02 Get your data in /Integrations
Files from any tool
Upload an export from any system: support tools, survey platforms, review sites, CRMs and spreadsheets.
Any system that can export a file can feed Dimension Labs. A file is also the route for history and, today, the only route for business data.
Uploads accept CSV, TSV and JSONL, of any size. They don’t accept .xlsx: give an Excel file to the converter or the file checker, which read it directly and write a CSV you can upload. Don’t resave it as CSV from Excel, which rewrites long ids and timestamps in scientific notation and loses digits.
The steps, from export to backfill, are on Upload a file.
Common exports
Section titled “Common exports”| Export | Conversation data (Unstructured data) | Extra columns, or business data (Structured data) |
|---|---|---|
| Support tool ticket export | One row per ticket comment, ticket id as sessionId. If the export has one row per ticket with the whole thread in one cell, see Split a transcript into turns; for HTML bodies, Clean ticket and email text. | Ticket-level fields (priority, CSAT, queue) as extra columns |
| Chat transcript export | One row per message | Channel, queue, tags as extra columns |
| Survey export | Questions and open-ended answers, one conversation per respondent (the converter reshapes it) | Scores as extra columns, or as business data keyed on the respondent id |
| Review export (Google, app stores, review sites) | One row per review | Rating, location, source as extra columns |
| CRM export (accounts, opportunities, renewals) | A business data file under Structured data (Business data format) |
Leave out names, emails and phone numbers, as ids and as extra columns (PII redaction). See Reviews, surveys and tickets for how each maps.
Monthly files
Section titled “Monthly files”If you send the same export every month, save the converter’s mapping (the Save mapping button) and load it next time, or keep a toolkit mapping file. Upload each month’s conversation file to the same source; overlap with the previous month is fine, because duplicates are removed when a file is ingested. Then run a backfill. Keep business data file names stable, because each file name becomes a table name.