Say‑Do Gap Diagnostic

Use this when surveys or interviews diverge from observed behavior.

1. Is prior behavior available
   • Yes: condition on prior behavior for recruitment and segmentation.
   • No: proceed to context sampling.
2. Can you observe in natural context
   • Yes: field observation or shadowing.
   • No: instrument a pilot or diary study with time‑stamped events.
3. Can you run a small field experiment
   • Yes: A/B a minimal intervention in real workflows.
   • No: use choice‑based tasks that mimic constraints, then validate in the field.
4. When to ignore self‑report
   • Aspirational answers, sensitive topics, or high social desirability.

Output: a short memo with your method choice, constraints, and expected Δ‑B window.