Elicitation that does not lead the answer
Leading questions feel efficient and quietly warp the requirement. Prefer prompts about current work, exceptions, and failure modes. Capture quotes; paraphrase later in a separate note so raw evidence stays intact.
Observation still beats another survey. Shadow a shift, note handoffs, and ask why a workaround exists. Those details become acceptance criteria that testers can actually use.
Desktopbranch’s elicitation lab drills these habits with peer feedback so analysts hear how they sound under time pressure.