Here's the situation. My larger place of work's electronic resources folks are processing renewals when one digital library comes back and says: your normal increase would be 3% but since your usage increased 24% since last year, it's a 6% increase. Ladies and gentlemen: we do not have an extra 3% to give them, nor did we agree to a usage based subscription fee (we think).
First thought was that the pricing model is AFU and what on earth actual publisher wants more money for more usage?
Second thought was whoa - 24% increase in one year is weird for something that is basically the same and a topic of perennial interest? Like not CRISPR-R-Us or all-AI-all-the-time or anything super hot... just a topic that all of the sci/tech divisions of the larger institution are interested in to some extent. No new programs in the area and no dramatic changes to the content on the platform. So maybe the logs are messed up?
The electronic resources folks got two bits of usage data for us: a file with all the IPs that have downloaded articles 2016-2018 and a file with article identification info and date-time stamp (to the minute only) of when it was downloaded 2016-2018.
First thing to check - are these all our IPs (we've gotten usage logs of other universities in the past from other vendors)? Yes. And interestingly, it divides up 1/3, 1/3, 1/3 between the 3 major sci/tech divisions of my larger institution.
Second thing - are they double counting? We can see by visual examination that there are 2 entries with the same time stamp for a single article. There are also a bunch that are 1 minute off. Question: is the double counting consistent across years or does it happen significantly more often in 2018?
So I imported the log into R, then counted up all the records that had the same article name identifier as the one below them and either the same time stamp or 1 minute off. Well bummer. Turns out the double counting was worse in 2016, the year with the lowest usage. (FWIW, we had a 2% increase in usage from 2016-2017 and the data show an 18% increase not a 24% increase from 2017-2018)
Third thing - is this one of the publishers that's weird about reserves? I'm not at all fluent in that since I don't work with students, but we can see articles with 230 downloads all clustering around the beginnings of semesters (August-September, January). So... ?
Another thought: we tell people to just send around the link and everyone should download their own copy. Sounds like a bad idea for this publisher!
This isn't a done deal but I'm not sure beyond looking at the license what we can do. We're definitely going to push back. Wish us luck!