While requirements elicitation remains a key success factor in software projects, there is little empirical research on the elicitation methods. We focus on fit-gap analysis, a requirements elicitation technique that is common in practice, but hardly studied in requirements engineering. Fit-gap analysis is a method for matching software products with the needs of customers, and identifyies needs that are supported (fits), and needs that are not (gaps). Through a grounded theory investigation of recording transcripts from fit-gap analysis, we provide empirical knowledge about this elicitation technique. We determine and discuss the different categories of the topics contained in a fit-gap analysis. Additionally, as a first step toward assisting analysts in processing and exploring their analyses, we build and share a set of keywords and phrases that can help automatically identify the categories within the transcriptions. As an early validation, we conduct an experiment, involving both students and practitioners, that determines the relative perceived importance of the identified categories. Finally, we derive implications for research in the field that include our perspective on how software tooling can assist analysts in fit-gap analysis.