We’re about to launch a nice integration between our e-learning system (Moodle) and our reading list system (Aspire) – I’ll detail more about some other time when launched. One of the things we wanted to do is monitor how many users were clicking from Moodle to Aspire once it was launched.

Something I had noticed before with our Google Analytics in the past is that we had almost zero referrals from our moodle system, not just to our reading lists, but also to every site and online service we run. This was odd as our moodle system here is very well used and much smaller and bespoke services which generating many more referrals to our sites.

In fact looking at the GA stats for the entire University website confirmed the same. Over the course of the year obscure sites around the world were generating more traffic in to the University website.

It wasn’t just Moodle. The in-house web administration portal (timetables, grades, library account, personal information), SussexDirect, which again is a major service used by everyone on campus was generating very few referrals.

Why was this? Were they being included in ‘sussex.ac.uk’? were these platforms doing something which lost the referral information, perhaps due to javascript linking or redirects?

Looking at the former. Other hostnames in sussex were being recorded separately, so why would these be treated differently? Plus, if I drill down by selecting ‘sussex.ac.uk’ from the referrals report it shows the pages from the domain that are generating clicks through to our pages, and looking at this list it was clear that the pages were all from www.sussex.ac.uk and not from the moodle and SussexDirect. Nor could I see anything in the way these systems created linked which would stop them from being counted.

This is the sort of thing that bugs me and I wanted to get to the bottom of it. I needed to do setup various tests to narrow down the problem.

My next test was to use the real time reports. I created links from other systems on campus that I can edit (a local wiki, the University blog platform etc) and linked to these various systems. Then using different browsers (otherwise it remembers your current Visit) I clicked through from the links to various web services.

Clicking the links from the local wiki to various services (while following the real time GA report for each service) showed it worked as expected. Each click showed up on the realtime report (it’s summer, so low usage makes it easier to spot myself) and I could see it coming from the wiki’s hostname. Good.

However links from Moodle and SussexDirect would show up in the sense it would show I have just accessed a page on the destination site, but the referral report would either show nothing or just ‘sussex.ac.uk’. The latter case was particularly odd, as if – as noted above – you then ran the normal referral report, it was clear that ‘sussex.ac.uk’ did not include pages that are part of Moodle/SussexDirect.

Was this a common problem? Googling around (actually, duckduckgoing around) produced little. Lots of people were concerned about self referrals, but little in the way of losing referral information from what should be a major source of inbound users.

The problem was frustrating, GA is very popular, so is Moodle, yet various searches provided no clues as to others having similar issues, yet I was not aware of anything locally, or with our GA configuration, that was non-standard.

The solution came from a colleague in IT Services (which is annoying as I didn’t get to the bottom of it). Moodle and SussexDirect serve their content via https – an encrypted http connection. Aspire, and other sites, use http. When following a link from http to https it is good practice for web servers and logging tools to not include the referral information (as it may include sensitive information). It was that simple.

The workaround we are looking at is to use GA Custom Campaigns – which includes adding some parameters to urls – to help monitor users following links from system to the other.

Leave a Reply