Export a complete list of all Stitch connectors, including source type, connection details, destination, replication frequency, and any custom transformations. Note which connectors are in use vs. archived. Create a spreadsheet with columns for source, destination, schedule, and Airbyte equivalent.
⚠️ Watch Out For:
- Archived connectors may not appear in the main list—check the archived section
- Some Stitch connector names differ slightly from Airbyte's (e.g., 'Salesforce' vs. 'Salesforce Account')—plan ahead
Decide whether to use Airbyte Cloud (hosted) or self-host Airbyte (Docker, Kubernetes, or cloud VM). Airbyte Cloud is easier to operate; self-hosted offers cost savings and customization. For a test migration, start with Airbyte Cloud. Document your choice and infrastructure requirements.
⚠️ Watch Out For:
- Self-hosted Airbyte requires Docker/Kubernetes knowledge—budget extra time if you lack it
- Airbyte Cloud has rate limits and data volume caps—verify they match your requirements
For Airbyte Cloud: sign up and create a workspace. For self-hosted: deploy Airbyte (docker run or kubectl apply depending on your environment). Set up database connections (PostgreSQL backend for Airbyte's state management). Configure destinations to match your Stitch warehouse.
⚠️ Watch Out For:
- Self-hosted Airbyte needs a PostgreSQL database for metadata storage—don't skip this
- Network connectivity for self-hosted must allow Airbyte to reach your data sources and destinations
Create destination connections in Airbyte for each warehouse (Snowflake, BigQuery, Redshift, Postgres, etc.). Test the connection with a simple query (e.g., SELECT 1). Verify schema and table naming conventions match your Stitch setup, or configure Airbyte's normalization to match.
⚠️ Watch Out For:
- Airbyte's default schema naming differs from Stitch—use normalization settings to match if needed
- Warehouse credentials and network access rules (firewall, VPN) must be updated for Airbyte's IPs
For each source in your Stitch inventory, create an Airbyte connector. Authenticate with the source credentials. Select tables/streams and columns. Configure sync mode (Full Refresh vs. Incremental). Set replication frequency. Start with a non-critical source as a test.
⚠️ Watch Out For:
- Airbyte's source connector versions can vary—verify you're using the latest stable release
- Some sources (Salesforce, Hubspot) require additional configuration (cursor fields, state)—read the Airbyte docs
Trigger manual syncs for your first connectors. Monitor Airbyte's logs for errors. Compare record counts and sample rows with Stitch to ensure data integrity. Verify that incremental syncs correctly track state (high-water marks). Test with different sync frequencies.
⚠️ Watch Out For:
- Airbyte's incremental sync can behave differently than Stitch—carefully review the first few syncs
- Large initial syncs may take longer in Airbyte if network bandwidth is constrained
Set up schedules for each connector (hourly, daily, weekly). Configure Airbyte's alerting (email, Slack, webhooks) for sync failures. Set up monitoring dashboards to track sync volume, latency, and error rates. Document the schedule for each source.
⚠️ Watch Out For:
- Airbyte's scheduler runs locally—ensure your deployment doesn't restart during scheduled syncs
- Alert thresholds should account for expected variability (e.g., weekend vs. weekday traffic)
Roll out remaining Stitch connectors to Airbyte. Prioritize by criticality and frequency. Batch similar sources together for efficiency. Continue validating data quality. Document any source-specific quirks (custom cursors, filtered schemas, etc.).
⚠️ Watch Out For:
- Custom connectors or obscure sources may not have Airbyte equivalents—plan alternatives
- Rate limits on some APIs (Stripe, Salesforce) may require Airbyte connection pooling tweaks
Keep both Stitch and Airbyte running in parallel for 1-2 full sync cycles. Compare outputs: record counts, schema structures, data accuracy. Flag any discrepancies. This parallel period is critical for confidence before cutover.
⚠️ Watch Out For:
- Timing mismatches between Stitch and Airbyte syncs can complicate comparison—sync them at the same time if possible
- Test downstream pipelines (dbt, dashboards) against Airbyte data to catch integration issues early
Once Airbyte passes validation, disable Stitch syncs. Keep Stitch credentials and backup tables accessible for 2 weeks post-cutover. Update documentation, team runbooks, and alert receivers to point to Airbyte. Remove Stitch from your integrations budget.
⚠️ Watch Out For:
- Don't delete Stitch connectors immediately—archive them for recovery if needed
- Update downstream alerts and SLAs to account for any changes in sync timing or latency
