Who is actually in the decision
Separate the economic sponsor, the operator, the security or compliance reviewer, the internal champion, and the person who inherits the system after go-live. Titles lie. A vice president may open the budget and still be unable to describe the weekly exception queue. An individual contributor may hold an unofficial veto because the work lands on their desk when the tool fails at 6 a.m.
Recruit for the veto and the inheritance, not for the most senior person who will take a meeting. If you can only reach the champion, write that limitation in the note. A claude reading does not upgrade a champion’s enthusiasm into “the market wants this.” The buying committee guide shows how those roles split the right to say no, which is the list you should be recruiting from.
Build a decision trace, not a quote bank
Walk the last purchase, the last loss, or the last renewal backward. Ask for the calendar: the first internal message, the first vendor conversation, the meeting where it nearly died, and the artifact that got forwarded. That artifact is often a security questionnaire, a one-page comparison, or a recap someone pasted into chat after the call. Stay with one deal long enough to see sequence.
Five shallow calls that ask for pain points produce slogans you could have written yourself. Two deep traces produce material the positioning synthesis page can actually use. Mark what was said apart from what you inferred. A month later you will not remember which objection was uttered and which one you added because it sounded strategic.
Questions that hold still
Ask what was already in place, and what became irritating enough to open a search. Ask who inside the account had to be convinced, and which question they posed that the champion could not answer on the first pass. Ask what a failed implementation would have cost the person in front of you, not “the company” in the abstract. Ask which touchpoint made the risk feel specific. Ask what they almost believed, and which sentence killed it.
Pair those questions with touchpoint observation when the buyer mentions an artifact you can actually see, and with purchase friction when the story is about delay rather than desire. The interview finds the sequence. The other two lenses test it against behavior and against the reasons a yes stalls.
What to stop asking
Stop asking people to rank your feature list. Stop asking them to forecast ROI. Stop asking how many companies “like them” exist and then treating the answer as a sample frame. Those questions invite fiction, and fiction is easy to slide into a slide. The claude standard refuses representative statistics built from whoever agreed to talk. Range is the goal: a won deal, a lost deal, a stalled renewal, and a non-buyer who stayed with the incumbent.
The protocol page turns that range into a recruiting grid. The evidence standard tells you which sentences may leave the building. If a quote only works when the champion is still in the room, you have not finished the interview.