Question

Does the idea of UI Driven Development make sense at all? Most of our clients like to convey their requirements in form of screens. For e.g. I want a screen to do THIS and THAT. Sometimes they even go as far as to dictate the layout of a screen by themselves (this may be because clients of today use software applications for most of their tasks already).

Also this method of requirement gathering seems to convey both the Data and Associated Behavior automatically.

What do you guys think?

Answer 1

Actually it does make sense in this regard. Use cases differ here in that they cannot be used in the same way. Use cases will be usefull to extract and formalize requirements.

I was workinng in a UI lab a while back and we toyed with this concept, though we did not call it as such. Basic idea here is that we would use agile iterative approach to development where we would use usability tesing to help converge on a desired solution.

Typical cycle would be :

• draft a few requirements and use cases, very limited in scope very focused.
• Create a test protocol that would allow us to gather data on this feature (usability, ease of use, acceptance, performace etc).
• Create or expand the application to include this feature
• have a few users test the application using agile adaptation of usability testing (see Ruben’s Handbook of Usability Testing) where we would limit test sessions to 15 minutes with a 15 minutes debreifing.
• Extract usefull data from testing to inject in future iteration.

This method was perticularely usefull when the users would either not know exactly what they wanted or could not tell us. We would therefore have to devise tests that would gather objective data on the actuall usefulness of the software to the user and try to ajust the next iteration consequently. (many of our ‘customers’ if I may call them so were heavily handicapped and could not talk so we had to be creative).

This way of working forced us to have a GUI available to the client very early in the software’s development life and because it was the central point where we would direct testing one could say it was GUI driven desing since the driving force for convergence was the user interactions with the system.

Although we developped this technique mostly for very specific cases we did some testing with normal users on a normal software and got very positive results. The desing would converge very quickly towards the usre’s needs. Also the fact that they participated in this desing approach had also very positive effects on the acceptance of the product in the target community.

Unfortunately internal strifes had the lab dismanteled before we could publish our results and expand this line of research, a shame really.

So UIDD (if you pardon the bad acronym) would be a member of the TDD family of approaches towards software development where the iterations hinges around user interactions.

---

Additionnal reading on the subject :

http://www.codinghorror.com/blog/archives/001091.html

http://cakebaker.42dh.com/2007/07/07/usability-driven-development/

http://www.springerlink.com/content/l413k76812896gnt/

http://www.agilemodeling.com/essays/agileUsability.htm

http://www.uxbooth.com/blog/how-test-driven-development-increases-overall-usability/