Nightly Stripe Export to Snowflake Warehouse

This workflow runs on a nightly cron and lands your Stripe revenue data in Snowflake without manual exports. It pulls every charge, refund, and dispute created in the last 24 hours from the Stripe API, writes the raw payloads to a Postgres staging table for an audit trail, normalizes the records (currency, amounts in major units, customer and invoice IDs, fee breakdowns), and upserts them into a `fct_payments` table in Snowflake keyed on the Stripe object ID so reruns never double-count.

Use this when finance or analytics teams need fresh Stripe numbers in the warehouse each morning but you don't want to pay for a managed ELT connector or babysit a brittle script. It fits teams that already run dbt or BI on Snowflake and want a single, idempotent revenue table. The Postgres staging step gives you a replayable landing zone, so if a Snowflake load fails you can re-upsert from staging without re-hitting the Stripe API or tripping rate limits.

A schedule trigger fires at 02:00 in your timezone and computes the prior day's window. The workflow queries Stripe for charges, refunds, and disputes in that range, paginating until exhausted. Raw rows are inserted into a Postgres `stg_stripe_events` table tagged with the run timestamp. A transform step flattens nested objects and converts amounts from cents to decimal. Finally a Snowflake MERGE upserts the cleaned rows into `analytics.fct_payments` on the object ID, inserting new payments and updating status changes (e.g. a charge that later refunded), then emits a row count summary for the run.