Putting Usage Statistics to Work in the Purchasing Life Cycle

1
Putting Usage Statistics to Work in the
Purchasing Life Cycle

• Terry Bucknell
• Electronic Resources Manager
• 26th February 2007

2
Overview

• How The University of Liverpool Library has used
  usage statistics in its e-resources purchasing
  decisions in the last 6 years
• Deselection
• Selection
• Monitoring / Renewals process
• Assessment / trials

3
Institutional Context 2001

• The University of Liverpool
• One of the UK's leading research-led universities
• A Russell Group member (top 19 research-led
  institutions)
• Very wide range of subjects (inc Vet.Sci and
  Civic Design)
• But
• Very cautious financially large cash reserves
• The worst funded library among the UKs Civic
  Universities
• Journal online access often gone unclaimed
• Very few e-journal Big Deals
• Major databases as networked CD-ROMs
• No e-books

4
My Context 2001

• Arrived from a similar institution
• but with a better funded library!
• Former role was as a subject librarian
• Positive reaction from users to web databases and
  Big Deals
• My ambitions for Liverpool
• Convert CD-ROMs to web
• More Big Deals
• A bigger share of the budget for e-resources
• My subject background is Physics, so
• What evidence (i.e. data) can I collect to
  support my ideas and free up funds?

5
Sources of Usage Statistics in 2001

• Initially, we werent collecting any usage
  statistics
• We found we could get usage statistics from
• 11 e-journal sources
• 10 database sources
• 0 e-books sources (what are e-books?!)
• 1 CD-ROM source (our own server, 46 CD-ROMs)
• With few e-resources, and few usage statistics to
  manage, we had plenty of time to analyse the
  statistics that we did receive!

6
CD-ROM vs. Web Database usage stats

• Were able to convert several medical databases
  from CD-ROM to web for little extra cost
• We compared the usage statistics for the two
  formats
• Slightly higher cost much higher usage

7
Thats Hardly Rocket Science!

• No, but the mindset in place was
• Web databases are probably better than CD-ROMs,
  but we cant afford them, sigh
• Some simple analysis allowed us to predict
• How much more usage wed get if we moved to the
  web version
• How much more satisfied our users would be
• How much better value for money wed be getting
• So lets make our users and our funders happier,
  and find the money!
• Figures like this help to build a case

8
Aside Whats in a Name?

• CAB Abstracts CD-ROM to CabDirect on the web
• Usage hardly increased
• Downgrade to subset Veterinary Science Database
• Usage more than doubled (at a fraction of the
  price)!
• Impressive increase in value for money!

9
Usage of Web-Based Databases

• For web-based databases, we calculated
• Cost-per-session and Cost-per-search
• Bore differences between resources in mind
• Full-text database vs. AI database
• Unique information vs. availability in other
  resources
• Concluded that two of our high-quality
  subject-specific databases werent worth the high
  cost
• Consulted academics nobody shouted
• Cancelled to free funds (for e-journal Big
  Deals?)
• Offered mediated searches instead zero take-up
• Researchers will make do with interdisciplinary
  databases, but want more full-text

10
Analysis of Big Deals

• We only had a couple of Big Deals
• ScienceDirect
• Blackwell Science
• Are we paying for titles we dont really want?
• Or are we restoring cancelled titles at a
  fraction of the price, and getting lots more
  useful titles too?
• Usage statistics can settle the argument
• Usage of subscribed vs. unsubscribed titles
• Is there a core and a periphery?
• When we pick and choose, do we choose the right
  titles?

11
Subscribed vs. Unsubscribed

• ScienceDirect usage stats (for 2006!)
• Blackwell Science showed a very similar pattern
• On average, a subscribed title is used more than
  an unsubscribed one
• Downloads per subscribed title 507
• Downloads per unsubscribed title 140
• We only subscribe to 18 of the titles in the Big
  Deal
• 56 of our downloads are from unsubscribed titles
• By taking the Big Deal our researchers are able
  to download 127 more articles (225,000)
• Were only paying Elsevier a few extra
• This Big Deal is good value for money for us!

12
Core and Periphery?

• Only a tiny proportion of titles are really core
• But a huge proportion of our downloads come from
  the tail

13
Core and Periphery?

• 750 titles with between 101 and 1,000 downloads
• 727 titles with between 11 and 100 downloads

14
Distribution of Subscribed Titles
15
Subscribe to the High-Use Titles?
16
Subscribe to the High-Use Titles?

• Departments decide which titles they want
• Theyll want the best titles in their subject
• Or at least the ones they can afford
• E-journal usage varies between subjects
• Unsubscribed biomedical titles will naturally be
  used more than subscribed archaeology titles
• If we subscribed purely to the high-use titles
• Wed need many more (expensive) biomedical
  subscriptions beyond our biomedical budget
• Would cancel (cheaper) subscriptions in other
  subject areas - but they still need to spend
  their budget!

17
Analysis of Big Deals Conclusions

• Big Deals suit us (particularly well?)
• Lots more titles and usage for little extra cost
• We subscribed to fewer titles than a library of
  our type should be able to very tight
  rationing!
• Restore access to cancelled titles at a fraction
  of the cost
• Cancellation clauses not too problematic we
  were down to the bone anyway!
• Price caps free up funds for more e-resources
  (difference between full and capped prices over
  several years)
• Which Big Deals to go for?
• Used ILL statistics for most requested publishers
  (not titles)
• Drew up a prioritised hit list
• Timed perfectly to feed into NESLi2 survey of
  publishers

18
Implications on Journal Purchasing

• Planning ahead to take a Big Deal next year
• Cooperation / collaboration to ensure
• No cancellations of that publishers titles
• No new subscriptions to titles in that Big Deal
• More holistic management of journals portfolio
• Look out for multiple new subscriptions requests
  from the same publisher
• Consider a Big Deal instead
• If so, renegotiate any planned cancellations of
  that publishers titles
• Much more interaction between Subject Librarians
  and Electronic Resources Manager

19
Institutional Context 2002 onwards

• University
• New Vice-Chancellor, Drummond Bone
• Strategy to improve league tables position
• Library Computing spend per FTE in particular
• Commitment to increase Library budget
  significantly for several successive years
• Library
• New Librarian, Phil Sykes
• Feeds broad usage statistics to the committee
  that sets our budget
• Able to prove that e-resources provide excellent
  value for money and improved user satisfaction
• Successful at bidding for development funds

20
E-Resources Portfolio in 2007

• Over 17,000 e-journals
• Subscribe to every NESLi2 agreement
• 27 e-journals Big Deals in all
• Over 280 databases
• Over 30,000 E-Books
• We think this is great, but it means

21
Usage Statistics Overload

• We obtain/receive usage statistics from
• 43 e-journals sources (11 in 2001)
• 35 database sources (10 in 2001)
• 8 e-books sources (0 in 2001)
• no CD-ROMs sources (stopped bothering in 2004)
• We want our usage statistics to serve a purpose
• We cant analyse them to the depth that we used
  to
• Wed rather spend money on content than on usage
  statistics management tools

22
Our Solution

• We compile our usage statistics only twice per
  year
• End of Academic Year
• E-Journals Total full text downloads
• Web databases sessions and searches
• E-books whatever makes sense!
• NOT title-level e-journal statistics at this
  point
• To feed into annual report to funding committee
• To coincide with SCONUL statistics collection

23
Look How Well Were Spending your Money
24
Our Solution

• We compile our usage statistics only twice per
  year
• End of Calendar Year
• E-journals Full-text downloads per title
• COUNTER JR1 report (or equivalent)
• To feed into our annual Journals Review
• Web databases sessions and searches
• E-books whatever makes sense!
• To make a prediction for the end of next academic
  year (to see if we need to promote resources to
  keep looking good!)

25
Annual Journals Review

• Journals budgets are still managed very carefully
• Journal subscriptions met from departmental
  allocations calculated from funding formula
• Information about subscriptions (cost, fund code,
  format etc) managed in an Access database
• Subject librarians generate reports and annually
  agree any cancellations / new subscriptions /
  changes of format with their departmental reps
• Subject librarians and departmental reps expect
  to see usage data / value information too

26
Usage Statistics in the Annual Review
Download 43 JR1 reports (or similar) into a
single Excel spreadsheet (one tab per report)
Tweak slightly so all sheets are same format
Import into our Access subscriptions database as
an additional Table
Join Tables by their ISSN fields
Create Query Type in Fund Code, get financial,
admin and usage data (inc. cost per download)
27
Advice to Subject Librarians

• Low usage might be due to
• The subject area concerned
• Statistics from a site where we dont have access
• Journal only recently became available online
• Few issues available online
• Also have access at another site that doesnt
  provide usage statistics (e.g. PubMed Central)
• ISSN mismatch
• Change in way vendor calculates/supplies
  statistics
• JSTOR, EBSCO, ProQuest excluded (but available in
  another column) not related to subscription

28
Is This Realistic in Practice?

• All usage stats URLs and logins kept in HTML
  files (available to subject librarians if they
  want them)
• It really doesnt take long to download 43
  reports
• Good proportion are true COUNTER R2 JR1 reports
• A matter of a few minutes each to change others
• Importing to Access is quick and easy
• Access query replaces any blank pISSN with eISSN
• Matches what we do in subscriptions table
• Subject librarians run same query as before, but
  more checking, if they want to
• A large proportion of our titles are not
  cancellable

29
Usage Statistics in Trials

• Surprised how rarely vendors send us usage
  statistics at the end of a trial
• But do trial usage statistics tell us much?
• Not realistic to compare usage of trial resources
  with subscribed resources
• For a realistic comparison we need to integrate
  trial resources into a librarys e-resources
  environment
• Trials are usually too short to bother doing this
• Scopus recognised that very long term trial was
  the way to break a completely new product to the
  market
• Vendors are you more interested in protecting
  your IPR or in making a sale? Give us more than
  30 days!

30
E-Journal Trials at Liverpool

• Try to secure a trial of 6 months (IOS Press,
  Bentham) or at least 2 months (Palgrave, Thieme)
• Spreadsheet of Titles, ISSNs, URLs, first
  available Year and Volume (from publisher if
  possible)
• Turn into MARC records with MARCMaker
• Use to update link resolver
• MARC records promote trial and encourage comments
• OpenURL means researchers will be pointed to
  these journals in their normal course of research
  reliable, comparable usage cf subscribed titles

31
Trial E-Journal Catalogue Record
32
Trial Comments Collated in Blog
33
OpenURL to Drive Usage (Statistics)
34
Thank You For Listening

• If You Have Been