Import data in.
Update in place.

Two import modes

• Data import — upload a wide CSV, insert new respondents only.
• Mid-field import — update specific columns on existing respondents by respondentId.

Both read the same canonical model as Export, so a CSV that came out of the Export wizard reads cleanly back in.

Full dataset import

Drop a CSV. New respondents only — safe to re-run a half-failed import without creating duplicates:

• Rows with an existing respondentId are skipped.
• Duplicate IDs within the file collapse to the first row.
• Rows without a respondentId get a freshly generated one.

For panelist drops, pilot seeds, or historical data from a previous platform.

Mid-field update

Update specific columns on existing respondents by respondentId. Column-scoped and non-destructive for unrelated data. Common uses:

• Post-field coding — recoded open-ends land in Q18_code only.
• Panelist weights — hWeight from the panel provider after fieldwork closes.
• Back-office enrichment — CRM segment code against a known respondentId.

Import label & batch tag

Every import carries two labels:

• Import label — stored on each imported respondent, filterable in Data view and Export (e.g. Pilot batch, Panel A seed).
• Wave / batch tag — optional, for cross-wave analysis in Reports and in the Data view's Source column.

Filterable in Data view and Export for cross-wave or by-source cuts.

Schema validation

Nothing writes until the schema validates. Validate runs the file against the live survey structure:

• Unrecognised columns rejected (not silently dropped).
• Code values outside the question's axis flagged.
• Type mismatches (text in a numeric, out-of-range dates) reported with row numbers.

A precise diff before you commit.

Import history

Every import logged: label, batch tag, filename, respondent count, imported-at, by-whom. Failed imports kept with their validation errors, so you can retry the same file after fixing the schema.

Dataset summary

Summary panel at the top of the Import tab:

• Total respondents — live + imported.
• Imported records — just the ones from this tab.
• Variables — total column count across the current datamap.
• Import batches — distinct wave/batch tags seen.

Click any number to filter the Data view down to that scope.

Recode & derive (SOON)

Post-field derived variables — net scores, bucketed numerics, combined codes — on the roadmap. Today the same work runs inline from a script block; the Recode UI will add a point-and-click layer on top of the same engine.