Notion CSV imports: merge adds rows rather than updating records

SOURCE-BASED ANALYSIS · NO HANDS-ON RATING

Merge is not an update contract

Notion's current CSV documentation says imports and CSV merges add rows rather than update existing rows. Rows become pages and columns become properties. Treat Merge with CSV as an addition workflow, not an upsert based on a matching title.

Our recommendation is to separate a one-time migration from ongoing synchronization. If existing records must change, define and validate a supported update process independently instead of repeatedly importing the entire worksheet.

A fictional second batch

Assume a team imported ten project requests yesterday and receives a revised twelve-row spreadsheet today. Ten rows describe existing requests and two are new. Importing the whole revised sheet is not a documented method for updating those ten existing pages.

Keep a stable source ID for each request. Reconcile the two new IDs separately from the ten records requiring changes. The row counts are hypothetical inputs, not measured behavior or a product limit.

Prepare the mapping

• Source ID: a durable reference that does not depend on the display title.
• Title: the intended page name after import.
• Owner and status: explicit target property definitions.
• Date: an unambiguous example checked against the documented import format.
• Relationships and calculations: list what requires separate configuration.
• Outcome: imported, deliberately excluded or rejected with an explanation.

Rehearse before the full batch

1. Keep an unchanged source copy and identify the authorized destination.
2. Use three synthetic rows, including similar titles and one empty optional field.
3. Inspect the CSV headers and match columns to intended properties.
4. Import the sample and compare every source ID with the resulting page.
5. Inspect dates, numeric-looking values and text containing commas.
6. Document how an interrupted import will be reconciled before repeating it.

The official guide also notes that a new CSV import does not create new formulas or rollups, and that relation handling differs between a new import and mapping into an existing relation property. Review those requirements explicitly.

Resolve ambiguous results

Our troubleshooting advice is to pause when a row cannot be matched confidently. Search by the stable source reference, inspect the imported values and record the exception. A matching title alone is weak evidence when several requests share the same name.

Do not remove the source worksheet just because the destination count looks plausible. A count can conceal a missing request and an extra duplicate. Compare identities as well as totals, then inspect representative page content.

Acceptance criteria

Approve the batch when every in-scope row has a matched page or a documented exception, required property values are readable and repeated execution has a clear policy. This guide offers a source-based migration checklist, not a completed importer test or an assertion that CSV can reproduce every feature of the original system.