We've been tracking technical problems in QuestionPoint chat for a little over a year and a half, but until recently, I didn't have a sense of the extent of technical problems in e-mail questions.
We get relatively few e-mail questions, about 100 per month, as opposed to 500-3,500 chat sessions, but every day there is at least one notification in my inbox that an e-mail to a patron has bounced. This means that the patron will probably never see what the librarian sends.
An e-mail address may bounce for a few reasons:
1. The patron entered a typo or doesn't fully know their e-mail address. This happens with school networks especially - on the internet a student's address may be student@school.district.k12.or.us but at school they are just 'student', and acronyms for the school and district seem like they may be hard to keep track of (try saying iwms.wlwv.k12.or.us three times fast).
2. To protect their privacy, the patron may enter a fake e-mail address. We don't label the e-mail address as a required field, but people enter fake information anyway.
3. The patron's e-mail server thinks our e-mail is spam. In the last case, we are lucky if we get bounced mail. Telling spammers the mail bounced doesn't make them go away, so it's a waste of bandwidth to reply to a spammer that someone doesn't exist. A server may think we're spam because it can tell that a server at oclc.org is sending e-mail that purports to be from oregonlibraries.net.
In October 2007, I tracked each bounced message and kept track of how many questions had e-mail addresses that bounced, and I received east 82. I say 'at least' because not every e-mail sent to a bad address bounces, and because bounced mail could may have been caught in my own mail server's e-mail firewall and spam filters.
Of the 82, 77 were chat transcripts and 5 were e-mail transcripts. During that month, we had 2,122 chat sessions, but patrons only gave us e-mail addresses 751 times. We had 120 e-mail questions, so math tells us that 10% of chat sessions and 4% of e-mail questions had bad e-mail addresses. This comes to 9% overall (82/871).
One of the things I am trying to decide is if we should ask for an e-mail address for chat sessions at all. For the same month, 56% of academic queue patrons, 40% of public library queue patrons and 29% of K-12 queue patrons entered an e-mail address.
This is enough to make me want to keep asking for e-mail addresses, but 9% is still an uncomfortably high error rate.
To address the problem, I'd like to make it more clear to patrons that the e-mail address is optional. It might look something like this:
Your e-mail address:
Second, one of the real problems here is that librarians never get notified that the e-mail to a patron has bounced. Ideally, a virtual reference system that sends people e-mail should also receive the bounces, and make notes in the record when it happens.
This is a bug and a feature of QuestionPoint. The software does not handle bounces, but part of the reason for that is that us users have prioritized appearances over functionality. In order to make it appear that the answer/transcript is coming from the local library, libraries have the option of overriding the questionpoint.org e-mail address that is used to send patrons e-mail. For example, our patrons get transcripts mailed from oregonlibraries.net instead of questionpoint.org or oclc.org. There is no way that QuestionPoint can handle an error if the e-mail bounces directly to the library.
Short of the QuestionPoint team drastically changing their priorities (which I am not going to encourage), this may only be possible if we end up switching software.
The only other thing really to do about this is to obsessive-compulsively add notes to questions in QP where the e-mail bounces. Does anyone want to volunteer?

Tweaking that idea
I have a suggestion for the email issue. Don't ask it on intake. Ask for it at the end of a session ("provide your email address to receive a transcript of this discussion"). I believe most commercial products work this way.
Re: Tweaking that idea
QuestionPoint does provide this, in a form. The patron can enter an e-mail address at the end of the session to have the transcript e-mailed.
I have a hankering for simpler entry forms, so I had been leaning towards this solution, but there are a few arguments against it:
Given the e-mail address up front, librarians can follow-up on an unsatisfactory answer. We question whether this is an invasion of privacy or not. We get both positive and negative feedback on this topic, though my sense is that most of it is positive.
Second, if the patron browses away, hits a bookmark, clicks 'X' on the window or if there is a technical problem, they never see the screen that prompts for the e-mail address. Happily, OCLC adds a message, so I know it happens in 25% of sessions.
Maybe, patrons would click 'exit' more often if they knew it was their only chance to get the transcript, but I don't believe our system should be so opaque.
Post new comment