How does Fivetran count a row?
Per Fivetran's usage-based pricing documentation, MAR is the number of distinct rows synced to your destination in a calendar month, tracked by primary key.
- Once a month: a row that syncs 30 times in a month counts as one MAR. Fivetran contrasts this with row-based models that would charge 30 times.
- By primary key: keys are counted separately per account, destination, connection, table, activation and activation sync. Without a primary key, Fivetran creates a hashed one.
- Incremental: most connections sync only new, updated or deleted rows, so you pay for the subset that changed.
- Historical rows: if you edit years-old records constantly, those rows are re-imported and count, so a busy source costs more than a quiet one.
From Fivetran's usage-based pricing documentation, reviewed November 21, 2026.
What is free and what counts?
Fivetran splits MAR into free, historical and incremental types.
Fivetran says some re-syncs are free under the pricing it launched on March 1, 2025: historical data from adding a column, unchanged rows in re-import tables, and post-trial MAR already seen in the trial. Annual contracts signed before that date move to the latest pricing on renewal.
| Type | What Fivetran says | Examples it lists |
|---|---|---|
| Free MAR | Rows synced that do not count toward paid MAR | Initial syncs; connection trials; Fivetran system tables; the Fivetran Platform connector; the Free plan; promotions; private preview connectors |
| Historical MAR | Data that already existed in the source or was synced when the table or connection was created | Initial table syncs; full re-syncs after a schema change; manual re-syncs; migration syncs |
| Incremental MAR | New or updated data in the source since the last sync | Overlapping incremental syncs; changes in re-imported tables; rollback syncs in reporting connectors |
What raises MAR without you noticing?
Fivetran's documentation names several cases worth checking.
Fivetran also documents how to monitor and optimize usage in its own dashboard. See the pricing guide for how rows turn into dollars.
- Duplicate connections: it is the connection that matters, not the connector. Two Salesforce connections, one to staging and one to production, count their MAR separately.
- Tables without a primary key: Fivetran hashes columns to make one, and each row in such a table counts toward MAR; changing the columns it hashes affects MAR.
- Rollback syncs: some connectors re-fetch the previous month, and those rows count as new in the current month.
- Phantom updates: a connection made without an initial sync can see updates to rows that are not in the destination, and these count.
- Null values: a column that syncs null and later changes counts; a column that had no value at all and gets one does not.
Frequently asked questions
What does MAR stand for at Fivetran?
Monthly active rows: the number of distinct rows, tracked by primary key, that Fivetran syncs from a source in a calendar month.
Does an update to the same row count twice?
No. Fivetran says a row counts once per month no matter how many times it updates that month.
Is the first sync free?
Fivetran lists initial syncs as free MAR, though tracked incremental updates during that time may count toward paid MAR outside the 14-day connection trial.
Do two connections to the same source cost double?
Fivetran says each connection's rows are counted separately, even when two connections sync the same data with the same primary keys.
Can you reduce our MAR?
Often. We review your connections, blocked tables and sync settings for the cases Fivetran documents. Book a call to start.
Need help with Fivetran?
We help small and mid-size teams choose, cost and run data tools. See our Fivetran guides and services, and the Snowflake guides, and the Power BI guides, and the Looker Studio guides, and the AI implementation.






