Customers Drop High-Value Fivetran Connectors
Fivetran: the $200M/yr Zapier of ETL
This kind of churn hits Fivetran revenue early, because the first connectors customers replace are usually the ones generating the most rows and the most billable usage. A payments, CRM, or marketing connector can run constantly and push huge tables into Snowflake or BigQuery. When Stripe or Salesforce offers that export natively, a customer can keep Fivetran for the long tail, but remove the most expensive syncs first and immediately shrink spend.
-
Fivetran charges on data volume, so revenue is concentrated in busy connectors, not in the raw count of connectors. Its model benefits when customers sync large, fast changing systems like payments, CRM, and production databases, and gets hurt most when those systems move off platform.
-
Native connectors are strongest exactly where Fivetran is most exposed. The source app already owns the schema, API roadmap, and customer workflow, so it can export data more reliably and often monetize it as a premium feature, as seen with vendors like Stripe, Salesforce, Segment, and Customer.io.
-
This is why Fivetran does not disappear when native exports arrive. Teams still need one tool for hundreds of smaller apps and internal databases, and Fivetran still wins on managed reliability. But the remaining footprint shifts toward lower volume, less strategic connectors, which are less lucrative per account.
The next phase is a barbell market. Big SaaS platforms will keep pulling their own high volume data flows in house, while horizontal ETL vendors defend the messy long tail and database replication layers. That pushes Fivetran to lean further into enterprise data movement, governance, and broader pipeline ownership, where replacing a single connector does not cut spend as easily.