If you build it, will they come? (part 2)

I hate being reminded of how quickly time flies!
Nearly a year ago, I added a post about how five of our OPAC tweaks were being used:
1) “did you mean?” spell checker suggestions for failed searches (more info)
2) serendipity keyword suggestions for failed searches (more info)
3) “people who borrowed this also borrowed…” suggestions (more info)
4) books with similar subject headings (more info)
5) other editions of books (more info)
I left the logging running, so we now have 11 months worth of data to show usage — by “usage” I mean when a user actually clicked on one of the links displayed by the tweak.
Click any of the graphs to view the full sized version.
Usage by month

Being an academic site, usage of all library services varies by month — usage drops heavily during summer, and dips during Christmas and Easter.
You can easily see that “did you mean” and “people who borrowed this” are the most heavily used.
The “people who borrowed this” and the “similar subjects” tweaks are both ones that promote serendipity when browsing items — you’re viewing one book but are being given links to other relevant items.
Although the “serendipity keyword suggestions” is the least used feature (partly because it only appears under certain circumstances), it’s getting enough usage to justify the couple of hours coding it took.
Here’s a breakdown of the average number of daily clicks, along with the peak number of clicks each tweak has received in a single day…

Usage by day

It’ll be interesting to see if Tuesday is also the busiest day for other library services.
Usage by tweak

There’s almost a doubling in usage going on here from the “serendipity keywords” to the “did you mean”. I said it last time and I’ll say it again now — if you don’t have a spell checker on your OPAC, then you need to hire a web developer and add one a.s.a.p.!
Usage per hour

That odd little bump in usage around 5pm has smoothed out now.
Even with the least used tweaks, there’s more than enough usage to justify the development time, so I’m extremely happy with the graphs.
Assuming I don’t get run over by a bus, call back next year to find out if the usage has increased!

8 thoughts on “If you build it, will they come? (part 2)”

  1. May I just say how great it is to see actual numbers regarding innovations like this? Particularly long-term numbers, getting past the halo effect.
    It’s one thing to see some useful effect (and I certainly agree with your conclusion). It’s another to tell us about it: That benefits the field directly and substantially. Thanks.

  2. Many thanks for that Walt.
    What I’ll try and do is get some accurate figures for usage of the OPAC in general.
    One of the difficulties is that we have around 50 dedicated OPAC PCs in the our libraries that are set to logout and reload the front OPAC page if left alone for a couple of minutes. So, if they all got left on for 24 hours (which sometimes happens) and no-one used them, they’d still rack up 36,000 page views a day in the logs.
    Just to compare the top two used tweaks with some other monthly library data, here is a graph that also includes monthly figures for:
    1) the total number of distinct students who visited the main library
    2) the total number of distinct students who checked a book out

    I’ve deliberately used “smooth” lines, as it’s easier on the eye for picking out any trends.
    There are a couple of things that stand out to my eye:
    Firstly, relative usage of the tweaks appears to have increased since the influx of new students in Sep 2006. We didn’t publicise the tweaks to the students — instead I prefer to let them discover the value (if any) of the tweaks by themselves. In a year from now, it will be interesting to return to the data to see if tweak usage relative to vists and CKOs has continued to increase.
    Secondly, there’s a lag of a month between the peak usage for “did you mean” (Oct) and “also borrowed” (Nov). This makes sense — at the start of the new academic year (Sep), a lot of the stock is available but as CKOs increase (Oct), students will need to consider alternative titles when the book they want isn’t available (Nov).

  3. One more graph…
    We’ve been keeping tabs on failed keyword searches since Jun 2006 and “did you mean” links are only displayed after a failed keyword search.
    This graph shows the monthly number of “did you mean” click thrus as a percentage of the number of failed keyword searches:

    What is shows is that there’s an overall 14.28% (or 1 in 7) probability that the user clicked on a “did you mean” link when their search failed.

  4. The last chart is particularly impressive. It’s easy enough to say that offering alternative spellings is a great idea (and it’s true)–but you’re showing that it’s a great idea that people demonstrably use. (Those of us who’ve spent time building Help facilities, for example, are uncomfortably aware that those facilities are rarely used no matter how wonderful they are–and also that they’re absolutely essential even if they’re almost never used. You’ve got a better situation: Something inherently useful that’s also getting used a substantial percentage of the time.)

Comments are closed.