Score customer churn risk and trigger CS follow-ups with Postgres, Mixpanel, Zendesk, Gmail, Calendly, ClickUp, and Slack

Description

Quick overview
This workflow runs every 6 hours to combine customer usage and NPS data from Postgres with Mixpanel engagement and Zendesk ticket signals, calculate a churn risk score, and trigger Customer Success interventions via Gmail, Calendly, ClickUp, Slack, and a Postgres intervention log.

How it works
Runs every 6 hours on a schedule trigger.
Pulls recent product usage and last-login data from Postgres, fetches engagement data from Mixpanel, fetches recent NPS responses from Postgres, and retrieves recent tickets from Zendesk.
Merges the four data sources into a single per-customer record keyed by customer_id.
Calculates a composite churn score and risk level (low/medium/high) based on login recency, feature usage change, seat utilization, NPS score, and ticket signals.
Skips customers below the churn-risk threshold and routes at-risk customers to intervention steps.
Sends a personalized check-in email via Gmail for churn-risk customers, and for high-risk customers also creates a Calendly scheduled event, opens a follow-up task in ClickUp, and posts an escalation message to a Slack channel.
Logs the intervention details back to Postgres.

Setup
Add Postgres credentials and update the SQL queries and table/column names for usage snapshots, NPS responses, and the churn_intervention_log insert.
Add Mixpanel API credentials and replace YOUR_MIXPANEL_PROJECT_ID in the Mixpanel request body.
Add Zendesk credentials and set any required account/subdomain settings to retrieve the right ticket data.
Add Gmail OAuth2 credentials and ensure your customer records provide contact_email and (optionally) contact_first_name.
Add Calendly API credentials and replace YOUR_CS_CALL_EVENT_TYPE_ID (or adjust the step if you only want to include a booking link in the email).
Add ClickUp and Slack credentials, then replace YOUR_CS_LIST_ID, set the target Slack channel and ACCOUNT_MANAGER_SLACK_ID, and adjust churn score thresholds in the scoring logic as needed.