We use it all the time for both individual journals and for entire databases, including A&I databases (for which we expect a lower cost per use than full text of course).

That said, we need to be careful that we don't open ourselves to accusations of hypocrisy by vendors/publishers. For instance, ACS a couple of years ago wanted to introduce usage into their pricing model for our national consortium (I'm in Canada), and our consortial office vociferously objected (with the support of the library directors). I had a private conversation with one of the sales reps, who challenged me on that point, as I had to admit that we use usage for determining value for cancellation, so how is it unfair for the vendors to use usage for determining value for pricing?  

I don't have a good answer to that and the only answer I've heard here is that it isn't a question of fairness, but just budget reality, that we simply can't handle costs fluctuating wildly from one year to another if researchers suddenly use one product much more or less than normal in any given year.

Melissa Belvadi