Run a specialist troubleshooting clinic with useful evidence

A colleague brings a problem to a drop-in specialist session: a workflow behaves differently from what they expected, but they cannot yet explain why. The useful outcome is not always an immediate fix. A troubleshooting clinic should produce a clearer example, an examined explanation, or a handoff another specialist can use. Arrange waiting separately through the team's existing process.

Describe the expected and actual result

Ask the visitor to name the task, the expected result, and the point where the actual behavior differs. Record the relevant starting conditions, including the application or document involved. Replace a broad description such as it is broken with an observable event. Avoid interpreting frustration as a technical diagnosis; acknowledge the inconvenience while collecting enough detail to investigate.

Build a safe reproduction

Use a fictional or non-sensitive example when the original involves private information. Ask the visitor to repeat the shortest sequence that still produces the problem. If screen sharing is available, verify what the specialist actually receives before relying on it. Otherwise, work from approved screenshots and a written sequence. Never request passwords or authentication secrets to make the example easier to reproduce.

Test one explanation at a time

State a hypothesis and the observation that would support or weaken it. Change one relevant condition, then compare the result. For example, inspect whether the starting document differs before assuming that a visible control failed. Keep the original state recoverable where possible. Several simultaneous changes may seem faster, but they make it harder to explain why the behavior changed or whether the proposed fix is reliable.

Separate a workaround from a diagnosis

A workaround can unblock the visitor without establishing the cause. Label it that way and record any limitation. If the proposed action could remove work, change shared settings, or affect other participants, stop for the appropriate owner and recovery plan. A casual help session is not authority to modify every system involved. Recheck the visitor's original task rather than celebrating an unrelated successful action.

Hand off evidence when time runs out

Record the reproduction steps, observations, hypotheses tested, and the next investigation owner. Note anything that could not be checked. Give the visitor a realistic route for an update through an established channel, without promising an automatic ticket or guaranteed resolution. A clinic succeeds when the next person can continue from evidence instead of asking the visitor to repeat the entire story.