Extracted data
Extracted data is exactly what Graftport pulled from the source platform. One row per source object, keyed by the source’s stable ID.
Where to find it
Open Data in the sidebar, pick the migration and the resource, and leave the view on Source. Every row is one extracted source object, shown in the source platform’s own field names.
From there you can:
- Put any source field on screen as a column, or filter on it — a SKU, an email, a status.
- Click a row to read the whole payload as a collapsible tree, with the mapped output directly underneath it.
- Narrow to the rows one run touched, using the run scope control.
See Data explorer for the screen in full.
What it’s good for
- “Did we pull this row?” — filter on the ID and see.
- “What does the source actually contain for this product?” — open the row and read the payload.
- “Why did this transform output X?” — read the source record and the mapped output in the same panel; the answer is usually visible without leaving the page.
Downloading a snapshot
You can take a whole snapshot with you. On a migration’s Sources tab, open any source row and it offers a Download extract button. Clicking it downloads everything that source has extracted as a JSONL file — one record per line, each one exactly the source row your mapping receives.
That is the same row the Source view shows and the same row the mapping preview runs against, which is not always one-to-one with what the source platform’s API returned. On Magento, for example, a configurable product arrives with its variants already gathered onto it, so its simple child products are part of that record rather than records of their own. Writing a mapping against a download is therefore safe: what you see in the file is what your JSONata gets.
A few things worth knowing:
- Each source downloads separately, so a resource fed by a store connection and two files gives you three files, one per source.
- The download reflects the most recent extract for that source. Run an extract first if the file comes back empty.
- Large resources take a while — orders in particular. The file streams as it’s produced, so your browser shows it downloading the whole time.
Because the format matches what Graftport accepts as a file source, you can edit a copy and add it back as a new source if you need to correct data the source platform can’t fix.
What’s stored
The full API response object for that source row, untouched. Graftport doesn’t strip or normalize before storing — that’s the transform’s job. The extracted data is the ground truth, and every downstream artifact can be re-derived from it.
Retention
Extracted data persists for the lifetime of the migration. Even after a successful load, it sticks around — that’s what makes re-transforming cheap, and what auditors need to see.
There’s no automatic pruning today. Workspaces that need PII purge can contact the Graftport team to arrange one.