While online discussion platforms have been widely used and received a lot of attention in the infor_x005fmation systems communities，its information organization structure is still in the original form．Research that explains what kind of information organization structure is more suitable for online discussion is rare and confused if any．Thus a theory for optimal information organization in online group discussion is highly desirable．In this study，we follow the perspective of task-technology fitness，describe the fitness between tasks and information organization structures from the three perspectives of communication，process structuring，and information processing，and predict group performances based on task-technology fit condition．We also designed an experiment to validate if a match between task and information organization structure will lead to increased performance．The results indicate that for idea generation tasks，using a groupware with linear structures will lead to higher group performances compared to tree structures．On the other hand，for judgment tasks，using a groupware with tree structures will lead to higher group performances compared to linear structures． The proposed theory and method is extendable，which means we can evaluate the fitness between new task types and unseen technology，and predict the group performances．