User-driven purchasing: a pilot project to test an alternative pricing model for Springer e-book collections

readeror patron-driven acquisition) become a viable way of purchasing e-book collections? This question arose during a discussion between Anne Donnelly from Springer, Steve Sharp from Leeds University Library and me back in October 2009. That discussion led to an invitation from Springer to participate in an extended trial of their e-book portfolio, in order to assess whether user-driven purchasing could be applied to e-book collections. The user-driven purchasing model is already being offered by the major e-book aggregator platforms (eg. dawsonera, EBL, ebrary and MyiLibrary) and is deployed in a number of libraries worldwide, but its established use is for buying individual e-book titles rather than collections. So what was the trial model proposed by Springer?

■ E-book content would be opened within a library's SpringerLink account and costs would only be incurred once a chapter was downloaded; there would be no ongoing platform fees. A pre-defined cost per chapter download would be mutually agreed upon and all usage would be based on COUNTER Book Report 2 (number of successful section requests, which for Springer equates to Chapter Downloads) which is updated monthly. Springer's initial thoughts on costs were £4 per chapter download. They felt this price mapped nicely with rental prices on e-book aggregator sites, so that even if a collection was not then purchased, £4 for the one-time download of an STM e-book chapter would seem to be good value for money. Repeat downloads of the same chapter would still incur the chapter download fee.
■ Springer would offer all available copyright years (namely 2005 onwards); at the end of 2009 this equated to approximately 18,000 titles. Libraries could choose to select only a subset of copyright years or subjects if they wished.
■ OCLC-enhanced Marc records would be provided for all e-books. Discovery would also be enabled by the fact that all Springer e-books are indexed on Google at the chapter level. Furthermore, as Springer e-books sit alongside the e-journals on SpringerLink, they would be cross-searchable with a library's journal content.
■ Each e-book would still belong to an e-book subject collection and have a copyright year. Once the number of chapter downloads equalled the price of a collection, your institution would then own that collection in perpetuity and you would not be charged any more for content from within that collection. Similarly, should the price threshold be reached for an entire copyright year, again your institution would not be charged any more for content from that copyright year -you would own the entire copyright year and all the collections within that copyright year in perpetuity.
■ A maximum spend amount could be provided and once that was reached, access could be switched off if that was deemed necessary.
■ Where access was switched off but a collection or copyright year had been purchased (due to the number of chapter downloads), perpetual access would remain switched on for users.
■ Where access was switched off but a collection had not been purchased, access to that collection would be switched off -but it could be reactivated at any time. Upon re-activation the value spent on content from within that collection before the cut-off would be retained and spending would start from that point, not from zero spend.
■ No funds would need to be deposited in advance and Springer could invoice quarterly, bi-annually or annually depending on requirements.
The University of York Library agreed to Springer's suggestion of a three-month test project with no charges, so that both parties could evaluate the merits of the proposed model. We also agreed to present our initial findings at Springer's UK Library Advisory Board in February 2010. But what was in it for us (except for a threemonth trial to thousands of e-books!)?
Our current e-book provision, though still small compared to print books and other e-resources, is becoming established and is increasing year on year. In 2008/09 around 5% of our total information spend (on books, journals, databases, etc.) was on e-books; we spent £105,551 on 19,592 titles. The majority of this spend was on one-off purchases of individual titles, but the majority of e-book titles we provide are supplied via subscriptions to aggregator packages and publisher collections. In the same year, 15% of our one-off resources budget (books, DVDs, etc.) was spent on individual e-book purchases.
We are very interested in evaluating different e-book purchasing models and have just finished a user-driven pilot with MyiLibrary. We currently subscribe to some aggregator and publisher e-book collections, we have purchased a small number of publisher collections in perpetuity and we have purchased and continue to purchase individual e-book titles that precisely map onto teaching and research needs. At the moment, no single approach meets all our requirements, and we are prepared to make use of every available model in order to acquire the content we want in the way we want it.
In addition, with the backdrop of the global recession and cuts to UK higher education funding 1 , our budgets are tight 2 . The subscriptions budget especially is under pressure, having risen alarmingly over the past couple of years due to adverse currency exchange rates. Most of our one-off purchases budget is spent buying books which appear on student reading lists, nearly all of which are print. Demand for e-books is not yet explicit; from undergraduate students there is a stated, and oft repeated, demand for more print copies of books. Our library surveys reveal that all categories of users (undergraduates and postgraduates, research and academic staff) would like us to provide more electronic resources, but while we are regularly asked for 'more e-journals', there is not the same level of demand for 'more e-books'. For all these reasons, it can be difficult to justify buying e-book collections where demand cannot be predicted. To only buy what is used is therefore very appealing.
The distinctions between categories of content are also blurring: Springer's e-books are hosted on the SpringerLink platform alongside e-journals and cross-searchable with them; they often consist of conference proceedings or collections of chapters on a particular theme, and it could be argued that their content has more in common with journals than with the traditional research monograph. Historically at York the majority of spend on science information has been on journals, meaning little left in the science subject areas to spend on research-level book material, and no way of finding the funds without diverting money from journals.
The Springer trial commenced at the start of December 2009, when access was switched on and MARC records for all the trial titles (c. 12,400 from 11 subject collections) were loaded into our catalogue. Springer currently offer 12 subject collections but we already subscribe to one of them, Computer Science. Anyone searching our library catalogue, SpringerLink or Google would have discovered these e-books, but we did no publicity about them until early February, when we announced we had trial access. We deliberately did not mention anything about the pilot purchasing model behind the trial; this cautious approach to publicity was taken because we did not want to risk artificially inflating usage. Instead, we wanted to see whether there was an unvoiced need for this e-book content, and whether there would be a difference in usage pre-and post-publicity. Table 1 above indicates usage across the initial three-month trial period, from December 2009 to February 2010. The total usage according to the COUNTER BR2 report is given in the first row; this is then broken down further into titles available through the trial, and other titles. These 'other' titles include those in our subscribed-to collection, Computer Science, 1997-2004 volumes of the book series Lecture Notes in Computer Science which we had previously purchased, plus some titles with freely available content like front matter and back matter.
During these three months 1,561 different trial titles were used, around 12.5% of the total number of titles available to us. If we had paid £4 per chapter download (£31,488 in total, 4 x 7872), the total cost would have been significantly more than 12.5% of the total purchase cost of the e-book collections, had we bought them the conventional way. This suggests that further work on pricing is needed, not necessarily that the purchasing model itself is unworkable.
Did any patterns emerge from our usage? The statistics show that titles published during the 2009 copyright year were the most used, and that although usage does decrease for the older material, it still remains healthy (see Figure 1).
The two most-used trial collections were Mathematics and Statistics, and, Biomedical and Life Sciences (see Figure 2).
Looking in more detail at just one copyright year, 2009 (see Figure 3), it is clear that the proportion of usage per subject area is similar to that across the entire range of copyright years. There are some differences: for instance, Biomedical and Life Sciences and Medicine are proportionally higher, reflecting perhaps the more current nature of those disciplines compared to some of the others. Analyzing the usage of the 2009 copyright year also reveals that, had we paid £4 per chapter download, we would have reached the purchase threshold for three of the 2009 collections: Biomedical and Life Sciences, Business and Economics, and Mathematics and Statistics, and therefore retained ownership of those in perpetuity.
What happens next? Springer have extended our trial by another three months, taking it until the end of May 2010. That will give us a bigger pool of data to examine, and it will be interesting to see whether a different set of titles is used, and by how much the notional costs increase. It will also be interesting to work with them on pricing, and to have discussions about what the cost per download amount could be.
There are some obvious potential pitfalls to adopting this purchasing model if the pricing isn't right, and these centre around the loss of budgetary control; if a library were to spend the money it had set aside earlier than anticipated and had to withdraw access, the expectations of its user community could be severely disappointed, especially if access were lost at a key time such as the main dissertation-writing period.   This has been, and continues to be, a worthwhile pilot to be involved with. At a time when library budgets are under pressure there is a continual demand from our clientele to provide more resources, as well as to retain those to which we subscribe at an ever-increasing cost. The funds available to invest in new products such as e-book collections are decreasing. Libraries need to engage with more active collection management, and consider using different purchasing models to provide different sorts of resources: e.g. core and peripheral, research and teaching. Libraries and publishers working together to create and evaluate purchasing and pricing models, not just for ebooks but for all electronic resources, is at least one pragmatic way we can make progress in these challenging times.