I think too often user interface is dictated by what a designer/manager wants instead of what a user needs. Too often when a decision needs to get made on a website there is this conversation:
Person A: I don't like that blue for the button color, I think Red would be better Person B: Well I like it, and I think person C likes it too Person A: I guess it's a consensus, we'll go with blue
Data Should Win
The reality is, their opinions are completely irrelevant. What is relevant is data that demonstrates what the preference would be of the target user (thanks, Gavan). This is a problem that is pretty easy to grasp, but hard for executives to imagine that their opinions are irrelevant. So it is the job for those of us who work building interfaces to use data to convince managers that the data is right.
This is pretty hard for some of these decisions, most of the time you don’t have data on what color your user likes, or where they like their buttons. But what you can do is infer from data you already have (or should have).
Use Analytics
Most anyone who is building or maintaining a website should have some google analytics or equivalent software to measure traffic. I like using Google analytics’ “Site Overlay” feature to collect some data about what people click on, and then we can make some inferences about what they are clicking on and figure out what on our site is working and what isn’t.
Quick Example
I manage a blog for a friend who writes about sex and relationships at www.ritawatson.com. We looked at the analytics data to see which posts where the most engaging and experimented with different topics to see which ones worked. We didn’t come up with much. What I did find when I looked at the overlay data is that the highest traffic link on the homepage was Rita’s bio. From that I concluded that people often came to the site to get info on Rita, not on relationships. We put a picture of Rita on the homepage, and moved a widget with her pic and name to the top of the blog, and enjoyed a nice boost in traffic.
