Methodological literature on qualitative research interviews often warns against imposing the researcher’s definition and framing of a problem on research subjects. By contrast, social constructionists consider interviews as conversations, and thus view ‘leading questions’ as a natural part in the interview interaction. They see the interview as mutually co-constructing meaning. But what if the interlocutors do not seem to agree on what they construct? What if the interviewee has a particularly strong agenda, far from the intended research topic? Are these ‘failed’ interviews?