A pipeline maps the same CSV row to several related entity types. See Single vs pipeline for when to use it.

Choose the entities

  1. Open Data In and Out → Upload a file, select your CSV and click Next.
  2. In What to import, select each required entity card.
  3. Review Processing order, then Continue. The wizard sets this order automatically; it does not offer an Add step or reorder control.
  4. Map the columns for each entity. A source column may be reused by several entity mappings.

Example: organisations and contacts 

Organisation,First Name,Last Name,Email Address
Example Charity,Alex,Example,alex@example.com
Example Charity,Sam,Example,sam@example.com

Select Organisations and Contacts. Map Organisation to the organisation Name, and map each person’s columns to the contact fields. For each row, the organisation is resolved before the contact, and the contact is linked to that row’s organisation. The automatically created relationship uses the linked connection type.

There is no arbitrary “join these two CSV columns” configuration. Links come from the entity steps for the same source row. For stable identities across later files, map each entity’s source ID under External Keys, with a consistent platform name. Existing contacts can also match by email; organisations by name. Review matching carefully before re-importing changed data.

Other linked records

A Contact or Organisation step can supply the owner for payments, touchpoints and recurring-payment arrangements. Memberships reference previously imported contact/payment external IDs. A fundraising-page member needs both the contact and page to resolve.

Connections are a special case: both ends must already exist from an earlier completed import. Select From side and To side types, map the relevant identifiers, and select or map a connection type. When using external keys, provide the shared External key platform. A Connections step cannot rely on records newly created by the same job.

Check and submit

Run the pre-flight check, fix required mappings and review row warnings. Confirm the file, mappings, date format and processing order, then Start Import. See Run an import for the shared stages.

A missing or unresolved reference can fail the affected row; it is not silently guaranteed to create a usable relationship. Other rows can succeed. Check overall status, per-entity progress and the error report, then correct the source data or import the missing dependencies before rerunning the affected records. A failed job is not a rollback. See Job statuses and errors.