Tuesday, April 24, 2007

(Nearly) Effortless Personalization

I want to see if I can pull together something from a couple of strains of thought I've encountered over the past few days.
  • At the pre-Computers in Libraries conference workshop on web design, we looked at the Amazon.com web site, a site that is hardly simple yet still seems to be very effective (how often have I heard ... we should be more like amazon.com?!). The workshop presenter suggested that one of the reasons for the site's success is that, although it is a complex page, most of the complexity is about me--"New For You", "Recommended for You", "Make a Wish", and so forth, based in part on my past activity on the site and the relationships between books that I've purchased, wished for, perused, etc., and other books they sell.
  • Our department (Digital Projects) recently surveyed several students about a redesign of the library's web site. One of the questions we asked was whether there was anything they'd like to see on the library home page that wasn't there now. Two of the four students for whom I was the notetaker specifically mentioned the addition of something about me--either the books I currently have checked out or the journals that I use most often--to the library's home page.
  • Conventional wisdom seems to be that, with a few notable exceptions, people do not take advantage of opportunities to personalize a web page (e.g., "my library") when it requires them to do something to make it happen.
So, putting these things together, I wonder if it would be useful to think about the amazon.com model of (nearly) effortless personalization; that is, the system personalizes the page for me, based on my past activity. Perhaps we could offer an option (not a requirement) to log into the library web site similar to what Amazon does, including an option to "remember me on this computer". Then maybe we could track things like the journals or databases that the person looks at and personalize the library home page back to them. (There could also be opportunities for them to do some thing(s) to even further personalize the page, realizing that few would likely take advantage of that.) Using the single sign-on capability of Webauth or Shibboleth, we may not even need to require an explicit login on the library's web site every time for this to work--there might be a way to detect whether we already know who they are.

There are obviously privacy issues with this approach that need more careful consideration than I have given them here but I'm not sure they are insurmountable. There are more and more services that offer a "remember me on this computer" option and it may be reasonable to think our users are becoming familiar with this capability (e.g., asking to be "remembered" on their personal computers but not on shared or public machines).

There is also the task of figuring out the best technologies to use to accomplish something like this--e.g., an entirely cookie-based approach, a user profile database.

But the first question, of course, is, would this be desirable functionality to offer?

3 comments:

thoscrich said...

Hi Jim,

What features, benefits, and special content would patrons get once our site "remembers" them? Are we just talking about "Materials on Loan" or would there be other personalized content?

It sounds like some students are using DukePASS. If the content and personalization that we offer are as compelling as those offered by DukePass, then I think this would succeed.

What tool would we use to publish/host the Library home page in order to allow for a DukePASS level of personalization?

I think it would be worthwhile to investigate this further.

--Tom

Jim Coble said...

I don't know exactly what kind of personalized content we'd want to offer. The ones that were suggested in the student interviews were books checked out (or info about one's account, more generally) and journals (/databases) one uses most often. In DPD, we've sometimes talked about "people who checked out this book [i.e., one you've checked out] also checked out ...". That's another possibility.

One university has an RSS feed for new books in each LC classification. That could be an option ... either based on patron sign-up or on the LC classes they check out most often.

I wasn't thinking so much about giving them direct control over the layout of the library page ... more like amazon.com where the site determines the layout and just populates the layout they choose with your personalized content.

--Jim

Paolo said...

This sounds great Jim. We'd need to have a way for people to opt out of it (for privacy), and we could always show some genericized information in those cases.

Another approach would be to recognize that most people don't use the library site as a portal, and to build widgets with library information that can be plugged into places they do use (DukePass, Google, Facebook, etc.) or even as plugins to common applications (a Duke Libraries sidebar or context-sensitive add-on for browsers, a widget for the Mac Dashboard or upcoming Vista equivalent, or the current Yahoo Widget system).