How to Reduce Fivetran Costs With Sync Frequency and Column Selection
Fivetran bills on monthly active rows (MAR), so every table you sync has a price tag attached. This guide shows which Fivetran settings cut that number without breaking your dashboards.
How Is Fivetran MAR Counted?
A monthly active row is any row inserted, updated, or deleted in a table during the billing month. Re-syncing the same row three times counts as three active rows in many connectors. That detail is why a churn-heavy CRM table can cost more than a much larger static one.
Which Sources Get Expensive Fast?
Event streams, ad platforms, and CRM systems with frequent updates. A single page-views table can hit millions of MAR. By contrast, a product catalog that changes weekly stays cheap even if it’s large, because few rows are active per month.
| Source type | Update pattern | Relative cost |
|---|---|---|
| Web event stream | Per-pageview inserts | High |
| CRM (Salesforce/HubSpot) | Constant updates | Medium-High |
| Product catalog | Weekly changes | Low |
How Do You Lower Sync Frequency Safely?
Most connectors default to frequent syncs. If your dashboard refreshes daily, a 15-minute sync wastes MAR on intermediate states. Drop it to 6 or 24 hours where the business can tolerate it.
- Open the connector and check its sync frequency.
- Ask the reporting owner what freshness they actually need — not what they’d like.
- Lower the frequency to match that requirement.
- Confirm the first few syncs still land before your morning reports run.
This single change often cuts MAR by 30-60% on high-frequency sources.
How Does Column and Table Selection Help?
Connectors let you exclude columns outright, and Fivetran charges less for excluded ones. Wide tables with JSON blobs and audit fields you never query are perfect candidates. Pause entire connectors during a migration instead of deleting them, so historical data stays intact.
What About Schema Drift and Downstream Breaks?
Fivetran adapts to new columns automatically, which is convenient until a renamed source field silently nulls a dbt model. Set up schema-change alerts and review them. Cheap isn’t cheap if your revenue dashboard goes blank.
When Is Fivetran the Wrong Choice on Cost?
If you sync one high-volume Postgres database into the same warehouse, self-hosted tools like Airbyte or a hand-written CDC job cost less in raw dollars — you pay in maintenance instead. Fivetran’s pricing makes sense when you run many sources and value not babysitting them.
Quick Summary
- MAR counts every insert, update, and delete — not unique rows.
- Event streams and frequently-updated CRMs drive most of the bill.
- Lowering sync frequency from 15 min to 6-24 hours can cut MAR 30-60%.
- Exclude unused columns and pause idle connectors.
- Watch schema-drift alerts so cost savings don’t break reporting.
