UIE Roadshow : Day 2
Today’s seminar was “Using Personas to Guide Design” given by [Kim Goodwin](http://cooper.com/content/company/executives.asp) of [Cooper](http://cooper.com). Kim focussed on Persona creation but also touched on the overall design process including user research and UI specification creation. (I’m summarizing the summarization of the Cooper design process here, so I may not get all the details right. I hope someone reads this and will correct me if I’m way off.)
Compared to [yesterday's seminar](http://jay-photo.com/blog/index.php?p=129), Kim suggested a lighter weight interview and observation process. For example, Kate and Ellen suggested video taping site visits was a very useful thing to do whereas Kim advised us to avoid video altogether. Kim also suggested a less scripted interview style. However, the goals are the same: discover the user’s tasks, attitudes, and frustrations and, most importantly, their true (not reported) behaviour.
From there, the Cooper design process puts a lot of emphasis on creating personas, which are the major deliverable summarizing the user research findings. Many of the deliverables discussed yesterday are either collapsed into personas or not produced at all. Part of the reason Cooper can get away with minimal documentation of the findings is because they ensure that the design team (an interaction designer and a “design communicator”) conduct all of the interviews. That way, they have the entire data set in their heads. Personas help communicate with stakeholders and are useful reference points for design, but they don’t appear to be knowledge transfer tools by themselves.
As I wrote, though, the seminar was about creating [personas](http://cooper.com/content/why_cooper/powerful_personas.asp). Personas are user archetypes. They amalgamate observation data from multiple interviews into a fictional user description. Each persona has a name, picture, and narrative description of their current work, attitudes, and goals. They also have some personal details to add some life to the description, but these are kept to a minimum when not relevant to the context of the product domain.
To create personas, you analyze your research data and determine what they call “behavioural variables”. Variables can be measured along a range (qualitative end points) or as multiple choice buckets. Kim emphasized that you want to find behaviours and ignore the mechanisms being used. For example, finding a photograph is a behaviour but it doesn’t matter if the user searched a shoe box or a software application. For each role you’re interested in, you map out each interviewee’s behaviours onto these variables and look for patterns where multiple people’s behaviours cluster together. You then create your personas by combining the interviewee’s behaviours into a single archetype.
There is more to it, of course, but the central point is that behaviours drive the creation process. We don’t care about the technology they use — we only care about *what* they do and *why* they do it.
This seminar was excellent because it got me thinking about the overall design process. Creo is quite far away from what Cooper would consider ideal and I wonder if we can (or should) change things. I also wonder how useful personas are if they are not created by the actual GUI designers.
More thoughts later as this all sinks in.