Fivetran

This Fivetran review covers what Fivetran is used for, its key features, Fivetran pricing and the monthly active rows (MAR) model, where it shines and falls short, and how it compares with other tools. Fivetran extracts data from SaaS apps, databases, files and other data sources, loads it into a cloud data warehouse or lake, and keeps these data pipelines in sync on a schedule. Schema changes are handled for you, so analysts get fresh tables without writing or maintaining pipeline code. Since June 2026 Fivetran and dbt Labs are one company, which puts extraction, loading and transformation under one roof.

What is Fivetran used for?

Fivetran is a data integration platform used to centralize data for analytics, reporting and AI, from small teams to global enterprise teams. Typical data sources are Salesforce, HubSpot, NetSuite, Google Analytics, ad reporting connectors, Stripe, and production databases such as PostgreSQL, MySQL, SQL Server and Oracle. Typical destinations are Snowflake, BigQuery, Databricks, Redshift and data lakes. Data teams use the synced tables for dashboards in Power BI or Tableau, finance and revenue reporting, marketing attribution and machine-learning features.

The main value is that the data pipelines are fully managed. Fivetran extracts data incrementally, moves only incremental data after the first load, replicates changes from the source database with change data capture, and adapts when a source adds columns or tables. A data engineer who would otherwise build and fix custom data integration scripts for each API can set up a new connection in minutes and spend the time on modelling instead. That trade of subscription cost for engineering time is the core of Fivetran pricing.

Key features

• 700+ connectors: prebuilt data connectors for SaaS apps, databases, files and events, plus a connector SDK for custom sources.
• Incremental syncs: after the first load only changed rows move, every 15 minutes on Standard and down to every minute on Enterprise.
• Database replication: log-based change data capture for large production databases, with enterprise database connectors on higher plans.
• Automated schema handling: new columns and tables appear in the destination without pipeline changes.
• Transformations: dbt Core integration to model data inside the warehouse, billed by monthly model runs.
• Activations: reverse ETL that pushes warehouse data back to 200+ business tools, the former Census product.
• Security: role-based access, SCIM, private networking options, customer managed keys and PCI DSS Level 1 on Business Critical.
• Deployment choice: fully managed SaaS, or hybrid deployment where data stays in your own cloud provider or network.

Fivetran pricing

Fivetran's pricing model is usage-based: you pay for connection MAR, activation MAR and transformation model runs, whichever pricing plan you choose. It is a MAR pricing model: the bill follows how many rows change, not how many users or connectors you have. The details below come from the Fivetran pricing page and usage-based pricing docs as of September 2026.

• Free plan: 500,000 connection MAR, 3,500 activation MAR and 5,000 monthly model runs a month at no cost.
• Standard plan: unlimited users, 15-minute syncs, 700+ connectors, dbt Core integration and role-based access control. Usage-based, with a $5 base charge per connection that uses between 1 and 1 million MAR.
• Enterprise: adds enterprise features such as 1-minute syncs, enterprise database connectors, custom roles, SCIM, multi-cloud and hybrid deployment.
• Business Critical: Business Critical plans add customer managed keys, PCI DSS Level 1 and private networking options.

Transformations are free up to 5,000 successful monthly model runs, then $0.01 per run up to 30,000, $0.007 up to 100,000 and $0.002 beyond. Annual contracts save up to 22% compared with monthly pay-as-you-go, and large customers can negotiate enterprise license agreements. Fivetran does not publish a flat price per million MAR; third-party guides put the Standard paid plan at roughly $500 per million MAR at low volume, falling as data volume grows along each connection's own consumption curve.

The cost surprise to check is how MAR adds up across connections. Under the new pricing model of March 2025, MAR is counted and discounted per connection rather than at account level across the entire account, so many small connections cost more than one large one. On 1 January 2026 three changes followed: the $5 base charge per active connection, deletes counting as paid MAR, and history mode tables billing every changed row (2026 pricing updates). Check your MAR usage per connection in the dashboard before the first invoice, because these changes are the most common source of increased costs. Deleted rows now count as paid MAR, so cleanup jobs in the source raise monthly costs too.

Fivetran pricing explained: how MAR pricing works

Fivetran's MAR-based pricing is built on one unit. Monthly active rows (MAR) are the distinct primary keys that Fivetran inserts, updates or deletes in your destination during a calendar month. A row is counted once per month, however often it changes, so a table that updates the same 10,000 rows every hour costs the same as one that updates them once. Unchanged rows are free. Initial syncs of historical data are free, as are re-syncs caused by adding a column, unchanged rows in re-import tables, and each new connection's 14-day trial.

MAR is tracked separately per account, destination, connection and table, using the distinct primary keys in each table. Re-synced rows beyond the free cases above count again. The same source sent to multiple destinations counts twice, and multiple connections based on the same connector each carry their own base charge. Tables without a primary key, history mode and high-churn event data are the usual reasons a MAR count grows faster than expected. For a rough Fivetran cost, estimate the number of rows that change each month per connection, not the total size of your data volume.

Fivetran cost examples

Small setups with a few low-volume connections often stay inside the free plan or cost a few hundred dollars a month. Estuary, a competitor, estimates 2 million MAR a month at $700 to $2,667 and says 10 million MAR often exceeds $10,000 a month (Estuary). Treat these as rough figures: MAR calculations depend on how often the same rows change, whether you run multiple connectors and whether you replicate data to multiple destinations. Annual contracts or enterprise license agreements lower the unit price for large MAR counts. Critics call the MAR pricing model less transparent than flat-rate tools and say it can create unpredictable costs, because monthly expenses follow data churn rather than a fixed seat or connector count. For stable data volume and few connections, Fivetran pricing can still be cost effective compared with building pipelines in-house.

Is Fivetran an ETL or ELT tool for your data stack?

Fivetran is an ELT data integration tool. It extracts and loads raw data first and leaves transformation to the data warehouse, where dbt models turn it into clean tables. That order keeps the pipelines simple and lets analysts rebuild models without reloading data. Since the merger with dbt Labs, completed on 1 June 2026, dbt Core 2.0 remains open source under the Apache 2.0 licence, and Fivetran positions the combined data stack as the data layer for analytics and AI agents (Fivetran press release).

Where Fivetran shines — and where it falls short

Strengths:

• The widest set of maintained connectors, with schema changes handled automatically.
• Little engineering time: pipelines keep running without custom code.
• Reliable database replication at enterprise scale, with strong security and compliance options.
• One vendor for ingestion, dbt transformations and activations.

Trade-offs:

• MAR based billing leads to unpredictable costs for high-churn data, and high costs at large volume.
• Per-connection tiering and the 2026 base charge raise prices for multi connector setups.
• Limited control over how data is modelled on load; transformation happens afterwards.
• Closed source; self-hosting is only partial through hybrid deployment.

How Fivetran compares

• Boomi Flow: part of the Boomi platform for application integration and workflows rather than warehouse ELT.
• Alteryx: analyst-driven data prep and blending that often runs on top of Fivetran-loaded tables.
• Dataiku: data science and ML platform that consumes warehouse data.
• Make, Zapier and n8n: app-to-app automation for single records, not bulk data replication.
• Power BI: a BI destination for the tables Fivetran loads.

Direct ELT competitors such as Airbyte, Stitch, Matillion and Estuary are not listed on LowCodeDevs yet.

Is Fivetran worth it?

Fivetran is worth it for data teams that want reliable pipelines from many data sources without building them, especially when engineering time costs more than the subscription. The free plan covers small teams and proofs of concept, and the Standard plan fits most mid-sized companies with moderate change volumes. It is harder to justify for high-volume, high-churn data, many small connections, or teams that prefer open-source tools they can host themselves; in those cases Airbyte or a custom pipeline can be more cost efficient. If you want help setting up Fivetran, keeping MAR under control or building dbt models on top, the data agencies and developers on LowCodeDevs can take it on.

Sources

• Fivetran pricing: plans, free plan limits, base charge and transformation rates.
• Usage-based pricing docs: how MAR is counted and which re-syncs are free.
• 2026 pricing updates: base charge, deletes and history mode changes.
• Fivetran + dbt Labs complete merger: merger date, leadership and dbt Core licence.
• Fivetran pricing model, Estuary: cost examples and per-connector tiering (written by a competitor; its claim that minor schema changes raise costs is only partly true, because Fivetran's docs make re-syncs for added columns free).
• Fivetran pricing explained, Weld and Windsor.ai: pricing history and cost estimates (both written by competitors).