← Back to Glossary

Fivetran

Fivetran is a managed, fully automated data integration (ELT) platform that uses pre-built connectors to replicate data from source systems into a data warehouse or lake.

Overview

Fivetran is a commercial, fully managed data movement platform built around the ELT (extract, load, transform) pattern: data is extracted from a source and loaded into a destination in as close to its raw form as possible, with transformation happening afterward inside the destination. Its core product is a large catalog of pre-built, maintained connectors for SaaS applications, databases, event streams, and files. Instead of writing and maintaining custom extraction scripts, teams configure a connector, point it at a destination, and Fivetran handles scheduling, schema detection, schema drift, and incremental syncing on an ongoing basis.

How it works

  • Connector: a managed integration for a specific source (e.g., Salesforce, Postgres, Stripe). Fivetran owns and updates the connector as source APIs change.
  • Sync: a scheduled run that pulls new or changed data since the last run, using each source's most reliable change-detection method (log-based CDC for databases where supported, API cursors/webhooks for SaaS sources).
  • Destination: the warehouse or lake the data lands in. Fivetran automatically creates and evolves destination schemas to match the source, handling nested and semi-structured data.
  • Transformations: Fivetran can trigger downstream transformation jobs, including dbt projects, after a sync completes, but the extraction and loading logic itself is not user-written code.

Why it matters

Fivetran's value proposition is reducing the engineering time spent building and babysitting brittle extraction pipelines, in exchange for a usage-based subscription (commonly priced on volume of rows/records synced, referred to as monthly active rows). This tradeoff is attractive for common, well-supported sources where a custom connector would mostly reinvent what Fivetran already maintains, and less attractive for highly custom or low-volume sources where the cost of a managed connector may exceed a simple custom script.

A typical modern stack pattern is Fivetran syncing raw data into a cloud warehouse or a lake of Parquet files, with a query engine like DuckDB or MotherDuck used downstream to run transformations or ad hoc analysis, often orchestrated by a tool like Airflow or dbt Cloud that kicks off after a Fivetran sync finishes.

Related terms

FAQS

No. Fivetran is a proprietary, commercial SaaS platform; it is not open source and does not have a self-hosted free tier equivalent to open-source alternatives like Airbyte.

Fivetran's core job is extract and load; transformation is expected to happen in the destination, commonly via a triggered dbt project, rather than inside Fivetran's own connectors.

Fivetran is generally priced on the volume of data synced (monthly active rows), rather than a flat per-connector or per-seat fee, though specific pricing plans vary.