Reverse ETL: Database to Events

Merge database records into your event pipeline with synthetic events — your way, with YAML.

What is Reverse ETL?

Reverse ETL connects your database systems to your CDP, transforming database records into event-like streams that complement your real-time event capture.

Whether you're working with orders from your MySQL database, users from your Postgres system, or product catalogs from your SQL Server — Growcado lets you map these to CDP events with simple YAML configuration.

🔄 Why it matters

• Unlock historical data that existed before your tracking implementation

• Enrich your customer profiles with data from systems that don't generate events

• Maintain consistency across all customer touchpoints and systems

• Stitch together fragmented identities by connecting database records to known users

No engineers or complex ETL pipelines needed. Just clean YAML and your database connection.

Config: Defining Synthetic Event Sources

In your YAML config, you'll define which database tables should be mapped to events, how to identify users, and what properties to include.

Here's an example:

synthetic_event_mappings:
  orders_to_completed_events:
    enabled: true
    source:
      schema: publuc
      table: store_mysql__order
      primary_key: order_id
      timestamp_field: order_cdt
    event:
      type: OrderCompleted
    condition: "order_status > 0"
    identity:
      user_id: user_id_fk
    traits:
      mobile_number: mobile

🛠️ How it works:

• Source: Tell Growcado where to find your data (schema, table, keys)

• Event: Define what event type this data should become

• Condition: Only generate events for records that match your criteria

• Identity: Map database columns to customer identifiers

• Traits: Extract fields that should become customer traits

• Properties: Control how database fields map to event properties

Example in action

Let's walk through a real-world example of mapping order data to synthetic events:

1. You have a MySQL orders table with customer purchases

2. You want these orders to appear as OrderCompleted events in your CDP

3. Each order has line items stored in a JSON column

Growcado will:

1. Connect to your database and read the orders table

2. Convert each order to a synthetic event

3. Map all the identifiers to the correct profiles

4. Transform line items into properly structured product arrays

5. Make these events indistinguishable from real tracked events

🔍 Advanced Configuration

Property Mapping

For each database column, you can control exactly how it appears in your events:

properties:
  order_id: 
    target: order_id
    type: string
  order_total_price: 
    target: total
    type: numeric
  order_shipping_price: 
    target: shipping
    type: numeric

Handling Complex Data

Work with nested JSON data using JSON mappings:

order_item_json: 
  target: products
  type: json_array
  json_mappings:
    - source_key: item_id
      target_key: product_id
      type: numeric
    - source_key: item_name
      target_key: name
      type: string
    - source_key: category_name
      target_key: category
      type: string

Deduplication

Prevent duplicate events when processing the same data multiple times:

deduplication:
  enabled: true
  match_keys:
    - source: order_id
      target: property_order_id

Adding Constants

Add fixed values to every generated event:

constants:
  currency: USD

Key Features

• Bidirectional identity resolution — connects database records to existing profiles

• Delta processing — only processes new or changed records

• Full history import — backfill your CDP with historical data

• Custom transforms — advanced property mapping and typing

• Schema enforcement — ensures data consistency across sources

• Just-in-time generation — creates events only when needed

📊 Where Synthetic Events Live

Once processed, synthetic events become indistinguishable from regular tracked events in your warehouse:

cdp_events table structure:

event_id	event_type	user_id	timestamp	properties
evt_a9f2...	OrderCompleted	user_789	2023-05-14T15:32:11	{"order_id":"12345","total":59.99,...}
evt_b3c1...	PageViewed	user_789	2023-05-14T15:30:22	{"page_path":"/checkout",...}
evt_d7e4...	OrderCompleted	user_456	2023-05-14T14:18:05	{"order_id":"12344","total":29.99,...}

The only difference? A metadata field (context_source: synthetic) that helps you trace data lineage when needed.

All computed traits, identity resolution, and segment memberships apply equally to both synthetic and regular events.