THE STARTING POINT
Make exceptions, unavailable dependencies and manual recovery part of the brief.
List the interruptions.
Consider missing inputs, unavailable systems and outputs that cannot be accepted. Describe what happens to work already in progress when a dependency fails. Distinguish a temporary delay from a case requiring a business decision, so the workflow does not treat every interruption as an automatic retry.
Keep a usable manual route.
Identify what information a person needs to continue the task safely. Make pending work and completed actions visible to avoid duplicate processing. Discuss how the manual route will be rehearsed during the pilot rather than assuming someone can reconstruct the task when the automation stops.
Review recovery ownership.
Assign responsibility for resolving exceptions and deciding when automated processing can resume. Record the checks needed to reconcile manual and automated work. A fallback should be part of the workflow’s operating instructions, not a separate document that nobody knows to consult.
TAKE THE NEXT STEP
Your discovery checklist
- List missing-input and unavailable-system scenarios.
- Rehearse a manual route without duplicating actions.
- Define reconciliation checks before automation resumes.