observation-analysis

Define the process scope, select 2-3 participants at different experience levels, and schedule during real work rather than staged demos. Record timestamped actions, artifacts, and signals like workarounds or delays, then validate findings with participants within 48 hours.

Passive observation (job shadowing) means watching silently and asking questions afterward, minimizing disruption to natural behavior. Active observation (contextual inquiry) involves asking questions during the task, giving deeper insight but potentially altering the workflow.

Review notes for patterns like workarounds, pain points, delays, errors, and handoffs, each of which signals an implied gap. Draft a candidate requirement per finding with the observation source recorded, then refine it using a requirements-writing skill.

This is the Hawthorne effect: people perform the ideal process instead of the real one when watched. Counter it by observing longer than 15-20 minutes, running multiple sessions, and comparing observations against error or defect data.

Use observation when interview data conflicts with defect data, when stakeholders claim a process works fine despite high error rates, or when you need to understand physical environment constraints. Interviews reveal what people think they do; observation reveals what they actually do.