Usage statistics

When databases come up for renewal, one of the first things we do is look at usage statistics. Well, I have a tax database due for renewal -- the vendor does not provide usage statistics and has no intention of doing so. When I asked, the reply was that the cost was the same regardless of usage and if we wanted to know, they recommend that their corporate customers send out an e-mail to gauge usage. Well, we're a library, so that's not really an option for us. So, do you just blow off usage statistics?

The first professional conference I ever attended was the Charleston Conference. I had been in my first full time job less than six months when I went, but I'm still learning from one of the sessions I attended. It was a session on Counter compliant statistics. They didn't teach that in library school, so it was all new to me. One of the presenters stated that they finally implemented a policy to not subscribe to a database/resource if the usage statistics were not Counter compliant. I didn't quite understand it until I spent a week gathering usage statistics for my library's annual report. Then, I got it.

Now, I'm wondering if I shouldn't have such a policy for databases that provide no usage statistics. The above mentioned database will cost us thousands of dollars and I have no way of gauging if it's being used. My business librarians tell me one facutly member in the business department uses it for one of her classes. We do have a MS in Taxation which adds a little wrinkle to it. However, ProQuest now has an Accounting and Tax database that offers most of what this database does in addition to accounting journals. Their interface is familiar to most and usage stats (Counter compliant ones no less) would be available.

I have no desire to take away a resource that is being used. I also have no desire to pay thousands of dollars for a resource for one class, especially if the information could be provided by another resource that would have broader applicability.

Which begs another question. Is any type of usage statistic better than none?