You read back a transcript and every answer is a generalisation: I usually check on Mondays, I would probably use it for invoices, we tend to reconcile at month end. Judged against story-based interviewing, what has happened here, and what changes in the next session?

The principle — a transcript of generalisations means the questions asked for generalisations, and that is fixable in the asking.

Why the key is correct — Torres's rule is to Keep the interview grounded in specific instances of past behavior. Every answer in this transcript is the opposite: usually, probably and tend to are all summaries the participant produced instead of recounting anything. The repair is in the question form. Her demonstration is Tell me about the last time you watched Netflix. — one occasion, named, with the person asked to walk through it. In the next session that means replacing every do you and would you with a last time you, and when a generalisation arrives anyway, following it immediately with a request for the most recent instance of it. The detail a team can act on lives in the instance: which day, what interrupted it, what was tried first, and why the person stopped.

Why the others are wrong — accepting a habit as a past behaviour treats an average as though it were evidence, and the averaging is precisely where the actionable detail was lost. Reading the answers as evasion misdescribes both the participant and the purpose, since summarising is simply how people answer questions about themselves and the technique exists to collect better material rather than to test anyone's honesty. And claiming a habit carries the same detail as an occasion is contradicted by the transcript itself, which contains no dates, no triggers and no outcomes.

Remember this — usually and probably are the sound of a question that asked for a summary.

Sources — Teresa Torres on story-based customer interviews, Product Talk.