Workshop on portal usability

The heterogeneous world of usability

The second workshop on usability at HDM dealt with portal issues and was supposedly quite a success. At least that's what the participants said. Why do I hesitate? The talks where all informative and to the point and gave me something to ponder over (more about it below). But what really fascinated me (and also came as a shock to me as I have to admit) was the fact that the participants where such a heterogeneous bunch of people basically coming from two totally different areas. One group where social engineers, business specialists and psychologists - the other one had computer science engineers, electronic engineers and worse. How do you meet the expectations of such an audience?

Surprisingly the audience had no problem at all with such different backgrounds. When I suggested that the next workshop would be much smaller and more specialized they rejected the idea of creating different interest groups depending on the different backgrounds and said that this was exactly what hat hindered usability for more than twenty years. Sounds good but another reason might have been that usability is still a vague and unknown topic and there is no clear professional picture of it yet.

This is about to change. Here at HDM we have a professor who specializes in usability - Michael Burmester - and it was a pleasure to run this workshop with him. We have some exciting plans for the future which I will mention later on. And the response from the industry to our workshop was enormous: we had more than 80 participants from many major companies and specialized portal companies attending the workshop. None of us had expected this. But the message was clear: in times of scarce budgets usability sells: Companies understand that they don't sell products if their interfaces lack in usability.

But now to the workshop and its results - some of them quite surprising at least to me. Modern portal usability is nowadays tracked through professional statistical methods. This was explained through the example of http://www.expedia.com. The site evolved considerably through the last 5 years and every change to the user interface is tested for its results using e.g. A/B tests. Usability experts check the site on a regular base and no change goes without them being involved. Obviously the company behind this portal links usability directly to success - expressed e.g. in flights sold. Another very interesting point here was the mixture of design issues (attractivity), marketing (sale offers) and usability (few elements on screen). Sometimes it is probably quite hard to distinguish the effects. The major result for me was that now portal development means not only to integrate design with development but that usability issues are an independent and equally important integration factor. If I think back to how hard it usually was just to turn a nice design (why is it always first in a project?) into something that can actually be developed and will run in a performant way) I shudder when another complex problem enters this area and needs integration.

A speech portal was shown and this was probably the funniest part of the whole workshop: people where shown on video how they attempted to book a trip on a travel portal or make a call through some telcom speech portal. The videos showed how tedious most of these portals really where. According to the specialist from Fraunhofer IAO the Quelle speech portal seems to be quite good though.

While I avoid to use speech portals (like most people I seem to hate the humiliation..) I learnt a lot from the talk. Because the speech portal is such an extreme form of portal it makes navigational and semantic problems much more visible. But the biggest shock was that the really bad speech portals don't look so bad without the videos. As a software programmer I could almost find myself in the way those speech based user interfaces where designed: they where basically only a direct mapping of the programmers categories (software designers would probably use the term "model" here) to the user interface layer. This is the wrong approach in graphical user interfaces as we all know nowadays because it leads to very annoying and tedious navigation. One can almost see the "if" statements of the business logic showing up in the GUI.

But if the direct mapping approach is wrong - what do we teach our students? Currently all they get is a concept of "MVC" - Model, View and Controller architecture. The better ones notice soon that this is even a recursive design pattern: it repeats itself within view components e.g. in browsers. And it is basically a direct mapping approach between data (model) and presentation (view). Perhaps the solution is the same that a student of mine, Christoph Birkholt - an experienced game designer - used in our design patterns seminar: He discovered narrative patterns for game design, e.g. how new characters are introduced or how a player is prepared to switch to a different level of the game. These are non-technical patterns in the sense that they are not programming patterns. But they are technically in the sense of structuring a game and making it understandable and finally playable. It is those patterns that we need to teach for GUI design as well.

How does usability know-how enter a project? Two students of Prof. Burmester ran a consulting project with the police. They did a number of analysis session (so called focus group analysis) and where able to improve the usability of the system considerably. One thing that was not clear to me was how they distinguished regular requirements analysis and usability aspects during their work.

But I would like to see our more technically oriented students working together with the information designers on software projects. My hope is that the students will pick up the social know-how needed to work successfully in large and complicated projects.