Detect Input Drift Before It Breaks an Automation

ROLE Act as a data quality monitoring specialist. OBJECTIVE Design input-drift monitoring for [AUTOMATED WORKFLOW]. CONTEXT Input sources and schema: [INPUT SPECIFICATION]. Historical examples: [BASELINE SAMPLES]. Known seasonal changes: [EXPECTED VARIATION]. Business-critical fields: [CRITICAL FIELDS]. INSTRUCTIONS Separate schema changes, missingness, new categories, formatting shifts, language changes, and unusual volume patterns. Identify signals that can be checked deterministically before involving an AI model. Define baseline windows and explain how sparse data limits statistical conclusions. Specify warning versus quarantine conditions based on downstream impact. Include a process for distinguishing legitimate business changes from upstream defects and for updating baselines only after review. Avoid treating every unfamiliar value as an error. Show how affected records remain traceable without logging unnecessary sensitive content. OUTPUT FORMAT Provide a monitoring matrix, baseline worksheet, alert examples, and three synthetic drift scenarios with recommended handling. QUALITY CHECK Flag assumptions and avoid fabricated statistics. Do not invent historical distributions. Label illustrative thresholds and explain how the team should calibrate them using actual observations.