Yesterday, we talked about filtering; today, Terry Bucknell suggests, we're looking at the opposite - buying by the bucketful. A multidisciplinary university like Liverpool (where Terry is e-resources manager) buys a wide range of content, for example, 19 ebook packages in 5 years. These have traditionally been bought with leftover budget at the end of the financial year (n.b. Liverpool decided ebook packages are better than journal backfiles here). As "leftover budget" becomes a thing of the past, Terry is analysing the value from these packages in more depth and always considering alternative purchasing models - individual title selection, patron-driven acquisition, etc.. Here are some highlights of Terry's analysis:
- 40% of Liverpool's e-resource usage is e-books - yet 95% of the budget goes on journals.
- Liverpool's usage is typical - approx 40% of titles in a collection are used in the first year; approximately 60% have been used by the second year
- Some subjects (e.g. mathematics, at Liverpool) seem to perform badly - is this a factor of how/when information is used in different disciplines? need to be careful before making collection development decisions based on this data
- All types of books get used at least a bit, but some content (e.g. conference proceedings) is used more than other content (e.g. monographs)
- Pareto principle applies! 80% of downloads from top 21% of ebooks - Terry doesn't think this should be a factor in how collections are purchased / priced. (Looking more closely, 35% of usage on one platform came from one title! - doesn't tell you anything about the broader collection, just that some books are heavily used)
- Even on aggregator platforms (where there's a greater level of individual title selection than a publisher package), a third of ebooks have had only 1 or 2 accesses during 2 years
- With patron-driven acquisition, all ebooks are used (because you don't buy them unless they are) - so should be better value? Terry used ebrary model (purchase triggered by 10 page turns / 10 minutes in a title / copy & pasting / printing) to analyse Liverpool's Springer ebook usage stats and calculated that PDA costs would overtake package costs in just one year in most cases (even when cheaper backfiles were excluded from analysis).
- Evidence from elsewhere (e.g. U Iowa ebrary pilot) also shows that PDA budgets run out quickly - libraries who started trials had to resort to buying packages after all
- ... other PDA models are available ... (and may show different results) but Terry found that a PDA model would have to allow for "6 chapters free" before it would be comparable to package pricing.
Experiments like this can give libraries and publishers an idea of what is a fair price to pay for an ebook package; Terry's conclusions:
- Some packages are better value than others, and libraries should prioritise these in collections
- Aggregated databases give cheap critical mass
- Single title selections are important for core texts
- PDA can fill the gaps, but not form the foundations
The implications for libraries:
- Need to centralise book budgets - stop fragmenting by formats etc - a hard sell for lots of faculty / librarians
- Rapid move to e-only book acquisition - implications for logistics / staffing
The implications for publishers:
- Packages need to be at least 50% discount for it to be worth it for the library - make it a "no brainer" (70% discount) for the library to purchase, and you'll solve the budget crisis
- Offer combined books / journals packages with appropriate cost weighting / discounting.
Labels: analysis, database, ebooks, model, package, pda, purchasing, uksg