Loading learning data into SuccessFactors with connectors

SuccessFactors Learning’s connectors load users, items, curricula and learning history from pipe-delimited files dropped on its SFTP server. A file that’s almost right is the worst kind: the connector runs overnight, skips the rows it doesn’t like, and the log tells you in the morning. These tools catch those problems before the file goes up, in your browser, so personal data in the file isn’t uploaded anywhere.

1. Start from the spreadsheet

Put the data in a spreadsheet with the connector’s field names as the header row. Copy the cells (header first) and paste them under Convert to connector format in the Connector Checker. It writes the rows with | between fields and !##! at the end of each, quoting any value that contains a |. Save as a file gives connector_data.txt; rename it to the file name the connector expects.

Sent a workbook (.xlsx) or a CSV export from another system? Open it in CSV ⇄ JSON and choose Show as a table to check the columns lined up: in a CSV, an address with an unquoted comma shifts every column after it. A workbook opens as CSV, with any dates written as 2026-01-31, ready to check and convert.

2. Make sure it’s UTF-8

Connector files must be UTF-8, and spreadsheets often save as Windows-1252, which turns José and Zoë into garbage or stops the file loading. Open the file in the Encoding Fixer with File’s encoding on Detect: it shows what it found and the result as UTF-8. Leave Start saved files with a UTF-8 BOM unticked, because a BOM ends up glued to the first column’s name, and Save as a file.

3. Check every row

Choose the file under Check a file in the Connector Checker. It works out the connector from the file name or the header and checks the header against the connector’s fields, then every value: required fields left empty, text longer than the field allows (counted in bytes, so accented letters count double), Y/N fields, numbers, time zones, and dates, which must be written MON-DD-YYYY HH24:MI:SS, as JAN-31-2026 00:00:00. Fix the errors in the spreadsheet, convert again and check again until none are left.

Reference fields lists values such as domains, job codes and user IDs that must already exist in the LMS: the checker can’t see the LMS, so check those there. Add your own rules under File settings and your own rules, such as ITEM_ID = [A-Z]{3}-\d{4}, to catch your organisation’s naming mistakes.

4. Split a very large file

A history load with hundreds of thousands of rows can time out or be rejected for its size. In CSV Split & Join, set Rows per file, tick The first row is a header so each part gets it, and set Row end marker to !##!, so rows containing line breaks aren’t cut in half. Split and save (ZIP) gives numbered files: rename each to the connector’s pattern and load them one at a time.

5. Read the log when it still fails

Paste the lines from the connector’s log under Look up error codes in the Connector Checker: each code is listed with its severity and what it means, next to the lines it came from.