This paper presents empirical research results regarding the impact of Interaction Analysis (IA) graphs on groups of students collaborating through online blogging according to a "learning by design" scenario. The IA graphs used are of two categories; the first category summarizes quantitatively the activity of the users for each blog, hence permitting the comparison of the students' activity level in a group, while the second category allows the comparison among different groups. The statistical analysis of the students' interactions shows the significant impact of the graphs presence based on the number of posts and comments produced by the groups. Furthermore, the graphs of the first category (intra-group IA) have stronger impact than the graphs of the second (intergroup IA). The results support the general claim that interaction analysis is an important component of self-regulation in computer supported collaborative learning environments. In addition the research conclusion suggests that collaborative learning by design is an effective model especially when combined with interaction analysis tools.