The impact of book suggestions/recommendations?

Whilst finalising my presentation for the 2009 UKSG Conference in Torquay, I thought it would be interested to dig into the circulation data to see if there was any indication that our book recommendation/suggestion services (i.e. “people who borrowed this, also borrowed…” and “we think you might be interested in…”) have had any impact on borrowing.
Here’s a graph showing the range of stock that’s being borrowed each calendar year since 2000…
interesting
Just to be clear — the graph isn’t showing the total number of items borrowed, it’s the range of unique titles (in Horizon speak, bib numbers) that have been borrowed. If you speak SQL, then we’re talking about a “count(distinct(bib#))” type query. What I don’t have to hand is the total number of titles in stock for each year, but I’d hazard a guess that it’s been fairly constant.
You can see that from 2000 to 2005, borrowing seems to have been limited to a range of around 65,000 titles (probably driven primarily by reading lists). At the end of 2005, we introduced the “people who borrowed this, also borrowed…” suggestions and then, in early 2006, we added personalised “we think you might be interested in…” suggestions for users who’ve logged into the OPAC.
Hand on heart, I wouldn’t say that the suggestions/recommendations are wholly responsible for the sudden and continuing increase in the range of stock being borrowed, but they certainly seem to be having an impact.
Hand-in-hand with that increase, we’ve also seen a decrease in the number of times books are getting renewed (even though we’ve made renewing much easier than before, via self-issue, telephone renewals, and pre-overdue reminders). Rather than hanging onto a book and repeatedly renewing it, our students seem to be exploring our stock more widely and seeking out other titles to borrow.
So, whilst I don’t think there’s a quick any easy way of finding out what the true impact has been, I’m certainly sat here with a grin like a Cheshire cat!

8 thoughts on “The impact of book suggestions/recommendations?”

  1. I’d hazard a bet that it has made a difference. Would be interesting to compare this graph to other similar libraries. It would also be nice if you could do some sort of meaningful survey.
    (Do you shop at the Co-op ever? I’ve been impressed by their ways of doing ‘impromptu’ surveys via the cash card console. Not necessarily do-able, but has me thinking beyond ‘Amazon’ incentives;-))

  2. Great stuff. As Joy says, I guess to really understand behaviour you’d need to ask the users the question. With self issue machines the idea of impromptu questionnaires a la Co-op sounds quite interesting.
    I wonder if there are some other factors you could plot against the number of bibs in circulation – e.g. average # copies per bib, registered users – this would help see if other factors are varying (and therefor related)

  3. I’ll see what other data I can pull together, although I might have to rely on the SCONUL stats for some of it. Using the circ data, I should be able to calculate the average number of unique bibs borrowed per user since 2000.

  4. Well the introduction of similar features at Dundee doesn’t appear to have had the same dramatic effect. The number of distinct titles borrowed has been declining slowly since 2005 (as we moved systems I do not have any figures for earlier than this). The decline is perhaps slightly less since we introduced the recommendations. Unique titles per user sounds a better approach, but looking at the real-terms book spend (excluding serials) or disposals of material could also be revealing.

  5. I’m curious at to whether data can specifically target authors that are popular and recommended so that the collection of that particular author’s work can be deepened. Many librries don’t have “all” of a particular author’s works, yet don’t always fill in missing titles. Having more titles of the popular and recommended authors, might help increase circulation as well.

  6. Hi Lizbeth!
    I’m wondering if there’s any way of using the Amazon Web Services to help locate gaps for a given author?
    Alternatively, if circulation data from multiple different libraries could be aggregated together, that would allow you to identify which would be the best of the missing titles to purchase (i.e. out of the missing titles, which are the ones that have the highest circulation at the other libraries).

Comments are closed.