Electronic Resource Metrics: Appropriate cost-per-search?

We recently finished a good healthy weed and evaluation of our print reference collection, and I promised a similar evaluation of our electronic resources at the close of the print ref project.  We do keep track of database use, with vendor statistics, some more reliable than others.  We've kept these stats for the past few years, so we have a relative idea of raw use--sessions or pages or log-ins per month.  Then we divide cost by use, and come up with a per-use/search/page view cost.  Year-end price-per-use ranges from a few pennies per search to a cost-per-search that comes out to about the price of a new hardback  before discount).   

What we're looking for now, are guidelines for what an appropriate cost-per-search is.  So far, all I've been able to do is compare our results with the few annual reports for other libraries that I've found on line.  I've also queried lots of my online colleagues, most of who are academic librarians. High cost-per-search is an accepted part of academic libraries' budgets, when a specific database, regardless of how little used, is essential to support the work of a particular program.  Someone said $5/search was high and $10, astronomical.  Someone else said $10 was not so bad. 

The few responses I've gotten, while appreciated, have not really given me a solid answer.  When I posted to wispublib, mostly I heard from other folks who would like to know what I learn, including our largest public libraries in the state.

Someone pointed me to this E-Metrics Instructional System (EMIS)  which has a bunch of tools about how to get data and a way to generate reports with it, but I didn't find the specific bit of information I was looking for.  (EMIS is an IMLS-funded project developed by Florida State University.)   The same person also pointed me to OPLIN (Ohio Public Library Information Network). I've browsed the site a bit, but nothing has jumped out.

So, I come to you, Raccoon readers, to mine you for your endless and collective cleverness and wisdom.  Your assistance is greatly appreciated!

Comments

[Ryan]

I was going to say a comparison against ILL costs would be a good scenario.

For instance, York University charges $12.00 CDN for non-profits that are not part of Canadian Academic Library Associations/Consortia. $5.00 if they are.

That tells me that a $10.00 per search is probably a bit steep unless it is an essential resource -- but neither is it gastronomical.

[Bill]

It seems to me, that this is like looking at Journal Citation Index.
You need to group the databases, and compare similar (or identical) ones.
If you have Gale and Ebsco and Modern Language Association literature databases, a cost per search evaluation makes sense.
Something exotic (Physics!) may entail a high per search cost, because it is used by a small number of students. If it is required for accreditation, you may be required to keep it, regardless of cost.
Full-text vs. citation MIGHT be less expensive than ILL costs, in a low use database. My $.02 (Your mileage may vary.)

[Jeff]

Everybody seems to hide their stats. I would just see how each database compares. If one cost per click is higher than another, I would examine the database as to why it is not used.

I talk about this in a post:
http://gathernodust.blogspot.com/2007/08/my-first-slideshareslidecast.html

Right now we are at 25 cents per click versus 81 cents last year. We were able to convince our County consortium and the state library to pick up some of the databases because of their success. Driving our cost way down so that we can buy into overdrive.

Hope this helps. I have an excel file if you are interested.

[rochelle]

Thanks, Jenica! You people are so sensible.

[Jenica]

We're all so DIFFERENT, in size, users, budgets, needs, collections... how could we pick one number for us all and declare it to be "right"?

And so my answer is thus: Figure out what you think your best-used, most appropriate database is. Figure out what its cost-per-use is. And then declare that to be "good" for your library, and compare the rest to it.