Import your backlog
If you already have feedback somewhere else, bring it with you. Open a project's Settings → Import, upload the export file, and check the preview before anything is written.
Live connections
A file import is a snapshot. If you run Zendesk, Intercom or Slack day to day, connect them instead and feedback arrives as it happens — Settings → Sources.
Set a filter. A busy helpdesk pushes every ticket, and importing a support queue as product feedback buries the real signal under password resets. Tag the ones you want — a Zendesk ticket tag, an Intercom conversation tag, or a Slack reaction emoji — and only those become items. Leave it blank and everything comes through, which the settings page warns you about.
Connecting gives you a webhook URL and a signing secret (shown once). Point the tool at the URL, send your project id in X-Project-Id, and sign the raw body with HMAC-SHA256 into X-Fdbck-Signature. Redelivery is safe: each item carries its source id, so a retried webhook never creates a second copy.
Like file imports, connected items land un-triaged. A busy support day never spends your agent budget on its own.
What you can upload
| Tool | Format | Notes |
|---|---|---|
| UserVoice | CSV | Ideas export. Uses Voters Count for demand, not the legacy Votes column. |
| Canny | CSV | Posts export. Reads Score (Total Votes). |
| Productboard | CSV | Notes export. HTML in note_text is stripped to plain text. |
| Jira | CSV | Issues export. Summary and Description; dd/MMM/yy dates are handled. |
| Zendesk | JSON | CSV has no ticket text — see below. Export JSON instead. |
| Anything else | CSV | Any file with a title and a body column; you pick which is which. |
Up to 10,000 rows and 20MB per upload. Split a larger export and import it in parts.
Nothing runs automatically
Imported items land in your inbox un-triaged. That is deliberate: triaging a few thousand items the moment you click upload would spend a few thousand runs on your own agent account without you choosing it. Triage the ones you care about from the inbox.
Vote counts come across as demand, so a request 200 people asked for arrives looking important instead of flattening to one. Original dates are kept.
If we don't recognise the file
fdbck detects the common exports from their column names. If yours isn't one of them, the preview shows the columns it found and asks which holds the title and which holds the feedback text. Nothing is guessed.
The simplest file that always works is a CSV with these headers:
title,body,votes,status "Dark mode","Please add a dark theme",12,Planned "Faster search","Search takes 8 seconds",7,Open
Zendesk needs JSON, not CSV
Zendesk's CSV export does not contain ticket text. Descriptions and comments are excluded from every CSV path — the account export, view exports, and Explore reports alike — so an import would create tickets with nothing in them. fdbck refuses that file rather than filling your inbox with empty rows.
Export as JSON instead (Admin Center → Account → Tools → Reports → Export) and upload that. Two things worth knowing: exports are off by default on new Zendesk accounts and have to be enabled by their support, and JSON export requires Suite Growth or Support Professional and above.
Importing twice is safe
Rows carry their original id from the source tool, so re-uploading the same export does not create a second copy — the preview will tell you how many were already imported. Every import is recorded as a batch, so a bad one can be identified rather than hunted for by hand.