Monday, August 31, 2009

Interesting Neilsen column on card sorting and keyword bias

Card Sorting: Pushing Users Beyond Terminology Matches

I read this, and thought about a usability study/site redesign I have been mapping out in my own mind for a few months now. Why not, before the card sort, perform an initial round of open-ended questioning of your actual (or possible) users? Start by telling your subjects (in as simple, vague a way as possible) about the type of information the site will contain (e.g. this is a site for a academic library, or this is a site for an online retailer that sell shoes, etc). Then ask the users "what type of information, tools, and services would you expect to find on this type of site?" and even "What type of information, tools, and services would you like to find on this type of site, even it seems like that would be impossible"- to make sure the don't limit their expectations unnecessarily at the outset. Then, for the card sorting, create cards with all the keyword/phrases you cull from those initial open-ended questions. Jumble the cards up and ask your card sorters (different people than answered the open-ended questions) to do the normal card sorting routine. Now no terms come directly from the site content authors or the people running the study, but from actual users. Then, ask the card sorters to do the normal sorting and grouping and labeling of groups, and you'd arrive at a less biased end result- or at least a result that's biased by actual user terminology, instead of site designers, content creators, and testers… Does this sound feasible and workable?