We believe that we need to put more emphasis on helping students develop problem solving capabilities in our teaching on software development.Our research question is how much mental effort students invest in the different phases of problem solving: Understanding, Solution Planning, Implementation and Review. Other questions are which category of thinking they focus on, and how the effort put into each phase correlates to success.We designed and performed an empirical study with the intention to gain insights into problem solving skills of our students. In this study, CS-students were asked to solve different typical programming problems and they were guided through the process of problem solving with a kind of coding interview. First results show that that the implementation makes up the smallest portion of the effort, measured in terms of number of characters in our interviews’ transcripts. Students who were more thorough in analyzing the problem had generally longer implementation phases, indicating more substantial changes to the implementation than those who could not solve the problem. Unexpectedly, the analyzing phase took the successful students just as long as the ones who were not successful.All students repeatedly switched between the typical phases of problem solving. Students who proceeded in a more structured manner were the most successful. We see evidence that a lack of programming knowledge was not a main cause of failure in the fact that those students who could not solve the task have already failed at the problem analysis.Using insights gained from this kind of study will help to improve teaching of problem solving in introductory programming courses the future.