The Big Deal Approach to Acquiring eBooks: A Usage-Based Study

This paper presents the results of an analysis of COUNTER eBook usage reports for SpringerLink eBook collections purchased since July 2008 at the University of Liverpool. The usage reports were augmented with contextual title-level information drawn from Springer’s eBooks Title List. The combined data was used to study how usage of eBooks is influenced by factors such as the subject area, the year of publication and the length of time since the collection was acquired. Analysis was performed to simulate the effects of user-driven purchasing to determine whether this model could apply to this type of content. The study concludes that big deal approach has worked well: all but one of the subject area have been well used, the number of unused titles continues to diminish each year, older titles continue to attract significant usage, and the cost-per-use is relatively low. The challenge remains to find pricing models that allow more libraries to acquire eBook collections when budgets may be largely committed to journals.


Introduction
The University of Liverpool Library's first significant foray into eBooks occurred in summer 2006 with a subscription to Ebrary Academic Complete. This provided a critical mass of titles and usage grew rapidly. User demand led to that collection being augmented through the acquisition of single-purchase titles through Ebrary and later other eBook aggregators, Dawsonera and Myilibrary.
Whilst the ability to access books online was positively received it became apparent that users disliked Digital Rights Management intensely. Feedback from library users was that they expected to be able to view, download and print eBook chapters just as they already did with eJournal articles: as DRM-free PDF files. As a result the library amended its strategy to increasingly buy eBooks on publishers' own platforms. These eBooks were often in bundles acquired at the end of the financial year if sufficient funds were available.
The library first acquired of Springer eBooks, the complete collection for copyright years 2005 to 2008, at the end of the 2007/08 financial year in July 2008. The following July the complete 2009 collection was purchased. The timing of these purchases inevitably resulted in some duplication, as titles that had been bought in print or as single eBooks through aggregators were later acquired in the collections. To avoid this double-purchasing, the library amended its strategy to purchase the 2010 collections at the end of 2009, mirroring the journal purchasing timetable. Any purchasing of Springer books outside of this arrangement is heavily discouraged. To help satisfy demand for print, Liverpool became one of the first libraries in Europe to join Springer's MyCopy program 1 . MyCopy allows users to order any Springer eBook (with a few exceptions) as a personal Print-on-Demand soft cover print book for a heavily discounted price.
As part of its acquisition of the 2010 Springer eBook collections, University of Liverpool Library agreed to partner with Springer to study how eBooks were being used and received, through a three part project: analysis of usage reports, an extensive online user survey, and through focus groups. This article reports on the first part of this study -the analysis of usage reports.
Standard COUNTER BR2 reports were augmented with data from Springer's eBook Title List, using Excel's VLOOKUP function 2 to bring contextual title-level information, like Copyright Year and Subject into the usage reports.

About Springer eBooks
Springer eBooks are sold in 12 (formerly 13 but Birkhäuser Architecture has recently been sold) annual subject collections, with each year's collections containing around 3,500 titles. From 2005 forward all Springer book titles have been published online as well as in print. The collection is strong in STM subjects plus business and management, with smaller number of titles in other humanities and social science subjects. The eBook collections are characterized by a high proportion of research monographs, contributed volumes and conference proceedings but also include textbooks and major reference works. They are mainly aimed at researchers from the advanced undergraduate level upwards.
The SpringerLink platform presents eBook chapters alongside eJournal articles on a single interface. Each chapter is a PDF file that may be viewed, downloaded or printed in its entirety. MARC records are provided for each collection from 2005 onwards, and COUNTER-compliant usage reports are provided.

Results of eBook Usage Statistics Study
Overall use of eBooks It is notable that the use of Springer eJournals increased significantly between 2008 and 2009, even after excluding the usage of an additional collection that was added to Liverpool's Big Deal in 2009. This suggests that having access to eBooks on the same platform as eJournals does have an inflationary effect on the usage of the eJournals. This will be closely watched to see if this pattern continues into 2010 and also if there are similar effects on other combined-content platforms.   Table 1 sets out the constituent parts of the ebook collections that were purchased together with their respective eBook usage. The majority (35%) of the eBooks in the collection are scholarly monographs and they attracted the highest volume of usage in absolute terms with their percentage of downloads being roughly equal to their volume contribution. Usage of Proceedings was below their contribution in terms of downloads but this is perhaps because users are more likely to just want a single article/chapter from this type of publication. Professional books and textbooks outperform their contribution which could be explained by the fact that these books generally have a broader appeal, especially at the undergraduate level. Reference works also significantly outperformed in terms of their % contribution to chapter downloads which can be explained by their large number of individual entries.

Year of Publication
One major area of interest was the longevity of interest in scientific eBooks: would they cease to be of much interest once they were more than a few years old? Table 2   springer.com Figure 2 shows that once the 2009 collection had been purchased, its monthly share of the downloads quickly surpassed that of the earlier collections, even though that collection was not complete until the end of 2009. Nevertheless, titles from each of the previous years continued to make a significant and sustained contribution to the overall usage, much more so than older eJournals. Figure 3 shows that different subject areas exhibit much consistency in the usage, with between 40% and 60% of titles used in all of the subject collections. The one exception was Mathematics & Statistics where fewer than 30% of titles were used. This mirrors the eJournals usage in Mathematics at the University which is lower than most other subjects. This may be a reflection that mathematicians read less text based material than other disciplines, still prefer to read from printed sources, or need to read older sources that are not available online.  The Big Deal Approach to Acquiring eBooks: A Usage-Based Study 5 springer.com

Differences between subject areas -Computer Science and Biomedical and Life Sciences
Computer Science and Biomedical and Life Sciences are two subjects that are generally perceived to have a heavily reliance on the most up-to-date research, generally based around journals and conference proceedings. These two subject collections were investigated in more detail to see if this affected usage patterns. Figure 4 shows that even these subjects show little decline in the proportion of titles used in older collections. Figure 5 shows that fewer chapters were downloaded from older collections in Biomedical and Life Sciences but that there was little effect in Computer Science.

Effectiveness of book selection policy
The University of Liverpool Library acquires print books only when specifically requested by an academic department, either for teaching or research purposes, or when usage of existing copies demonstrates that additional copies are required. The library does not engage in approval plans and although some titles are acquired through standing orders the use of these is declining. Figure 6 shows the relationship between the total circulation figures for print books and the usage of the same titles in eBook format in 2009. Overall, there was little correlation between loans of print copies and their online usage. Indeed, if anything it appears that the titles that were borrowed most were used least online. This confirms that when users want to immerse themselves in a book thoroughly they prefer to read a print copy, but they are happy to use an online version if they just want to read a chapter or two. Nevertheless, the titles that had been bought in print performed slightly better online than the collection average. Of the titles that were held in print, 63% were used online in 2009, compared to 48% of the all titles in the collection. The library only owns 3.4% of the titles in the 2005-2009 collections, but they accounted for 4.4% of the titles that were used and 6.2% of the chapter downloads.
Overall in 2009 alone, 48% of all eBooks from 2005 to 2009 were used at least once which compares well against the widely-quoted figure of 40% of print books being unused six years after purchase 3 . In contrast, fewer than 40% of the purchased Springer eBooks haven't been used within two years of purchase.

Pattern of eBook usage from year to year
Several institutions and aggregators are currently experimenting with a user-driven purchasing model for eBooks. This seems an attractive option for many libraries to ensure that every title that is purchased is used at least once. These models typically open up access to all titles for a limited period of time (either a fixed period or until the money runs out). At the end of the period, access to the non-purchased titles is switched off. Would this approach work if made available for Springer eBooks? To illustrate this further, only 11 of the 50 best-used titles in 2009 were also among the best-used titles in the first quarter of 2009. Thus if usage in the first quarter of 2009 had been used to select 50 titles for purchase, then 39 of the year's most heavily-used titles would have been missed, including the best-used title, and the 4th , 5th, 8th, 9th and 10th best-used titles. Equally, the library would have purchased some low-use titlesdown to the 473rd best-used for the year.

Frequency of use
Analysis of the number of months for which each title was used showed that only one title had been used every month in 2009 and fewer than 5% of the titles had been used in 5 or more months of the year. More than 50% of titles were used between 1 and 4 times a year, and more than 45% of titles were not used at allsee Table 3. Note that the 'long tail' of infrequently used titles forms the bulk of the usage: titles used between 1 and 4 times in 2009 constituted 77.6% of the chapter downloads in 2009. This is not a case of a few highuse titles dominating usage of the collection: the breadth of the collection is key.
It should be noted that these figures just refer to usage in one calendar year. The longer a collection has been owned, the greater the proportion of the collection that will have been used. Figure 8 shows a remarkable consistency in the way that an increasing proportion of each collection is used over time: About 12 months after acquisition, around 45% of the titles in each collection have been used; by 21 months over 60% have been used. It is quite remarkable to consider that 45 titles published in 2005 and purchased in July 2008 were used for the first time in March 2010 -and 4 of those had 10 or more chapters downloaded in that month. It will be interesting to investigate whether collections purchased from other publishers exhibit the same shaped curve.

Measurement of eBook and eBook Collection Value
It is difficult to calculate and compare the value of eBook collections, in part because of the variability in how eBook platforms function and how they report usage 4 . COUNTER released its code of practice (CoP) for books and reference works in March 2006. This includes the COUNTER Book Report 2 (BR2) Number of successful section requests by month and title which has been implemented by many aggregators and publishers since. However, the COUNTER CoP defines a 'section' merely as 'a subdivision of a book or reference work' . This means that (depending on the platform) a dictionary definition, a single page, a group of five pages or a chapter can all count as a 'section' . Neither does the report state how a 'section' is defined for that platform; that is left to the librarian to know, or find out. onwards that should be included in the calculation. This would mean it would take until the end of time to calculate the true value of the purchase! However, these arguments apply equally to eJournals where it is the norm to just divide the year's payment by that year's usage, so it seems justified to apply that same simplified calculation to eBooks purchased in an annual collection.

Conclusions
Through fairly simple analysis of COUNTER reports it has been shown that acquiring Springer eBooks as a 'big deal' has been a good value investment for the University of Liverpool. Each subject area has been well used (with the usual exception of Mathematics), the number of unused titles continues to diminish each year, and older titles continue to attract significant usage.
Acquiring a 'database of book chapters' seems to be the best approach for research-level eBooks (research monographs, contributed works, conference proceedings, etc.). Coupled with Springer's MyCopy programme this approach could help shift the library to a 'primarily electronic' book acquisitions policy for this type of material.
The study has highlighted that STM students and researchers make good use of book material even though STM faculty prefer to allocate their library budgets almost entirely to journals. At the University of Liverpool this leaves little scope for finding funds to acquire eBook collections, except through end-of-financial year purchases. As more publishers start producing bundles of eBook content, perhaps with a view to bundling this with eJournal content, they need to understand that book budgets are often dwarfed by journal budgets and they may need to fundamentally rethink their pricing models so that libraries' book budgets can accommodate such offerings.