We explain, and illustrate with examples drawn from our own work, how interpretive researchers analyse comparative data. We argue that a comparative project compounds the uncertainty, confusion and paralysis that can set in when confronted with a 'mountain' of qualitative data. We argue it is not possible to 'somehow capture' this full complexity. We outline and defend the need for a consciously impressionistic orientation to data analysis. Rather than searching for a ‘Eureka!’ moment that confirms or refutes a narrow theory (in naturalist mode) or makes sense of the whole picture (in idiographic mode), a comparative focus on dilemmas enables the use of a kaleidoscope of different analytical lenses and tools to explore complex specificness in context. We outline rules of thumb for helping along the way.