Repeatability
High
The transformation logic is structurally identical for every record: flatten nested donation history, compute four aggregate fields, and standardize address formatting. This runs the same way every quarter with no judgment variation.
Ambiguity Tolerance
High
Success criteria are concrete and verifiable: 1,100 output rows, correct aggregation totals, consistent address format, and no dropped records. A script can self-validate against the source data.
Data & Tool Availability
High
The agent needs only the exported JSON file and a Python or similar scripting environment — no external APIs, live systems, or special permissions required. Everything needed is self-contained.
Error Cost
Medium
Errors like miscounted gifts or wrong totals could corrupt donor records in the new CRM, which matters for fundraising accuracy. However, the migration is reversible — the source JSON is preserved and output can be validated before import.
Human Judgment Required
Low
Address standardization follows well-established rules (USPS formatting, regex normalization), and missing phone numbers can be flagged rather than guessed. No subjective or relational judgment is needed.