We present the analysis of an episode of mathematical problem solving in a group, where data came from multiple advanced recorders, including multiple video cameras, Smartpen recorders, and mobile eye tracking glasses. Analysis focused on a particular group that was ineffective in their problem-solving process. Relying on the commognitive theory of learning on the one hand, and on quantitative descriptors of eye-tracking data on the other hand, we ask how do the interpretations of the discourse analysis and gaze data complement each other in understanding the obstacles to problem-solving in this episode. The setting included four Finnish 9th grade students solving a geometrical problem in the students' authentic mathematics classroom. The commognitive analysis revealed intensive social communication (subjectifying) along with the mathematical one (mathematizing), which seemed to interfere with the problem-solving process. Specifically, it masked the differences in students' interpretation of the tasks, and did not allow explication of meta-rules according to which students endorsed mathematical claims. Diagrams of quantified gaze data enabled a more macro-level picture of the full 15 min interaction, revealing differential loci of attention of the group members and thus triangulating the micro-analysis.