# CSV import Bring existing feedback into Usero from a CSV file or a pasted spreadsheet range. The rows land in your inbox as feedback items, run through the same AI clustering as everything else, and any of them can become a GitHub pull request. This is the fastest way to migrate off another tool (Canny, Productboard, Featurebase) or to pull a support backlog out of a spreadsheet. ## What it does - **Bulk-load feedback from a CSV.** Upload a `.csv` file or paste a range copied straight from a spreadsheet. Usero parses it in the browser, reads the headers, and walks you through mapping, preview, and commit on one screen. - **Auto-detects common exports.** Canny, Productboard, and Featurebase exports are recognized and their columns are mapped for you, so you usually just confirm the mapping. Any other tool that exports CSV works on the generic path. - **Tags the source.** Imported rows are marked as a CSV import and carry an importer id, so you can tell them apart from widget, email, and GitHub feedback and undo a bad import as a unit. - **One-hour undo.** Every completed import can be undone for one hour after it finishes. ## Prerequisites - A Usero account and a project. You need to be a member of the project you are importing into. - Your feedback as a CSV file, or open in a spreadsheet so you can copy a range. If your old tool only exports JSON or XML, convert it to CSV first (most spreadsheet apps do this in two clicks). ## Prepare your CSV You do not need to rename columns before importing; you map them in the wizard. But a clean file imports more cleanly: - **One header row.** The first row is treated as column names. - **One feedback item per row.** Each row becomes one feedback item. - **A title column.** Every row needs a non-empty title. Rows with a blank title are flagged in the preview so you can fix or skip them before they import. - **Optional columns** Usero can map: a content/description column, author email, author name, votes, status, and a created-at date. Anything else is ignored. - **Dates** in any format `Date.parse` understands (ISO `2026-01-15`, `Jan 15 2026`, and similar) are read as the item's created date. Unparseable dates are left blank rather than failing the row. - **Emails** that do not look like an email address are dropped rather than imported as junk. Canny exports the post body as HTML; Usero strips the tags so you get clean text. You do not need to clean it yourself. ## Upload and map columns 1. Open your project, go to **Integrations**, and find the **CSV import** card. Click **Start a CSV import**. (Direct URL: `usero.io/YOUR_CLIENT_ID/import/csv`.) 2. **Bring it in.** Drop your `.csv` file onto the drop zone, or paste a copied spreadsheet range. Usero reads the headers immediately. 3. **Match your columns.** If Usero recognizes a Canny, Productboard, or Featurebase export, the mapping is pre-filled and a banner says so; confirm it or override to the generic mapping. Otherwise, map each Usero field to one of your column headers. **Title is required.** Content, author email, author name, votes, status, and created-at are optional. 4. **Map status values (if you mapped a status column).** Usero builds a default translation from your raw status values (for example `under review`) onto Usero statuses. Adjust any that are wrong. Source values you do not map fall back to **open**. 5. **Preview.** You see the row count, unique authors, unique statuses, the date range, and a count of problem rows. Fix a missing title inline or skip the row. Then choose how duplicates are handled: - **Skip** (default) leaves any row that already exists untouched. - **Merge** updates the existing item. - **Replace** overwrites the existing item. 6. Click **Import**. The file finishes uploading if it has not already, and the import starts. ## What happens after commit - The rows ingest in the background while a progress bar fills. You can reload or share the URL mid-import; it carries the job id and resumes on the progress step. - Each row becomes a feedback item with `source = csv_import`, tagged with the import job so the batch can be undone together. - **Dedup runs in two passes.** Exact repeats inside the same file (matching title, author email, and created-at) are collapsed to one. With the **skip** strategy, rows whose key already exists in this project are dropped so re-running the same import does not double up. - Imported items run through the same AI clustering as the rest of your feedback, so near-duplicates that exact matching missed get grouped semantically once they are in. - When the import finishes you are taken to the dashboard, and the inbox offers an **undo** control for **one hour**. After the hour the items are normal feedback you can edit or delete individually. ## Troubleshooting **"Upload not found; please retry the upload."** The file did not finish uploading before commit. Go back and re-drop the file, then import again. **Rows skipped for "missing title."** Those rows had a blank title column. Map a different column to title, or fix the titles in the preview (or in your source file) and re-import. **Author emails missing on imported items.** Values that do not match a basic email pattern are dropped on purpose, so a name or a handle in the email column will not import as an email. Map the correct column, or leave author email unmapped. **Duplicates still appeared.** Exact dedup keys off title, author email, and created-at. Two rows that describe the same request in different words are treated as distinct on import; the AI clustering groups them afterward in the inbox. **I imported the wrong file.** If it has been under an hour, use the **undo** control in the inbox to reverse the whole batch. After an hour, the items are normal feedback you delete individually. **My export is not CSV.** Usero imports CSV only. Convert a JSON or XML export to CSV in your spreadsheet app first, then import. ## Next - [Integrations overview](/docs/integrations) - [Find your clientId](/docs/find-your-client-id) - [GitHub (AI pull requests)](/docs/integrations/github)