Integrations, Imports, and Exports
Moving data in and out
Imports and exports are defined by templates: file format (CSV, TSV, XML), encoding, delimiter, header handling, column mapping and a reconciliation key. The import wizard supports insert, update, and update-or-insert modes; rejected rows go to a reject file with a reason. In workflows, the pattern is Load file → Reconciliation → Enrichment or Update data, with Transfer file and Unzip for SFTP sources. Exports use an export template, a target mapping, and optional compression or encryption. External accounts hold SFTP or API credentials and should be referenced, never hard-coded.
Traps
- Without a reconciliation key, every import line becomes a new record.
- Date and number formats follow the instance locale unless explicitly mapped.
- Use first line as column names changes the mapping; skip it and the header becomes data.
- Reject files are not automatic retries; they need handling.
- An export of a workflow target exports the work table, not the source schema, unless you re-query.
- Encoding mismatches produce mojibake, not errors.
Rule of thumb
Normalize the reconciliation key before import (trim, lowercase email or a CRM ID) and run a test import into a small segment. Always download and inspect the reject file.
Worked example: Daily CRM contact sync: Transfer file from SFTP → Load file (CSV, UTF-8, comma, header) → Reconciliation on crmId with update-or-insert → Update data to lowercase email → reject branch to a log delivery. If crmId is blank, route to reject rather than matching on email.