Thank you to all --there were lots!--who replied. I've copied the comments here so that everyone can take advantage of this wonderful and generous resource. Happy weekend to all!

==============   Preparing for a Slimmer Bottom Line: analyzing and reducing a serials collection using a mixed methods approach.

Cannot add anything more "authoritative," but one aspect I don't see addressed here that has been relevant to some of my own decision-making is what portion of your measured usage is for your own users vs what you send to requesting libraries on ILL.  I have on occasion cancelled an expensive journal with decent usage, only to find that 75% of the usage was actually to fill other libraries' requests for it!  My hard-won dollars should go primarily to support my own folks, so an imbalance like that needs to be taken into consideration.  Overall, my journal use runs about 4-5% for ILLs (and I am a net lender!).  Just my 2 cents.
Our cost-per-use cut-off has been $50 for the last few years.   We arrived at this cut-off by a very un-scientific method.  We had to cut our budget back to a certain substantial amount, so we started subtracting the prices of our highest cost-per-use journals until we had our expenses down to the required point.  We just happened to land on $50.  Nevertheless, using this figure has resulted in the cancellation of other journals since the original budget cut.  
As a secondary measure, we have also looked at what we would have to pay for articles if we drop a particular journal, including copyright fees.  However, I have observed that dropping a journal often doesn't bring a corresponding rise in article requests that have to be filled via ILL.
I don't have anything "authoritative".  
Just some things to consider.

You could try to come up for a "cost per ill"

Figure that an ILL costs around $11-$15
The amount of time you spend on the request in total... perhaps 15-20 min ?  Divide by an hourly wage.
Look up the ILL CCC cost for ILL for that journal article (per page or per article) - at After the first 5 per year.

That will probably give you around $20 for the first 5 articles and around $60 for other articles after that.

I always compare the usage stats from the vendor with any local stats I have...

ILL Borrowing
ILL Lending
Local AtoZ list ClickThrough counts
Anything that might give a trend or indication that there is interest in the Title.

I've used Impact Factor - divided by the cost  --- $/impact

The general idea is to get a big picture and build a case...
We use all of the usual metrics (cost per use, ILL requests, etc.) but we also have a Library Advisory Committee to help us gather feedback about possible cancellations. They are advisory in nature and don't have the authority to direct us, thankfully, and so far this has worked out well and they've proved to be quite reasonable and helpful. 

It's been great for added input/justification, because we're in a position where some of our programs are much smaller than others, making it hard to go by strict number guidelines since they won't have as many hits on their resources. I realize we could go to a pay-on-demand model for some of these journals where the cost-per-use is a bit high but we use it too much to rely on ILL to fill our needs without violating copyright, but we just don't have the staff right now to successfully manage that. Plus, smaller programs still need to have instant access to their core resources. I guess just my two cents for some ways to get input beyond the numbers. 
I looked at usage statistic and measure that against the subscription price for a cost per download and then look at how much it would  cost for me to purchase an article from that title. Different publishers/titles charges different pay per use.
I'm afraid I use a very  unsophisticated way to aid in decision making about cancellations:

If the cost per use over the last 12 months is more than $11-$12, it goes on the  list as a potential cancellation. I don't waste time figuring out how much it cost in staff time, etc. If it is costing me more than an ILL would per use, it might be headed for the chopping block. 

I'm actually more generous  in the cost per  use than my predecessor - her guide was - if it  cost more than an extra-large latte, it was pretty much history.

Of course, I take into consideration whether it is the only title in the subject area. But that's about it.

When I receive requests for adding new titles, and they are affordable, we take them on for 1 year - and the requestor is bluntly  told that if the  usage does not justify the cost, they will be dropped. In the tight budgetary environment at our institution, everyone seems to understand. If they are expensive  and there is no ILL request history to back up the request, it's a no go.

We look at our ILL borrows each year and pay special attention to any title we requested more than 5 times - but the heavy hitters turn out to cost thousands of dollars for a subscription, so we never end up meeting the threshold where it would be more cost  efficient to have the subscription. 
I think the figure in #3 is way low at this point, probably outdated. The eJournal calculator (#4) may have been updated (in one place, but inadvertently not both) to reflect the increased cost of Copyright Fees over the last few years. It's definitely closer to $45, if not more. It's a lot of work, but the number is probably different for every organization, depending on the mix of things you use.

I look at cost per use, but also evaluate it against cost of paying copyright fees if we borrow in excess of the CONTU guidelines. You can go to CCC website and look at the copyright fees for ILL of various titles to get some sense of what your costs would be.

So let's say we use a particular journal 12 times during the year. If we cancel it and remain at the same level of usage, we'll have to pay copyright fees on 7 of the 12 interlibrary loans. If the copyright fees are $45.00 per article, that's a total of $315 year. So if we're talking about a journal that costs $1500 year, I'll probably not renew it. If we're talking about one that costs $315 or less, I definitely will. If we're talking about $315 - $500, I almost definitely will. Other things I think about to make that determination are the subject matter, and what else we have in the same area, how important it might be to have instant access (clinical journals vs. policy journals, for example), the time &  hassle factor of ILL. Basically what I'm figuring out are the unseen costs in borrowing vs. having on hand.
We don't factor in the cost of our time. We also don't use impact factor ($/impact) in our decision either.  We cut journals that don't average approx $11-$15 per usage.  I say approximately because if it is a new journal or has had historically good usage and just had a flukey year we give it another year to get its usage stats up. We look at how much we loan it out too (if it is an ejournal that can be loaned out) but that doesn't always save it. For us, the decision to keep the journal is our patron usage. If it gets enough usage to offset ILL costs then we keep it.

Of course there are exceptions. I mentioned the flukey year. We will cut a journal if it has significant and continued access problems.  We don't need to bother with a journal that we continually have problems with.
That not only eats up our time which is better spend elsewhere but it impacts the usage stats as well.

We have also cut journals that have had historically high usage with us because the publisher raised the price too high.  Due to budget constraints and increasing journal costs, we don't have any chaff to separate from the wheat. The journal either makes muster or it is cut the following year.  Because we don't hold on to under-performing journals, this leaves us with a narrow tolerance for journals that significantly boost their costs.  There have been occasions when a publisher raised the price of a journal too much, to the point it put it into the same price league as other more highly valued items, we cut it despite its good usage.
for what it's worth, we use the same $30 number.  Basically, we highlight $30+ in yellow (for watch this title) and $100+ in pink for cancel. The $30+ titles we watch cost per use for trends over ~3 years, or to see if it is single/small set of users.

Jo-Anne Aspri, MLS | LIBRARY | Kent Hospital | 455 Toll Gate Road | Warwick, RI 02886  | 401.737.7010 x31309 | jaspri AT


-----Original Message-----
From: Medical Libraries Discussion List [mailto:[log in to unmask]] On Behalf Of Aspri, Jo-Anne
Sent: Tuesday, December 10, 2013 3:35 PM
To: [log in to unmask]
Subject: Usage Criteria for discontinuing E-Journals?

Esteemed Colleagues,

Please pardon the cross-posting.

I am trying to gather various measures that have been used to assist in decision-making regarding discontinuation of individual E-Journal titles. Somewhere in my head I had a >$30/use or <30 uses/year figure but I can't verify how I came up with that. Looking through a few resources, I came across the following:

1.  Shearer BS, Klatt C, Nagy SP. Development of a new academic digital library: a study of usage data of a core medical electronic journal collection. J Med Libr Assoc. 2009 Apr;97(2):93-101. doi: 10.3163/1536-5050.97.2.005. PubMed PMID:
19404499; PubMed Central PMCID: PMC2670205.

"...each library must develop its own set of criteria on which to base selection and retention decisions."(p. 100)

2.  Bloom, M., & Marks, L. (2013). Growing Opportunities in the Hospital Library: Measuring the Collection Needs of Hospital Clinicians. Journal Of Hospital Librarianship, 13(2), 113-119. doi:10.1080/15323269.2013.770658

"Any <journal> title that had been accessed less than 10 times in a year was deemed eligible for cancellation." (p. 116)

3.  Prottsman MF. "Financial Management" In: Bandy, M. M., Dudden, R.F., eds. The Medical Library Association Guide to Managing Health Care Libraries 2nd ed., 2011. p. 65

Fig. 4.1 Interlibrary Loan Cost-Benefit Analysis: "Commercial costs (i.e. for PPV) are estimated at $25.00 each."

4.  Perhaps "the best" is this from except that at that URL, the figure for "E-journal articles accessed" is given as $45, whereas if you click the "E-Journal articles" link it takes you to:



Estimated cost to purchase a single article from an ejournal vendor

Vendors provide use statistics for e-journals. As for print journals decide what that use represents. If it's hits on Tables of Contents, or hits per issue, that could represent, on average, one article read per hit. If the vendor supplies article level stats you can use that. You may need to adjust different vendor stats so the common denominator is articles read

Does anyone have anything more, or more authoritative, to add?

I value your input!
Jo-Anne Aspri, MLS | LIBRARY | Kent Hospital | 455 Toll Gate Road | Warwick, RI 02886  | 401.737.7010 x31309 | jaspri AT

<br><hr><font face='Verdana' color='Gray' size='1'><br>This e-mail and any files transmitted with it are confidential and intended solely for the use of the individual<br>or entity to whom they are addressed. If you are not the intended recipient, you are hereby notified<br>that any disclosure, copying, distribution or taking of any action in reliance on the information contained in<br>this e-mail is prohibited. If you have received this e-mail in error, please notify sender by reply e-mail and<br>delete this message and any attachment(s) immediately. Thank you for your consideration in this matter.<br></font>