Run an import

This is the hands-on version of Importing from your current stack. You need to be a company admin.

Before you start

• Create the target project in os.construction, with its cost codes and permission templates.
• Have the source company ID and project ID from your current system, and an access token for its API.
• Try it on sample data first if you like: the sample-data source needs no credentials.

Create a job

Run it

Open the job. For a connected import, paste the access token and select Run connected dry-run (or Run connected import for a live job). The token is used for that request only and isn’t saved. For sample data, use the run button on the job.

Read the results

The job page shows:

• Per-entity stats: for each record type, how many were Fetched, Created, Updated, Skipped and Failed.
• Run log: a timestamped line for each step and each failure, with the reason.
• The job’s source, mode, status, target project and start and finish times.

A dry run fills in the counts without writing records. Check them against the old system before you run live.

Re-run safely

Every imported record remembers where it came from. Running the same live job again updates the records it already created rather than duplicating them, and picks up anything new in the source. If a run is interrupted, the job page offers recovery: retry with the same source, and records already committed are kept.

Fix failures

Failed records don’t stop the run. Read the reason in the run log, fix it at the source or in os.construction (often a missing person or company), and re-run the job.