Page |1 ebooks@cambridge and PDA This document contains a summary and full report on the trial of Patron Driven Acquisition (PDA) for ebook purchasing at Cambridge University April - May 2013. The trial was administered by ebooks@cambridge and covered three subject areas using the dawsonera platform. It was financed by part of the interest from the Connell Fund. While librarians were notified that the trial was taking place, it was not publicised to staff and students. Summary Out of the 8,329 selected ebooks which were made available during the trial, 117 were purchased at a total cost of £9,784.45. This gives a mean average price per ebook of £83.63 (inclusive of VAT). Each ebook will be available in future to Cambridge staff and students via LibrarySearch, according to the licence restrictions imposed by the publisher. Time needs to pass before we can judge how useful are the titles purchased using PDA, and how this compares with ebooks purchased by other methods, such as ordering titles by librarian selection and reader recommendation, or as packages. Running the trial has given the ebooks@cambridge service very useful experience in using the PDA model. A number of issues which emerged are highlighted below. Usage With PDA there will no doubt be a myriad of reasons why our users have clicked on titles, wanted to continue reading past the 5 minute preview and unwittingly triggered purchases. It will be interesting to chart, through studying usage statistics over time, if some titles are obviously in-demand type ebooks, and whether those popular ones are or perhaps more interestingly, are not on reading lists. It will also be of interest to note if there are any titles which are only used once or twice. Can we assume these titles are more likely to have been triggered by research students rather than undergraduates? 27% of titles were purchased despite there being print copies in the UL, and some Faculty and College libraries; does this indicate a real need for an ebook copy to supplement the already popular and well-used print copies, or is this a sign of a user taking the easy option of browsing through the ebook first to ascertain whether to make the effort to go and borrow a print copy? The use of the PDA purchased ebooks should be compared with the use of those purchased through other methods, possibly in January 2014. It would also be interesting to compare these figures with those for the issue of print copies, where held. Without further research, we cannot say yet what the value of these ebooks will be, or how this relates to the value of print copies. Anecdotal evidence from the Education Faculty Library is conflicting. When a couple of their readers expressed an interest in certain titles, staff encouraged them to persist with reading in order to secure the purchase of the book, which suggests the purchase was of value, but when a reader who had suggested a title for purchase once the trial had finished was asked if he would be happy with a print copy, he replied “at the moment I do not need this book”. This suggests that some, at least, of the PDA purchases were titles stumbled upon and browsed out of passing interest, very much like browsing the shelves of print books. It is also possible that readers viewed the ebooks before deciding to make the journey to a library to borrow a print Jayne Kelly, ebooks Administrator June 2013 Page |2 copy. Catalogue During the trial, access to the PDA collection was provided on the supplier’s platform and through records loaded into the LibrarySearch catalogue. Adding a record to the catalogue traditionally implies that whatever it refers to has been acquired by the library. However with PDA there is no certainty that the ebooks which appear in the catalogue will remain there when the purchasing period is over. This is a particular issue for academics who rely on the continued availability of texts, and for librarians who are checking the availability of material on reading lists. Some method of identifying in the catalogue which titles are included in a PDA offer would be useful. Alternatively, we could ignore importing catalogue records until a purchase has been made, in which case the availability of the trial collection would have to be promoted. Duplicates The issue of avoiding duplicating ebook purchases needs to be considered. Every effort was made during the trial by the ebooks@cambridge team to avoid purchasing an ebook which we already held and, despite extensive list preparation and checking, duplicates were discovered as a result of titles appearing in more than one subject profile and in three cases duplicates of existing purchases were made (subsequently withdrawn and credited). Avoiding such duplication occupied considerable administrative work before and during the trial and is not likely to be scalable. Should we therefore accept that duplication in ebook purchasing and associated costs is inevitable, or should we ensure that it is kept to a minimum? There is a cost either way. Also, an issue this trial did not touch on, is the extent to which we balance ebook acquisitions with existing print holdings, i.e. should we concentrate on purchasing ebooks of titles not held in print? If so, how? Purchase orders Purchase orders and invoices had to be created manually in Voyager after purchases were triggered. This was labour intensive and would be far more so if the PDA budget was substantially larger. Vendor error Potential purchasing was compromised in the first two weeks of the trial due to an error by Dawsonera which meant that readers could only preview the first 50 pages of any title within the 5 minute previews. We had assumed this was standard and how the PDA should work. Apparently not, as when dawsonera realised they had not switched on the correct set up and changed this, our ebooks suddenly started allowing users to access the whole ebook content. I queried this, thinking it was a bug and found out that our users should have had full-text access during the previews all along. I would have preferred dawsonera to have admitted their mistake and informed us when they changed the settings, rather than having to go chasing for an explanation. Dawson also halted the trial before the budget was exhausted, leaving a £300 balance. This possibly deprived the trial of further PDA purchases, but the balance has been used to purchase in demand ebooks. Jayne Kelly, ebooks Administrator June 2013 Page |3 Full report PDA (Patron Driven Acquisition) allows a batch of selected ebooks to be made available to readers on a suppliers’ platform, either for a predetermined period of time or until a specified budget has been exhausted. To facilitate discovery, a record for each ebook is added to the library catalogue. Readers can access all the content and trigger purchases by specific actions. When the limit of the budget is reached, or the period of availability ends, the purchased content only remains available and access to the remainder is switched off. Preparatory work: June 2012 Following agreement by the ebooks@cambridge Advisory Group, a talk was given to Cambridge librarians by a guest speaker, Sarah Thompson, Content Acquisition Librarian at the University of York, summarising PDA models offered by academic library ebook vendors. November 2012 At a meeting of the Connell Fund group of librarians, together with Jayne Kelly and Sarah Stamford of the ebooks@cambridge Advisory Group, the decision was taken to proceed with a PDA trial in 3 subjects: Archaeology & Anthropology, Education & Philosophy. The budget was set at £10,000 sourced from the Connell Fund interest 2012-13. The other half of the Connell Fund interest was set aside for purchasing individual ebooks as requested by the SPS and Psychology Libraries. It was decided not to publicise the trial, other than to inform Cambridge librarians that it was taking place via the lib-list email. With limited funding the group did not want purchasing to be “hogged” by an individual or particular interest group. Vendor The Connell Fund librarians and the ebooks@cambridge team selected dawsonera for the trial, following a meeting with their representative, Cheryl Fellows. Dawsonera was already an ebooks@cambridge supplier, with a good track record of PDA provision to other universities. Purchase triggers It was decided to use the following two triggering options in combination • “Continue Reading” – a user reading an ebook would be asked after 5 minutes if s/he wanted to continue reading. If “yes” a purchase was triggered. • “Minimum number of Previews” – Each ebook could be “previewed” for up to 5 minutes. Each ebook could be previewed three times without charge; at the 4th attempt a purchase was triggered. An option was also included on the platform to “Suggest a purchase” which sent an email to the ebooks Administrator. Alternative triggers, such as the rental model, were considered but on this occasion discarded. Jayne Kelly, ebooks Administrator June 2013 Page |4 Profiling and selecting the range of ebooks for the trial: December 2012 - February 2013 Excluding owned ebook content It was necessary to exclude from the trial any ebooks which had already been purchased. In liaison with UL Digital Services staff, an Excel spread sheet with over 18,000 line items was provided to dawsonera. This necessarily included the isbns for the print copies of the titles which corresponded to our ebook holdings (it is not possible to match on e-isbns as they can be unique to each ebook supplier). Each print book generated multiple listings, due to the multiple isbns in the catalogue records. The dawsonera model would perhaps work better for those libraries who use 1 bibliographic record and different holdings records for their print and e content. Profiling Cheryl sent a PDA profile template for completion by the participating Faculty librarians, Aidan Baker, Angela Cutts/Louisa Brown and Jenni Lecky-Thompson. This profiling allowed the subject librarians to ensure that the ebooks included in the trial would be those which would normally be appropriate purchases for their faculty staff and students, and to exclude the rest. The criteria included in the initial selection were: English language; LC or Dewey code; excluded isbns; publisher and imprint; readership level; maximum price; publication date. The 3 completed templates were sent to dawsonera on the 20th of December. At this stage we learned that dawsonera could only deal with 1 subject profile at a time, so we decided that they should send the list of titles in the following order: 1. Philosophy 2. Education 3. Archaeology & Anthropology The initial lists of titles came with Dewey class marks (some with LC codes where provided by publishers). At this stage some date and price limits were imposed (on advice from dawsonera): Philosophy: no date or price limits. Education: no titles published pre 2008: £200 upper price limit Anthropology/Archaeology: no date limits: £200 upper price limit The profiles were further refined as follows. 1. The Philosophy title list contained 6,233 titles. These were sent to Jenni and through a process of gradual refinements (listed below) and quite a few emails back and forth to dawsonera this was whittled down to a final selection of 2,514 titles. a) a price limit of £200 b) excluded Dewey classes: 100-109.9, 133-139, 174.2-174.9 & 180-189 Jayne Kelly, ebooks Administrator June 2013 Page |5 c) excluded 'Technical', 'General' & 'College' readership levels d) excluded anything not in the English language e) excluded any titles published by CUP (as the promised Credit fund should be used in future to purchase ebooks on CBO) Jenni also removed about 300 already owned titles from the list. Jenni reported that the profile checking process was 2 – 3 hours of work in total. The final list was returned to dawsonera and the resulting MARC records were loaded into the FTP space for collection. There were issues with the MARC record files which are detailed later on. 2. The Education list arrived next and contained 3,905 titles. Louisa refined the LC classmarks slightly and, with the removal of a few extra titles the list was finalised at 1,928. The refinements were as follows: a) price limit of £200 b) only included Dewey classes: 100, 150-153, 155-158, 207, 300, 370-379, 307, 507, 607, 707, 780.7 & 907 c) excluded 'Technical' & 'General' readership levels d) excluded anything not in the English language e) excluded any titles published by CUP (as the promised Credit fund should be used in future to purchase ebooks on CBO) f) no titles published before 2008 Louisa spent a minimal amount of time on the profiling as she and Angela wanted to keep the range of titles quite broad. The Education MARC records were then supplied to the FTP server for collection. 3. When the Archaeology & Anthropology list arrived it contained 22,000 titles, this was even with the £200 price limit and excluding all CUP items. Thanks to Aidan’s efforts this total was refined to 3,887 titles, the exclusions follow: a) price limit of £200 b) ) only included Dewey classes: 069, 133, 155, 170, 291-299, 300-307, 909-919, 930-942, 960 & 972 c) excluded 'Technical', 'General' & ‘College’ readership levels d) excluded anything not in the English language e) excluded any titles published by CUP (as the promised Credit fund should be used in future to purchase ebooks on CBO) f) no titles published before 2006 g) restricted publishers to not many more than those included in the Haddon’s classification backlog Many of the titles included in the trial, especially in the Archaeology & Anthropology profile, were interdisciplinary, e.g. titles in History, Art, International Studies and Sociology were purchased alongside those of a more defined anthropological or archaeological nature. Aidan noted that some publishers that he would have liked to have seen included were missing from the profile. Dawsonera have been advised of this. A list of these is included at the end of this report, Jayne Kelly, ebooks Administrator June 2013 Page |6 in Appendix A. Aidan spent approximately 3 hours working on the profile. MARC catalogue records The MARC files containing the 8,329 records for the PDA were sent as .usm files and very large ones at that, so initially we had to request that dawsonera split the records so that they could be opened in MarcEdit. Then I asked advice about converting .usm to .mrc for loading as this was not proving to be simple. I had requested .mrc files from dawsonera but they maintained that the files could be converted to .mrc by simply changing the suffix on the files. While at UKSG in early April I discovered that other libraries had requested files in .mrc and dawsonera had provided them, so I had a word with Cheryl Fellows on the Dawson stand and the correctly formatted files turned up in my inbox the very next day! All 9 file batches were checked in MarcEdit for correct coding; work was need on the 006 field, a 948 field was added along with holdings. The records were loaded on the 23rd and 24th of April and were all searchable in LibrarySearch by lunchtime on the 25th. A selection of URLs were checked to ensure they linked to the ebook content correctly. Dawsonera reported that enhanced MARC records are sent as those received from the publishers for the PDA were not always of a high quality, but at this stage there has seemed no need to load the enhanced versions. Dawsonera went live with the PDA trial at 4pm on Thursday 25th April. During the trial: April - May 2013 The trial began on 25th April and finished on 20th May, and was run at this time of year because we were ready to go by then, not because we had decided this was the optimum time. However, the librarians felt that the exam revision period was an appropriate time as those taking exams would be likely to be consulting books which were essential reading. The total budget for spending on titles was £8,750 (£10K + 5% discount – 20% VAT). The only unowned content our users could now access on dawsonera were the 8,329 titles that comprised our subject profiles. Almost immediately the trial began we started to spot duplicates in the collection, 15 of these were reported to dawsonera who removed them from the platform. The ebooks@cambridge team removed 40 catalogue records from the ERDB (LibrarySearch). Some of these records were a result of titles appearing in more than one subject profile, thus generating 2 catalogue records but only one incidence on the platform. Every time we were notified that an ebook in the trial had been bought, we checked that the title was not one we already owned as an ebook; and this effort was justified in 3 cases by our discovery that an existing purchase had been duplicated. Number of hits 464 titles were previewed (or hit) during the trial period; of this 117 titles (25%) were purchased. The breakdown of previews is shown in the table below. From this we can deduce that 3 titles were purchased by the “Minimum number of previews” trigger and 114 by the “Continue Reading” trigger. Jayne Kelly, ebooks Administrator June 2013 Page |7 1 preview 2 previews 3 previews 4 previews Total titles No. of titles 385 60 16 3 464 Percentage 83% 13% 3.4% 0.6% 100% Purchasing and Orders Order confirmations were sent to the ebooks@cambridge team email from dawsonera throughout the day (on the hour) and the electronic order confirmations containing the ebook URLs were sent in the evening. The first purchase was triggered within an hour of the trial going live, and continued steadily until the fifth day when the trickle became more of a stream and 7 titles were triggered, a likely outcome given that the triggers for purchase were repeated instances of reading. From then on there was a constant flow of PDA batch confirmations arriving from dawsonera. The trial ran over 5 calendar weeks and during weeks 3 and 4 the triggers reached over 40. Overall Mondays and Tuesdays were the busiest purchasing days, Saturdays the quietest. The purchasing peak was Tuesday 7th May with 11 ebooks purchased. Just under 67% of purchases were generated in the afternoons and evenings. Purchases by date 12 10 8 6 Purchases 4 2 0 During the course of the trial we were invoiced for 3 ebooks we already owned through MyiLibrary. I am pleased to report that dawsonera were happy to remove these titles from the platform after a purchase had been triggered, we were not invoiced for the cost. Invoices Invoices and enhanced Marc records for purchases were emailed weekly; this could have been twice Jayne Kelly, ebooks Administrator June 2013 Page |8 weekly if required. There was a blip with invoicing of the earlier titles and dawsonera seem to be playing catch up with these, there remain 2 titles still to be invoiced. Feedback from librarians and users Feedback was limited as the trial had not been publicised beyond informing librarians. Four queries were received from Faculty librarians : one asking about record removal once the trial was finished, another one querying the 50 page limit for the preview (on behalf of a student) and another querying why the 50 page limit has disappeared. There was also a query about checking our holdings in the case of reading lists. With the unowned content in LibrarySearch it would be necessary to click on beyond the catalogue record and into the dawsonera platform to check whether any title was purchased or not. Once there, if a title is highlighted in green then we own it, if purple we don’t. An email to ‘liblist’ was sent mid-trial to highlight this issue. Suggestions for purchase 7 suggestions for purchase were received from users during the trial, 4 of these titles were subsequently purchased as part of the PDA. Any purchase suggestions are sent in weekly batches by dawsonera; users have to type in their email address and they can leave a short comment if they wish. All users were emailed a response which outlined how they could continue reading the fullcontent if they wished and 2 readers responded to the emails to express their thanks. After the trial The PDA trial ended with the purchased of the 117th ebook at 02.27am on the morning of Monday 20th May, it had been live for 26 days. Removing un-purchased content I cannot yet report back on this step of the process. I was told that dawsonera will send me a series of MARC files which I can use to delete the un-owned titles; I have since discovered that there is a standard 6-8 week wait for these files but there has been no explanation to justify this length of delay. I have received 9 suggestions for purchase since the end of the trial. These are being passed to appropriate Faculty and Departmental Librarians for information and if necessary for purchase where budgets allow. A meeting between the Connell Fund librarians, Jayne Kelly and Sarah Stamford was held on 20th June to review the trial. Those attending had been circulated with full details of the purchases, usage statistics and the budget for 2012-13. The meeting proposed a further PDA trial should be considered, possibly with a different supplier and subject areas, and at a different time of the academic year. One College reports having used the list of PDA purchases to identify three titles for purchase in print for their library. Jayne Kelly, ebooks Administrator June 2013 Page |9 Conclusions Analysis of purchases Subject coverage The spread of purchases across subject profiles is shown below (please note that 1 title appeared in the Arch & Anth and Education profiles so has been counted twice). Subject profile Archaeology & Anthropology Philosophy Education Total Number of titles purchased 72 Number of titles in trial 3,887 31 15 117 2,514 1,928 8,329 Costs £9,784.45 was spent on 117 titles, and the average cost per title was £83.63. The cheapest ebook cost £11.39 and the most expensive cost £162.11, in the context that there was a £200 price limit on profile titles. 32.5% of titles cost over £100. A graph showing the range of prices is detailed below. Price (inc. discount & VAT) 180 160 140 120 100 Price (inc. discount & VAT) 80 60 40 20 0 0 20 40 60 80 100 120 140 Print holdings The graph below shows the extent of print holdings for titles purchased as ebooks in the trial. Of the 117 purchases, only 10 were not held in print somewhere, 33 were held in the University Library only. As we are not able to track which users purchased which titles, we can never really know which type of student or academic staff or other purchased these titles, the only surety is that they had a Raven login. Jayne Kelly, ebooks Administrator June 2013 P a g e | 10 Print holdings FDL & College UL, FDL & College UL& FDL FDL only Series1 UL & College College only UL only No print copies 0 5 10 15 20 25 30 35 Purchase model The credit model accounted for 113 of the purchased ebooks, the other 4 were sold as managed user access which allows a designated number of simultaneous users at once, and the default seems to be 3 users. Publisher coverage Routledge publications proved the most popular with 31 purchases (26.5%), with Palgrave Macmillan (12%), Wiley-Blackwell (10.3%) and Oxford University Press (10.3%) following behind. A pie chart illustrating publisher distribution is shown below. “Other” includes those publishers with less than 3 titles triggered. There were no Cambridge University Press titles included in the profiles. Publisher coverage 3.4% 3.4% Routledge 2.5% 2.5% Palgrave Macmillan 3.4% 4.3% Wiley-Blackwell 26.5% Oxford University Press 5.1% Other Duke University Press 6% 12% 10.3% 10.3% 10.3% Bloomsbury Continuum UK Boydell Springer Verlag Sage University of Chicago Press Pluto Press Open University Press Jayne Kelly, ebooks Administrator June 2013 P a g e | 11 Appendix A. A list of publishers noted by Aidan Baker as missing from the Archaeology & Anthropology subject profile. Archaeopress British Academy British Museum Columbia University Press Editura Mega Harvard University Press Manohar Max Planck Institute Mora Ferenc Muzeum Norton Oxbow (David Brown) Pabst Peeters University of Arizona Press University Press of Florida Wachholtz Wits University Press Jayne Kelly, ebooks Administrator June 2013