Integration

Connect Databricks to your
customer knowledge graph

Connect Databricks to your customer knowledge graph

Overview

If your product usage, billing, or support data lives in a Databricks lakehouse, it belongs in your customer knowledge graph. Today you can connect a Databricks SQL warehouse directly to Sonora and pull structured customer data straight from your tables — no exports, no scripts, no waiting on data eng for a one-off pull.

What syncs

Connecting Databricks lets you bring in any table that carries a customer identifier, and resolves each row to the right account automatically:

  • Account and usage tables: any table keyed by customer ID, account ID, or domain, matched to a Sonora customer
  • Product usage metrics: feature adoption, login frequency, seat counts, or whatever your product team already models in the lakehouse
  • Billing and revenue data: ARR, invoice history, and contract data pulled straight from your finance tables
  • Support and ticket volume: if your support data lands in Databricks before it reaches a BI tool, it can land in Sonora too
  • Meetings, issues, and custom fields: anything else your team has modeled, mapped during setup and queried in chat and Canvas

Everything lands on the same unified customer record as your CRM and support tool data, so chat, Canvas, and playbooks can reason across product usage and lifecycle stage in one place instead of two.

Two ways to bring in accounts

You choose how Databricks rows relate to your customers during setup:

  • Create customers from the warehouse — your lakehouse account table becomes the source of truth, and Sonora creates a customer for each row.
  • Attach to existing customers — Databricks rows match customers you already sync from Salesforce or HubSpot by account ID, and the warehouse enriches them without ever creating or overwriting a record.

Secure, read-only connection

We connect using a Databricks service principal over OAuth 2.0 — not a personal access token — scoped with a read-only SQL role to only the schemas you approve. No credentials are stored in plaintext, and Sonora never writes back to your lakehouse.

Set it up once, point us at a workspace and SQL warehouse, and Sonora proposes a mapping from your tables. After that, syncs run continuously, pulling only new and changed rows.

Custom field mapping

Map any column to the attributes Sonora tracks. During setup, you pick the join key (customer ID, account ID, or domain) and then map the columns that matter:

  • ARR and usage tier
  • Renewal and contract dates
  • Product usage fields (whatever your team already models)
  • Health score inputs, if you’re calculating one upstream

You can generate the mapping straight from a live connection, or upload a schema (CSV, Excel, JSON, YAML, or Markdown) and let Sonora draft it for you. A dry run validates that every mapping resolves before the first sync starts. Anything Sonora doesn’t track natively can be added as a custom field and shows up in chat and Canvas alongside the rest of your customer data.

Get started

Head to Settings → Integrations → Databricks and connect your warehouse with a service principal. For the full setup walkthrough, SQL grants, and column reference, see the Databricks integration docs.