Choose your path

There are two questions: what kind of data is it, and how will it get in.

Conversation data

ChatsCallsTicketsEmailReviewsSurvey answersSocial posts

Data Sources

Unstructured dataRead in full

Enrichment

Dimensions

Business data

RevenueOrdersRenewalsAccountsLocationsUsage

Data Sources

Structured dataStored as is

Not enriched

A table

The agent joins the two on a shared idJoin business data ›

What kind of data is it?

Conversation data is anything a person wrote or said:

• chats with a bot or a human agent
• support tickets and emails
• call transcripts
• reviews (Google, app stores, review sites)
• open-ended survey answers
• social posts and forum threads

It goes under Data Sources > Unstructured data and is enriched into Dimensions: automatically every day for a live integration, and by a backfill for a file upload. Format: Conversation data format.

Business data is anything in plain rows and columns:

• revenue, orders, refunds
• renewals and churn dates
• accounts, plans, tiers
• store or location lists
• product usage
• survey scores without the text

It goes under Data Sources > Structured data, by file upload (an API for it is on the roadmap). It is queryable within about a minute and is never enriched. Format: Business data format.

Most valuable questions need both: what customers said, joined to what they did. See Join business data.

Tip

A survey export usually holds both kinds. The open-ended answers are conversation data. The scores, dates and respondent fields can travel with them as extra columns, or go in as business data keyed on the respondent id.

How will it get in?

Route	Use it when	Start here
Upload a file	You can export a CSV, TSV or JSONL file (or an Excel file, which the converter turns into a CSV). Also the right route for history. Any size: Dimension Labs splits a large file for you.	Upload a file
An integration	You want new conversations to keep arriving from Zendesk, Intercom, Qualtrics, Snowflake or another tool.	Integrations
The API	Your own systems send messages as they happen, to /track. Anything older than an hour goes to /trackhistorical.	When to use the API

Before you upload

1. Export the file from your source system. Avoid opening it in Excel and saving it again: that is the cause of most format problems. If your source gives you an Excel file, keep the .xlsx.
2. If your columns are already named text, userId, incoming, sessionId and dimensionlabs_timestamp, run the file through the file checker. Otherwise, use the converter to map them, then run its output through the checker. Both work in your browser, without uploading anything.
3. Leave out personal data: no emails, phone numbers or names as ids or extra columns (PII redaction).
4. Upload the file, then start a backfill from the Backfill tab.

If your export needs more than renamed columns (a transcript in one cell, HTML bodies, nested JSON), see Transform your data.