Configure a CSV Import Payload and its schema
Use a CSV Import Payload to read a Salesforce File in chunks and process its rows through Actions. Define the shape of one row, then add the Queries, Targets, Flows or child Payloads that should process each chunk.
Create the Payload and settings
Open the Integration, select New Payload and choose CSV Import.
Save the Payload and select its required CSV Import Action.
Select Edit CSV settings.
Choose the Delimiter and Quote character that match the source file.
Set Rows per chunk to a whole number from 1 to 10,000.
Save the settings. If this Payload will be started from Flow, configure the Salesforce file connection described below.
Start with a small chunk while testing. The configured number is a row limit for each processing transaction, not a guarantee that the transaction fits Salesforce CPU, heap or DML limits.
The file connection is required for Flow imports and is also used by the manual runner when configured. See Set up the Salesforce file connection.
Import the row structure
In Body, select Import CSV headers. Paste the header row from the CSV, a tabular row copied from a spreadsheet, or a supported Markdown table with its separator row. Select Import.
Importing headers replaces the existing Body structure. Review existing mappings before replacing an established schema.
Use dots for nested objects and indexed brackets for arrays. For example:
customerKey,name,address.city,contacts[0].email,contacts[1].email CUSTOMER-1042,Example Customer,Dublin,[email protected],[email protected]
The two indexed email columns share one Array element definition with an email String child. The Body models one CSV row; at runtime the CSV Import Action provides a collection of those rows.
Leaf values remain Strings. Configure conversions in the consuming Target fields or other supported mappings when a Salesforce destination needs a number, date or Boolean. Do not assume the sample values make the CSV schema numeric.
Define one row, then add the Actions that process the collection.
Add row processing
Select a plus button after CSV Import and choose the next Action. For a Target that processes every source row:
Select the destination object and operation.
Choose Multiple Records.
Set Value Source to Payload Data.
Select CSV Import and CSV Rows.
Save the Target, then add its destination field mappings from the appropriate row fields.
Multiple Records uses the CSV Rows collection from the earlier CSV Import Action.
Later Queries, Flows and eligible child Payloads can participate in the sequence. Observe the same dependency and callout-before-DML rules as other Actions.
Check before running
Use a small CSV that includes normal, blank and invalid values. Check nested paths, conversions and the Target's matching identifiers. Confirm that repeated runs will not create unwanted duplicates.
Do not change schema, settings or Actions while an import is active: later chunks read the current Payload configuration. The file version remains pinned, but the configuration is not snapshotted for the entire run.
Continue with Run a CSV import.