Design choices: affected by user feedback? affected by system performances?: lessons learned from the TERENCE project

Iterative design is nowadays indicated as the approach that most likely produces a successful system, and the application of usability evaluation methods is identified as the primary key that allow designers to reveal and fix problems early or, to better say, at the right moment during the design of the system. These approaches often overlooks issues that get revealed only when system "goes live", namely system performances and a different attitude of the users towards the system. With respect to these issues, in this paper we discuss the lessons learned from the TERENCE project -- a technology enhanced learning project for improving the text comprehension in children 7-11 years old -- and in particular from a large-scale evaluation. Our experience suggests modifications to the classical UCD lifecycle: on the one hand post implementation activities should be foreseen since the beginning (and assigned reasonable time and appropriate budget), and on the other hand system performance evaluation should be anticipated and integrated in the lifecycle, to be able to predict system behavior and variables that affect it, and consequently produce "performance informed" design iterations (with performances considered at interaction, architecture and coding levels). We also sketch a possible prediction approach.