Modeling Guide
Knowledge Guides /How Should Spreadsheet Requirements Be Imported? Check Field Mapping and Record Identity First
Last updated: · New Horizon Digital Outsourcing Team
Before importing requirements, agree on what a row represents, required fields, record identity, and handling of empty or invalid values. Trial representative records in a separate workspace and reconcile records and fields. A success message does not establish that meaning has been preserved.
Prepare the Source Structure
- Keep a copy of the source and identify the worksheet, header row, data range, and treatment of hidden rows. Separate notes, merged cells, and group headings from requirement records so that presentation does not become an unintended hierarchy.
- Define stable identifiers and source locations. Identical names need not represent duplicate records, and changing row numbers should not create new identities. Hold records without identifiers for resolution instead of treating titles as unique keys.
Map Columns to Defined Data Types
- Define the target field, type, unit, and required status for each column, distinguishing requirement text, provenance, and notes. Agree on enum mappings and handle dates, integers, and identifiers with leading zeros explicitly rather than inferring meaning from displayed text.
- Prepare required target types and attributes before the trial. Define rejection, retention, or manual review for unmapped columns, invalid enum values, and empty values. Do not hide incomplete data by silently dropping columns or inserting defaults.
Validate Rules with a Small Trial
- Include ordinary records, long and multiline text, special characters, empty fields, and duplicate identifiers. Import into a separate workspace and record accepted, rejected, and unresolved records with reasons.
- Import the same sample again to determine whether existing records are created again, updated, or rejected. Compare observed behavior with the agreed identity rules before making bulk writes to the production requirements repository.
Reconcile Records and Meaning before Scaling
- Reconcile source identifiers against target records, field values, and hierarchy, with separate counts for rejected and unresolved items. Equal totals can hide one omission and one duplicate, so compare identifiers rather than row counts alone.
- Deliver import rules, field mappings, exceptions, and sample review records, stating whether relationships, attachments, and formatting are in scope. These are recommended acceptance checks; platform and version capabilities must be established by an actual trial.