What are some best practices for displaying lists of items having many attributes, but no natural identifier?

Question

I have a database of full of items with nondescript names like "X0001", "X0002", etc. where each item has ~30 features which are usually strings of ~10 characters, but could be ranges or 50 characters. These features uniquely identify the item and are the things that the end-users already understand and want to be able to see.

In fact, the users would like to see all of them, all the time. However, their monitors aren't large enough and we'd like to avoid horizontal scrolling in our browser application.

So far, we've come up with an unordered list of feature/value pairs displayed horizontally and which wraps. But it's kind of difficult to compare one items feature set with 2 or 3 others, so I'm interested in learning the best practices for displaying this sort of anonymous feature-driven data in a user-friendly way.

Answer 1

Prioritize. While you have 30 different columns, not all of them have equal utility. Nor, likely, do all the columns apply to every type of widget. With these assumptions (which could be wrong), one optimization could be to make the process two steps: first limit the type of widget (sprocket, doohickey, thingy, whatsit), which will notably restrict the number of relevant columns.

Second, once you have filtered the list of relevant columns you can return to horizontal scrolling. It is certainly not optimal, but it is the best for comparing arbitrary columns. Users should rarely need to scroll if you have done a good job at prioritizing the leftmost columns.

To help calibrate the columns, this interface might be helpful:

^{download bmml source – Wireframes created with Balsamiq Mockups}

Users can add whichever columns they need to the right automatically, and you get valuable metrics about which columns they needed to improve which columns you show by default.