how many use saved records feature?

I was interested in how many users of our existing interfaces used a ‘saved records’ or ‘bookmark’  feature, to inform how much development time should be put into such for our Blacklight application.

In particular, I was interested in how many records users that save records save; because  unless your saving more than 20 or 30, a ‘tagging’ feature or a ‘search within my saved records’ feature isn’t all that useful.

So I went about trying to find out by looking directly at the underying rdbms for two systems:

1) HIP, our legacy in-production Horizon OPAC, which offers a ‘my list’ feature.  This feature works in some odd ways I don’t entirely understand, and additionally I had to reverse engineer the Firebird rdbms schema and may not entirley be understanding the data, but it seems to be decent. (Thanks RazorSQL for connecting to firebird, woo!).  The My List feature is in some ways kind of hard to use, but I know at least some people use it.

2) Xerxes, the front-end for Metalib used for federated searching of mostly article databases.  The saved record interface is pretty polished, although it’s saving a different sort of thing from HIP, so not entirely comparable.

In both cases, I’m not actually sure how big the total number of possible users is, this is harder to find out in our enterprise environment than you might think, so I can’t express number of users that have used the feature as a percentage of all users.


3611 users currently have a ‘My List’ existing. Actually a pretty decent number. [Some of these may be old lists for users no longer active, not easy to say. The oldest list in the system was created in 2003].

Oddly, around 100 of them don’t actually have any items in their ‘My List’, probably had them at one time (causing ‘My List’ to be lazily created), but deleted them.

  • Only 445 of that 3611 (12%) have more than 10 items saved.
  • 232 of that 3611 (6%) have more than 20 items saved.  (yes, this group is included in the previous group, etc., for all these numbers)
  • 152 of that 3611 (4%) have more than 30 items saved.
  • 83 (2%) have more than 50 items saved.
  • 16 (0.5%!) have more than 100 items saved.

(Updated, 10 minutes later). While HIP has no tagging feature, it DOES theoretically allow the user to create more than one ‘my list’, which could be a similar way of dealing with lots of records. However, this feature is VERY confusing in HIP; not only can I never remember how to use it myself, but I think in some circumstances you can accidentally create a second ‘My List’ without meaning to (I have actually done that before), and in some cases your first list may ‘expire’ forcing you to create a second ‘my list’, so I’m not entirely sure what the number of users with more than one ‘my list’ tells us.

However, there are 457 users with more than one ‘my list’. 12% of all users who have used ‘my list’ at all.

Xerxes (Metalib, Federated Search)

3787 users have at some time saved records associated with their account. Interestingly this is about the same number as in HIP. (On the one hand, Xerxes saved record feature is a lot easier to use; on the other hand I’m not sure if total usage of HIP is greater than Xerxes or not).

[Note to Xerxes developers — remember to filter out username like ‘local%’ from such queries to avoid counting temporary users which may really be one of your logged in users not yet logged in, just saving records to the session].

  • 1051 of those 3787 (27%) have more than 10 saved records.
  • 591 of those 3787 (15%) have more than 20 saved records. (again, yes, this group is included in the previous group, etc as we go)
  • 386 of those 3787 (10%) have more than 30 saved records.
  • 205 of them (5%) have more than 50 saved records.
  • 50 (1%) have more than 100 saved records.

So interestingly, Xerxes users save somewhat more records than HIP users, although still a distinct minority save very many.

Xerxes already allows the user of tagging. Could that be encouraging more record saving? Not sure, it doesn’t seem like you NEED tags just to save 10-20 records, and even 10-20 is significantly higher in Xerxes than HIP. I think it’s just Xerxes better interface that encourages a bit more saving, but still mostly moderate.

1855 users have applied at least one tag in Xerxes. 48% of all users who have saved records, somewhat surprising. Again, Xerxes interface makes it very easy and quick to add tags. (Nice job David Walker).

I’m not sure that neccesarily means tags are useful; many of those users have fewer than 10- saved records, why do they need tags to distinguish them? Tagging may just be a habit from other applications, or because it’s so easy to do they just tried it out, or perhaps they erroneously believe that their tags will be shared publically in a ‘social’ way.  Or they don’t realize what they are doing when using the tagging interface. Or they have a use for tagging even with fewer than 10 saved records, I just don’t know what it is!

  • 1199 of those 1855 tagging users have saved more than one tag. (64% of tagging users, and 31% of all users that have any saved records)
  • only 637 of those 1855  tagging users have saved more than 3 tags. (34% of tagging users, 16% of all users who have a saved record)
  • An actually kind of astonishing 391 users have saved more than 10 tags.  (21% of tagging users, 10% of all users who have a saved record.)


In Xerxes, which supports tagging through a very nice interface, an actually surprisingly high minority use the tagging feature.

However, in both HIP and Xerxes, not very many users save more than a couple dozen records — only a distinct minority.  The existence of tagging in Xerxes does not lead to a significant number of users saving lots of records (only 10% of all record-saving users with more than 30, only 5% with more than 50).

It is unclear to me what the purpose of tagging records is if you only have a couple dozen.  My guess is that Xerxes interface makes tagging so easy that people just go ahead and do it even though it doesn’t in fact end up being particularly useful to them. However, it would be interesting to find some of those users and ask them why they tag.

Enough users use ‘saved records’ in both interfaces that I think we DO need a saved records feature in new interfaces, such as our Blacklight implementation.

However, barring more information on why users in Xerxes tags, the numbers of saved records in both HIP and Xerxes lead me to suggest that we should not prioritize features which are really only neccesary (so far as we know) if you have a large number of saved records, such as tagging and searching within your saved record collection.

Some users, although a distinct minority not a vanishingly small one, do save enough records that more complicated organizational features may be neccesary. We should try to serve these users by making sure our systems export as cleanly as possible to existing bibliographic management applications like EndNote, RefWorks or Zotero, that offer more sophisticated organizational features than we are likely to, and which we already need to support good export to for other use cases as well.


4 thoughts on “how many use saved records feature?”

  1. There IS a saved records implementation in Blacklight core, it’s just kind of crappy.

    Sadly while Forward is based on Blacklight, they’ve ‘forked’ enough in their own direction, as I understand it, that making their new feature good to go back in shared Blacklight is non-trivial time for me or for them. May or may not be more efficient than simply starting from ‘scratch’ (ie, what’s already in shared Blacklight core), depending on how neat their features are and how ‘forked’ they are.

    However, regardless of how neat their features are, my tentative suggested conclusion for the moment is that I do should not in fact spend ANY time on supporting tagging or search-within-your-bookmarks features for our users; I should spend my time on other higher priorities instead, and only focus on a very basic straightforward bookmarking implementation, which is mostly already there, although it needs to be cleaned up a bit.

  2. As a reference point, the UMich Mirlyn “Favorites” implementation has 2873 distinct users.

    I can’t give absolute numbers on the “Select a record for later review/email/export-to-refworks/whatever” since they’re not made to log in for that, but about 4.25% of all sessions include someone “selecting” a record, with less than 1% of them eventually doing an export of selected records in some way.

    Of course, like so many little features, some of the folks that use the favorites use the *crap* out if it.

  3. Oh yeah, just because a minority of users use something doesn’t mean we don’t need to support it.

    It’s pretty clear that we need to support both a way to ‘bookmark’ records, AND (this one even MORE important) a way to export records to other systems and formats.

    I’m just not sure we need to support tools for managing large numbers of bookmarks, like tagging and search-inside-my-bookmarks. I think it might make more sense to in fact instead focus on making our “export” functions as AWESOME as possible, so people who need complex organizational tools like that can export to a system that does them better.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s