A cognitive walkthrough is task-specific, whereas heuristic evaluation takes a holistic view to catch problems not caught by this and other usability inspection methods.
The method is rooted in the notion that users typically prefer to learn a system by using it to accomplish tasks, rather than, for example, studying a manual.
The effectiveness of methods such as cognitive walkthroughs is hard to measure in applied settings, as there is very limited opportunity for controlled experiments while developing software.
However, Gray and Salzman called into question the validity of those studies in their dramatic 1998 paper "Damaged Merchandise", demonstrating how very difficult it is to measure the effectiveness of usability inspection methods.
After the task analysis has been made, the participants perform the walkthrough:[1] The CW method does not take several social attributes into account.
The method can only be successful if the usability specialist takes care to prepare the team for all possibilities during the cognitive walkthrough.