This repository contains the production dbt transformation layer for Sylvan’s subscription, deal, and invoicing analytics.
It runs on Snowflake, ingests data from HubSpot and Airtable via Fivetran, and produces analytics-ready tables used across Finance, RevOps, and Customer Success.

This is a live, deployed production project. Changes flow through GitHub and are deployed via CI only.

• Warehouse: Snowflake
• Transformation: dbt (v1.11.0-rc1)
• Ingestion: Fivetran (HubSpot + Airtable)
• CI/CD: GitHub Actions

Materialization: table
Purpose: Canonical subscription fact table for analytics and reporting.

Key characteristics:

• Built from HubSpot staging models (stg_deal, stg_company, etc.)
• Enriched with invoice data from Airtable
• Join key:
  airtable.INVOICES.associated_hubspot_deal_id → hubspot.deal_id

Important logic constraints (intentional):

• contract_start_date and contract_end_date do not exist
• Subscription timing is derived only from:
  • booking_date
  • contract_term_months
  • free_trial_duration
  • free_trial_placement
• next_invoice_to_raise:
  • Returns NULL when all invoices are raised
  • Never returns placeholder strings (e.g. "Complete")

• HubSpot: via Fivetran staging models
• Airtable:
  • Source: airtable_sylvan_customer_management
  • Table: INVOICES

• Database: SYLVAN_INTERNAL_DATABASE
• Schema: SUBSCRIPTION_MGMT_DBT
• Warehouse: SYLVAN_INTERNAL
• Role: DBT_SYLVAN_ROLE

All production runs happen via GitHub Actions.

1. Go to GitHub → Actions
2. Select the dbt workflow
3. Manually trigger the run
4. CI executes:
   • dbt build
   • dbt test

No manual dbt runs are performed directly against production.

• Some source tests are expected to fail and are currently accepted.
• This repository has transitioned from tests-only to a full transformation project.
• Treat all changes as production-impacting.

This is a production analytics repository. Changes should be intentional, minimal, and reviewable.

Do not:

• Re-architect existing models without alignment
• Invent new fields or business logic
• Run dbt directly in production outside CI

git checkout -b feature/<short-description>

2. Edit Models
	•	Update models under models/
	•	Follow existing patterns and naming conventions
	•	Preserve existing business logic unless explicitly changing it

3. Test Locally

Before opening a PR: dbt build
Focus on:
	•	Model compilation
	•	Downstream impact
	•	Tests relevant to your changes

⸻

Pull Requests

PRs should include:
	•	Clear description of what changed and why
	•	Callouts for any logic changes
	•	Confirmation that dbt build was run locally

Avoid large, unfocused PRs.

⸻

Deployment

Deployment is handled only through GitHub Actions.

To deploy:
	1.	Merge your PR into the main branch
	2.	Go to GitHub → Actions
	3.	Manually trigger the dbt workflow

This will rebuild and test production tables in Snowflake.

⸻

Rollbacks

Rollback = revert the commit and re-run the GitHub Action.
There is no manual intervention in Snowflake.