting companies to reach wider audiences. Suppose a prospective student visits Example University’s website, but doesn’t request information or take other action. A tag recording the visit makes data about the prospective student available to a re-targeting partner. The partner would then show the prospective student Example University advertisements, when she visits other websites in its network, with the objective of getting her to re-visit the Example University website.
Five of the sites in our survey use the Adroll RoundTrip tag which in turn runs scripts for 10 other tags (Open X, Rubicon, Outbrain, Advertising.com, Index Exchange, LiveRamp, PubMatic, Taboola, TripleLift and Bidswitch). The impact of this tag may be at odds with the site owner’s objectives – only an audit will tell.
Tag management recommendations for higher education digital marketing teams
All of the websites in our survey use tags. We doubt that in every case these site’s ‘business’ owners know which tags are installed and the purposes they serve.
Lack of knowledge carries with it risk exposures, diminishes the reliability of marketing data being collected and potential reduces the on-page user experience.
Here are five recommendations to resolve the types of issues describe above and address the privacy concerns in the next section:
- Identify all of your institution’s relevant websites
- Identify all of the tags on all of your relevant websites
- Prepare a simple tag management document that records the following:
- A non-technical description of what the tag is measuring, why this data is needed and on which websites it will be used
- Define and describe the measurement variable(s) the tag records (for example: the tag is designed to measure how far down a page a user scrolls)
- Define and describe the measurement conditions (for example: this measurement will only be triggered on scrolling down 50% of the page)
- Define and describe on which parts of a website the data will be collected (for example: only collect data on pages containing /course-description/ in the URL)
- Implement tag governance. Governance means determine who’s going to decide which tags get implemented and how you record those decisions.
- Implement a tag management system to allow marketing teams to add/delete/modify their tags
The privacy impacts of tags for higher education digital marketing teams
As higher education websites reach international audiences and most institutions actively recruit international students, marketing teams need to be aware of potential data privacy concerns.
On 25 May 2018 the European Union’s General Data Protection Regulation (GDPR) came into effect. Under GDPR EU citizens are entitled to a well-defined set of digital data privacy protections. And, those rights are enforceable worldwide.
In other words, under GDPR, EU citizens visiting US, Canadian or other non-EU university websites are entitled to expect the same privacy protection that they would receive from EU higher education institutions.
Yet, our research shows that few non-EU higher education websites are actively complying with GDPR, even though their websites are a destination for EU citizens looking to study abroad.
Identifying EU site visitors and responding to their data privacy requirements
For EU visitors you need their permission to collect personally identifiable information (PII) and when you receive permission, you also need a corresponding data retention policy.
In practice, this means presenting EU visitors with cookie identification notices and then allowing them to opt out. It also means implementing geolocation identification tags to present appropriate visitors with a privacy and opt out notice.
If EU visitors opt out, you can’t track them using any of the tags currently installed on your website(s). To fulfil this requirement, you will likely need to implement further technical changes, such as cookies recording their choice for the duration of a browsing session.
If EU visitors opt in you need to understand if any tags collect personally identifiable information: an IP address is considered personally identifiable data. That data will need anonymising or should not be collected from EU visitors.
As 96% of the sites in our study use Google Analytics, they will need to ensure they do not collect user IP addresses, avoid accidentally collecting PII (for example, email addresses collected via URLs embedded in email newsletters) and ensuring that Google Analytics data is deleted for EU visitors within a reasonable time period. Google Analytics currently defaults data retention to 26 months.
For Google Analytics managed via Google Tag Manager IP anonymization is readily implemented via the tag properties. For other third-party tags compliance with the GDPR privacy requirements may be more complicated.
Responding to privacy concerns
It is not unreasonable to expect similar legislation to GDPR coming into force in other jurisdictions as a response to public concerns about website data collection and the trustworthiness of organizations collecting that data.
If many site visitors opt out of having their data collected measuring digital marketing program effectiveness will be much less reliable and returns on investment harder to calculate and demonstrate.
As trusted and ethical institutions, non-EU universities and colleges should demonstrate that they can be trusted with data collected on their websites and minimize the number of visitors opting out.
Trust can be developed by being transparent about what data is being collected and why, by validating the tags used for data collection, eliminating over-collection of data and anonymising data wherever possible.
The first step is to identify the tags currently in use and re-justify why they are in place.
Footnote: There are two main ways of uncovering a website’s tags. The first method scans the source code when a web page has fully loaded. That’s the approach we took to collect data for this article. But, this approach potentially underestimates a web page’s full tag payload.
A more sophisticated method traces all the external servers a tag calls, but this relies on a comprehensive tag list. Our list numbers hundreds (the ones universities generally use), but software such as Ghostery’s recognises thousands. The second method is superior, because it identifies piggy-backing, when further tags run on a page invoked by some of the original tags.