Low usage?

We’ve had quite a few questions in recent months from libraries and publishers who’ve noted a decrease in usage. The first question we always ask is, of course, whether they’re looking at the right report. Using a TR_J1 is going to exclude all open access and free materials, so the TR is a much better report for understanding real usage!

The second question we ask is about AI. It’s early days, but we’ve seen indications that libraries that have licensed AI tools are seeing bigger drops in usage than those which haven’t. AI tools can be a fantastic support, but reading a generated summary of research often means that students and faculty members don’t go on to check out the source material.

Even where AI tools are embedded within a publisher site, unless the user clicks through to the source materials it’ll only count as search (i.e. no investigations or requests). The rationale behind our current AI consultation is that libraries and publishers want to have visibility of how much the AI tools are drawing on published materials – hence our proposal for Access_Method: Agent tracking usage by generative and agentic tools. 

Then of course you have the issue of third-party tools, whether that’s one intended for research (e.g. Consensus, Perplexity, Scite) or a generic LLM (e.g. ChatGPT, Claude). Even where the library has licensed content to allow their faculty and students to draw on publisher platform within a third-party tool, that kind of off-platform usage isn’t going to show up in the publishers’ COUNTER reports.

So if you’re seeing drops in your COUNTER usage, consider how widespread AI adoption is in your institution.