Mixed Level Designs

[RaagaMudra]

Dear All

I am planning to use DOE in web analytics. As a start we are working
on optimizing the design of a pop-up window. Typical experiments would
involve varying the levels of different components of the pop-up -
size,layout, scroll direction etc. In such scenario mixed level design
becomes unavoidable since we are purely dealing with categorical
variables.Can you suggest me a book or a software or an algorithm that
can help me in designing mixed level experiments with least number of
runs?.

Also I would like to know the similarity between taguchi OA and
Resolution III fraction factorial designs?

Thanks
Mathangi