If you build it, they’ll come back for more!

I’m just busy putting together slides for some of the upcoming presentations and I thought it was about time I trawled through some of the OPAC usage stats to see if our students are still using some of the OPAC tweaks we’ve made.
The good news is that they are, and then some more!
First up, here’s the overall usage for 4 of the tweaks (May 2006 to July 2007):
OPAC tweak usage
At first glance, nothing too surprising — the overall trend follows the academic year, with the lull over summer.
What did leap out was the blue line (clicks on “people who borrowed this, also borrowed…” suggestions) — since this April, the usage has been higher than the “did you mean” spelling suggestions (red line). So, either our users have suddenly become better spellers, or they’re making much higher usage of the borrowing suggestions. If I was a betting man, I’d say it was the latter.
We’ve now got enough data to compare the same 3 months in 2006 and 2007 (May to July):
OPAC tweak usage
OPAC tweak usage
That second graph is why I’m sat here with a grin like a Cheshire Cat 😀
I’ve dug out the circulation stats for the same period and that re-inforces the statement that the students are making much higher usage of borrowing suggestions in 2007 than in 2006. You can see that the number of check outs (bold pink) pretty much matches the number of clicks on the “did you mean” spelling suggestions (red line in the first graph). Check outs have also risen in 2007 when compared to the same months in 2006.
circulation stats
Interestingly, I don’t think we’ve ever had a student go up to a member of staff and say “I’ve found the suggestions really useful” or “thank you for adding spell checking”. I wonder how many complaints we’d get it we turned the features off?

5 thoughts on “If you build it, they’ll come back for more!”

  1. Yikes — I’m in Cites & Insights! Many thanks to Walt Crawford for making my ramblings seem coherent 🙂
    Since posting these stats, I’ve been pondering the increase in the “also borrowed” clicks…
    As I mentioned, circulation in 2007 is up on 2006, although it’s not dramatically higher — around 3%. Also, the check out graph pretty much matches the “did you mean” graph.
    Unfortunately what I don’t have is detailed logs of OPAC usage (e.g. unique users, length of session, etc).
    My gut feeling, which is partly confirmed by the graphs, is that the dramatic increase in “also borrowed”s indicates that out users are doing a lot more browsing within the OPAC than they used to.
    Looking back a couple of years, the full bib pages on the OPAC had very little in the way of exploratory links — you could either click on the author’s name, or on one of those cryptic LCSH links.
    The privacy issue is an interesting one and the PATRIOT Act has little in common with our Data Protection Act. I must stress again that we take borrower privacy extremely seriously and that the borrowing suggestions are based on fully anonymised and aggregated data.
    When we launched the feature, I did wonder how long it would be before a student or an academic member of staff raised the privacy question (either in anger or from curiosity). Nearly two years on, I’m still waiting.
    So, don’t our users care about privacy issues? …are they 100% confident that the library is acting within the guidelines of the Data Protection Act? …or are they simply so familiar with the idea from seeing it on Amazon that they don’t even think about it? Either way, they certainly seem to have embraced it.
    Walt also mentions that “also borrowed” is the only one that can be considered as “Library 2.0”, which I’d partly agree with — certainly the spelling suggestions should be “Library 1.0” and it’s to our collective shame that so many OPACs don’t even have that feature.

  2. I can imagine that one among several reasons for the extraordinary rise in the use of the “People Who Borrowed This…” is that the amount of data that is used for the calculations is reaching a critical mass that gives dramatically more relevant suggestions than just after the service was introduced.
    I can imagine that your construction of the spell checker doesn’t improve over time in the same way.

  3. Hi Jens
    I could see that would certainly be the case for an “opt in” system where the data gradually built up over time, but when we launched the service it was already using 10 years of historical circulation data (around 2,000,000 transactions). In other words, the critical mass was already there.
    The suggestions will certainly have improved and increased in number over the last couple of years (as another 500,000 transactions have been added to the melting pot), but that in itself would only account for about a 20% increase.

Comments are closed.