Importing Feedback

A .csv file, comma-separated rather than semicolons, UTF-8, with a header row. Then Settings → Integrations → Import CSV and upload it.

Annsa auto-detects the usual names:

• Feedback text: feedback, text, message, content
• Customer name: name, customer_name, first_name, surname
• Customer email: email, customer_email
• Revenue: revenue_band, tier, plan, sku

Use the column mapper if yours differ.

Over 10,000 rows an import can take several minutes, with a progress bar throughout. Splitting into files under 10,000 rows is more reliable.

Connect Google Drive first — Sheets uses the same connection. Then Settings → Integrations → Google Sheets, pick the spreadsheet and sheet, preview, map and import. Good for NPS responses, survey data and support exports.

On Priorities, + Add → Add Feedback. Paste the text, optionally add a customer name, email and revenue (what they pay you, as a number).

Authenticate with your API key from Settings → Integrations → Survey. Duplicates are skipped by text hash, same as CSV.

POST /feedback
{
  "text": "The export is too slow on large datasets",
  "customer_name": "Jane Smith",
  "customer_email": "jane@example.com",
  "revenue_band": 199,
  "source": "api"
}

Or a batch:

POST /feedback
{
  "items": [
    { "text": "...", "customer_email": "..." },
    { "text": "...", "customer_name": "..." }
  ]
}

See the API article for the full reference.

Anyone on the team can add Slack, GitHub, Google Sheets or a CSV from inside Ask — Annsa checks the column mapping first and stops rather than let a single row land wrong (a CSV opens straight into the import wizard).

Click Undo on the completion screen and every item from that import is removed. It applies to the most recent upload only, and priorities and specs recalculate automatically.

Bring it into one pool, then let it group itself. Annsa classifies every piece by intent, urgency and sentiment, folds different wordings of the same problem into one theme, and drops duplicates. You don't sort it into folders — you read the ranked themes it produces.

Put a survey where people already are: paste one embed before </body> and responses arrive in real time. Beyond that, connect Slack for the conversations you already have, import a CSV or Google Sheet of what you've collected, or post to the API.

Pick at least one source and let it run. Slack and in-product surveys collect continuously once connected; CSV, Google Sheets and transcripts bring in what you already have; the API and MCP take programmatic submissions. Roughly 10 pieces is where the ranking starts to mean something.