One mandate of good information architecture is finding concrete, measurable answers. So, my objectivity radar goes off when I see people trying to slip questions about aesthetics or subjective preferences into a round of Web site usability testing. It’s against best practice, but it’s not as uncommon as you might think.
Companies want (and sometimes need) user feedback to validate visual design decisions. As a result, they often resort to a seemingly straightforward approach – simply asking usability participants which design they prefer and why.
But if you’ve ever tried it, you know that the answer isn’t always as easy as asking the question. In fact, asking the question often results in contradictory findings, confusing recommendations…and, ultimately, lots and lots of noise. Many users don’t know why a design makes them feel a certain way, and those that do often struggle to articulate it. The few who can provide reasons often introduce feedback too subjective to wager a business decision on.
A recent article at UXmatters.com addresses the practice of asking usability participants which visual design option they prefer. The author shares my belief that such feedback creates more noise than it does clarity — but he proposes an interesting alternative and offers a case study on how the method played out at his company.
The approach they followed was based on a white paper by Joey Benedek and Trish Miner of Microsoft. Summarized, the process would be something like this:
- In the creative brief, identify the top three to five descriptors that the creative design would ideally invoke.
- After information design is complete, prepare multiple creative concepts that include Web page comps with the same layout and content but differing aesthetics.
- Arrange one group of test participants for each visual concept you intend to test.
- Develop a set of adjectives that participants might use to describe the site. Be sure that these adjectives are words your participants will understand, are salient to your research, and include a mix of descriptions that people might consider positive (around 60%) and negative (around 40%). Microsoft used nearly 120 descriptors in their testing.
- Ask each participant to view a comp, then select the three to five of these adjectives they think best described it.
- Ask them why they selected each of the words they did.
- Track the responses.
I’ve often been faced with this research dilemma, but I have never found the solution– so I’m eager to try this method. Have you ever tried a test technique similar to this one? Tell us about your experience in the comments.


