The purpose of this paper is to present a technique, called Knowledge FMEA, for distilling textual raw data which is useful for requirements collection and knowledge elicitation. The authors first give some insights into the diverse characteristics of textual raw data which can lead to higher complexity in analysis and may result in some gaps in interpreting the interviewees’ world view. We then outline a Knowledge FMEA procedure as it applies to qualitative data and its key benefits. Examples from a case study are presented to illustrate how to use the technique. Proposed Knowledge FMEA brings many advantages such as forcing the analysts to become deeply immersed in the raw data, identifying how the information is connected in causation, classifying the data according to why, what, how formulations and quantifying the findings for further quantitative analysis.