Labels are a vital but often glossed-over element of successful information architecture. Designers (or worse yet, Department Managers) come up with their own labels and stick with them. A University of Minnesota grad student shows us why we’re missing out.
There’s a big gap between staring at the results of a card sort and launching a new and improved information architecture. Fortunately there are several tools and techniques to help you translate your results into usable navigation. But even experienced designers sometimes skip two other important gaps between a card sort and the user experience: what content to include, and what labels to display. Both of these questions need to be addressed before you run a card sort.
The earliest working prototype of WebSort (circa 2001) required participants to complete two distinct steps. First, they were shown long descriptions of items (which represented a feature of an application, or the content found on a page) and asked to provide a short label for each item. Then they continued on to a card sort of those items, which now showed the labels they provided. We thought this was fantastic.
Turns out, not many other people did.
As development of WebSort progressed, the whole feature was scrapped. This summer when I chatted with our long-time clients and card sorting experts at UPA, I was reminded of the importance of content labeling. Two months later, Jakob wrote about a similar issue of avoiding bias of terminology matching. Finally, Josh Carroll contacted us about a labeling study he was planning, and we were happy to provide a bit of technical support and consultation.
Below are links to a short case study of Josh’s experience, as well as a much longer paper. Let me just highlight one of the unexpected gems:
Some … responses were long, rambling narratives that would never be used on a website. However, the lengthier labels provided more insight into participants’ vocabulary and context.
Just as the organization of items in a card sort does not equal a perfect final web site architecture, the labels provided by users don’t need to be perfect usable link text; the real value is in getting a better understanding of how your users think about your content.
You can download the 5-page and 25-page versions of Josh’s report below. Josh is a Senior Usability Consultant at the University of Minnesota. You can contact him at carr0234 at umn dot edu.
