Airbyte
Airbyte is our preferred ELT tool — 600+ connectors on a managed cloud, a no-code builder for custom APIs, and an open-source option when you need full control.
What It Is
Airbyte is a data integration platform with 600+ pre-built connectors. It handles extraction from sources (APIs, databases, SaaS tools) and loading into destinations like BigQuery. Our default deployment is Airbyte Cloud — managed, quick to stand up, and easy to maintain. For clients who need full control or have specific compliance requirements, Airbyte also offers a mature open-source self-hosted option.
Why We Chose It
Airbyte's pricing is more predictable at scale than Fivetran's per-connection MAR model, and when sources don't exist out of the box, we can build them — either with the no-code Connector Builder (good for most custom APIs) or the Python CDK for truly complex cases. That flexibility is the main reason we default to Airbyte on new implementations.
How We Use It
Set up Airbyte Cloud as the default deployment — managed, fast to start, easy for clients to maintain
Configure and monitor sync jobs from CRMs, marketing platforms, databases, and internal APIs
Build custom connectors with the no-code Connector Builder for APIs Airbyte doesn't ship by default
Fall back to the Python CDK for truly complex sources (auth flows, pagination quirks, unusual schemas)
Deploy self-hosted Airbyte on GKE for clients who need full infrastructure control
Implement normalized schema conventions that work cleanly with downstream Dataform or dbt models
Airbyte or Fivetran?
Choose Airbyte when:
- You want managed ELT but more predictable pricing than Fivetran at scale
- You need custom connectors — the Connector Builder covers most APIs without code
- You want the option to self-host for compliance or cost reasons
- You prefer a stack where ingestion and transformation come from separate vendors
Choose Fivetran when:
- You need the most polished managed ELT available and cost is secondary
- Your sources are all mainstream SaaS (Salesforce, HubSpot, Stripe, Shopify)
- You're standardizing on the Fivetran + dbt stack and want the vendor consolidation
Choose custom Cloud Run Functions when:
- The source is unusual, volume is small, and neither Airbyte nor Fivetran has a connector
- You already use Cloud Functions heavily and a full ELT tool is overkill for one pipeline