Uploading via S3

Drop people and company CSVs to your Primer S3 location for automated loading.

Beta — opened on request. Contact [e-mail] to request access.

You can load people and companies into Primer by dropping CSV files to an S3 location Primer reads. This is the automated equivalent of importing a CSV and works well with pipelines like Snowflake, Hightouch or Fivetran.

Where files go

Put each record type in its own folder under your dataset prefix:

• People — {prefix}/datasets/people/

• Companies — {prefix}/datasets/companies/

Your prefix comes from your sync configuration; support provides it when you're set up.

File format

Each file is a CSV with a header row using the canonical people or company columns. Each file is a complete snapshot of that dataset — records absent from a new file are dropped, not merged.

What happens after you drop a file

Primer picks up the file, processes it, and moves it into an archived/ subfolder on success or a failed/ subfolder on failure, so you can tell whether a drop was ingested.

Exporting to Primer's S3 from your warehouse

Any tool that can write a CSV to an S3 prefix can feed this drop. The pattern is the same whatever you use: produce a CSV with the canonical columns and land it in the right people/ or companies/ folder under your prefix.

• Snowflake — COPY INTO '@your_stage/...' (or an external stage pointed at the prefix) to unload a query result as CSV directly to the S3 location.

• Hightouch — configure an Amazon S3 destination targeting your prefix and sync a model to it on a schedule.

• Fivetran — use an S3 destination / transformation that writes the CSV to the prefix.

Because each file is a complete snapshot, schedule the export to write the whole current dataset each run, not just changed rows — anything absent from the newest file is dropped. Your prefix and the exact folders come from your sync configuration; support provides them when you're set up.