wait times in the 24/7 cooperative

Now and then people ask me if they can work with L-net as an intern or volunteer. Usually*, I don't have projects for anyone and I turn them away.

But a few years ago, I did have a project, and Joanna Burgess, then a graduate student at the University of Washington's iSchool, helped us to figure out why patrons either never appeared in the session, or disappeared partway through.

I started her off with the hypothesis that K-12 students were more likely to disappear than other types of patrons, and less likely to ask formal reference questions. While she did find this to be true to some extent, Joanna also suggested that a more likely indicator of whether a session will be successful or not was the length of time a patron waited.

Neither the Tutor.com software we used then, or the OCLC QuestionPoint software we use now, give any indication to the patron how long they will have to wait. I think this is a shame - patrons clearly give up when they wait too long, and maybe they wouldn't bother waiting if they knew how long it would take.

To compensate, I added messages to our about page, which loads when a patron wants to chat, saying that most people got connected in less than one minute.

That's no longer true - the median wait time has jumped up to 1 minute and 44 seconds, so yesterday I updated the about page to tell patrons about how long patrons have been waiting at that hour of the day in recent weeks. Currently, this varies from 8 seconds at 1 am to about two minutes or more, on average, between 5pm and 9pm.

Three weeks ago, the 24/7 cooperative switched to a system that gives "local" librarians the opportunity to pick up "their" patrons before the cooperative or QuestionPoint librarians do.

If the patron's library is online, cooperative and backup librarians wait 40-50 seconds before being able to pick up the patron, regardless of whether the patron's library is busy with other patrons or not.

Backup librarians generally wait 2 minutes before picking up anyone as long as cooperative librarians are available to do the job.

In most cases, the result is that patrons connecting to the cooperative wait longer than patrons connecting to Oregon librarians, and patrons connecting to backup librarians wait the longest of all.

In the last three weeks, patrons connecting to our different queues wait different amounts of time (in seconds):

I don't think these numbers say anything about the quality of service being provided by different groups of librarians to our patrons, just that under the current system, wait times vary drastically depending on who the patron is and what librarians are available to help them.

Ideally, all of our patrons should be getting the same service, and they aren't. I'm also willing to bet that cooperative librarians and backup librarians deal with a higher percentage of disconnecting patrons.

In theory, if we want our patrons to be picked up as soon as possible, having more Oregon librarians online is the way to do it. The exception is that our academic patrons get picked up even faster by the cooperative, and I think that is because there are reliably few Oregon academic librarians online when most academic patrons need help. If an Oregon academic librarian were online, and busy, the patron would wait at least 40 seconds.

The 24/7 system assumes that patrons are best served by their local library and that this is the most important thing. I agree that local service is important, but I think that reducing patron wait time is more so.

I'd like to see cooperative virtual reference models move to more of a free-for-all, and especially when all the librarians are busy and everyone has to wait. But given the system we have to work with, should we spread our librarians out on the schedule, or clump them more densely together?

* One last note - I may have another project for interns or volunteers now.