Editor’s note: We combine independent analysis, data collection, and hands-on testing to review data and AI tools. This Fivetran review weighs the areas that matter most when you pay by usage: pricing mechanics, connector coverage, reliability, and support.
Quick verdict: We recommend Fivetran for teams standardizing on a cloud warehouse who value engineering time over software spend, and for enterprises that need Oracle, SAP, or Db2 replication without a dedicated pipelines engineer. Go in with eyes open on one thing: MAR-based consumption pricing that resists forecasting, which makes the free 500,000 MAR tier the right first step. The June 2026 merger with dbt Labs folded transformations into the same platform that runs your pipelines.
In this review, I’ll take a closer look at Fivetran’s pricing mechanics, connector catalog, and what the dbt merger changes, so you can see exactly whether it fits your stack.
Key Takeaways 🔍
- Fivetran is fully managed ELT with 700+ connectors, and it handles schema drift automatically, so pipelines rarely need babysitting
- The company merged with dbt Labs on June 1, 2026, which makes dbt transformations a first-class part of the platform rather than an external dependency
- MAR-based pricing is genuinely hard to forecast: billing went per connection in March 2025, and a 2026 update added a $5 base charge per Standard plan connection plus billable deletes
- The free tier is real: 500,000 monthly active rows, 15-minute syncs, and initial historical syncs never count against your total
- Users report charges that are hard to trace back to a cause and connectors that resist customization when a source object is not supported
Pros and Cons
Pros
- 700+ fully managed connectors, including Oracle, SAP, and Db2
- Initial historical syncs are free, so backfilling years of data costs nothing
- Automatic schema drift handling, with no manual mapping step
- Change data capture moves only changed rows after the first sync
- dbt transformations are in-house since the June 2026 merger
- Usable free tier: 500,000 MAR and 5,000 transformation runs a month
Cons
- MAR billing resists forecasting, and volume discounts cannot pool across connections
- $5 base charge on every Standard plan connection between 1 and 1M MAR
- Connections, transformations, and Activations bill on three separate meters
- Managed connectors cannot be bent to pull unsupported objects
- PCI DSS Level 1, customer-managed keys, and private networking require Business Critical
How Much Does Fivetran Cost?
Fivetran’s pricing is split between four plans, from a free tier capped at 500,000 monthly active rows to Business Critical at $1,067 per million, and every dollar flows from a consumption meter you need to understand before you commit.
Fivetran bills on Monthly Active Rows, or MAR: the number of distinct rows inserted, updated, or deleted in a calendar month, tracked by primary key. A row updated ten times in one month still counts once. Initial historical syncs are free, which makes onboarding painless.
Since the 2026 pricing update, deletes are billable too, along with the optional history mode. The formula itself is base spend plus an incremental rate for each million MAR above your threshold, with partial millions prorated.
Free ($0): for measuring your real MAR before committing. Standard ($500 per million MAR): for most teams going paid. Enterprise ($667): for 1-minute syncs and custom roles. Business Critical ($1,067): for regulated industries.
| Plan | Price | Sync frequency | Notable inclusions |
|---|---|---|---|
| Free | $0, up to 500,000 MAR/month | Every 15 minutes | 5,000 free transformation runs |
| Standard | $500 per million MAR at the 1M threshold | Every 15 minutes | Core managed ELT, all standard connectors |
| Enterprise | $667 per million MAR at the 1M threshold | Every 1 minute | Custom roles, VPN tunnels on annual contracts, choice of cloud provider |
| Business Critical | $1,067 per million MAR at the 1M threshold | Every 1 minute | PCI DSS Level 1, customer-managed encryption keys, private networking |
Per-million pricing falls as volume grows within a connection, so large accounts pay less per million than the threshold figures, and annual contracts discount up to 22%, starting at 5% for smaller commitments.
A typical Facebook Ads connection at 34,479 MAR runs $22.06 per month, and a Marketo connection at 847,574 MAR runs $424.54. An Activations pipeline pushing warehouse data to Salesforce at around 12,007 MAR costs $202.25 per month on its own meter.
Small, stable connections really are cheap. The pain arrives with volume and connection count.
In March 2025, billing moved to a per-connection basis for new contracts; existing annual contracts keep their terms until renewal.
A 2026 update put a $5 base charge on every Standard plan connection syncing between 1 and 1M MAR, and made deletes plus history mode billable. Activations, Fivetran’s reverse ETL product, arrived in 2026 as well, metered separately from your ingestion connections.
Transformations run on their own meter as well: the first 5,000 model runs each month are free, then pricing tiers down from $0.01 per run toward $0.002 at high volume.
Is Fivetran Good Value for Money?
- Good value if your row churn is modest and stable: a handful of connections with predictable volumes lands well under the cost of an engineer maintaining pipelines
- Painful for high-churn data: event streams and frequently updated tables drive MAR up fast, and nested JSON can flatten one source record into many billable rows
- Painful for connector sprawl: with 20+ connections, the $5 bases and the inability to pool volume discounts stack against you
- The free tier removes the guesswork risk: you can measure real MAR before a dollar is committed
- dbt shops get extra mileage: 5,000 free transformation runs cover small projects, and one vendor for ingestion plus transformation is one fewer contract to manage
My recommendation: start on the Free plan, run your actual sources for a month, and size your true MAR from the usage dashboard. Move to Standard when you cross 500,000 MAR. Only buy Enterprise if you need 1-minute syncs, custom roles, or VPN tunnels, and treat Business Critical as a compliance purchase rather than an upgrade path.
My Experience With Fivetran
Signup asks for an email, a few optional background questions about your role and objective, and nothing else up front. I started on the free plan with the goal of getting one real source flowing into a warehouse.
The setup pattern is the same for every source. You pick a connector from the catalog, authorize access to the source, choose a destination and a schema name, and start the initial sync. If you do not have a warehouse yet, Fivetran offers a free Snowflake trial account as the destination, which removes the biggest blocker for a team evaluating from scratch.
That first historical sync runs free regardless of size, and the 14-day trial that follows unlocks Business Critical features, so you evaluate the platform at its best tier before paying for any of it. You can add transformations, more connections, and team members during the trial, before or after that first sync completes.
The dashboard shows connection and destination status, recent activity, a sync history chart, and spend and usage metrics, and about a week into the trial the billing page produces a usage estimate you can plan around.
Each connection carries a Setup tab for testing the connection, changing sync frequency, adjusting schema-change behavior, and triggering a historical re-sync. Connections can be paused, resumed, or manually re-synced from the same view, and email alerts are configurable per connection as well as in your account settings. When a sync fails, no data is lost; the dashboard flags it, and the sync resumes from the failure point once the underlying issue is fixed.
What the dashboard will not do is make the bill self-explanatory. Users report charges they struggle to trace back to a cause, and with three meters accruing at once, the spend view needs regular attention rather than an occasional glance.
Since the merger, dbt jobs run from the same platform that manages ingestion, which collapses what used to be two vendors into one screen.
Connectors and the Sync Engine
Fivetran lists 700+ connectors spanning SaaS applications, databases, files, and event streams, and the count includes the enterprise sources that thin out rival catalogs: Oracle, SAP, and Db2.
Connectors come in four types:
- Standard connectors: fully managed integrations for common SaaS, ERP, database, event, and file sources
- Lite connectors: lighter builds for niche SaaS tools, with less depth than the standard tier
- Partner-built connectors: maintained by technology partners rather than Fivetran itself
- Connector SDK: code templates for building your own when nothing above fits
On the other side of the pipeline, 200+ destinations cover the major warehouses and lakes, and Activations reuses that reach to push data back out into business applications.
Database replication uses change data capture, so after the initial load only changed rows move, which keeps both latency and MAR consumption down. Schema drift handling is the maintenance win: new columns and renamed fields propagate to the destination without anyone editing a mapping.
Managed connectors resist customization by design, and users report limited options when a source object or a complex transformation falls outside what the connector supports. Depth also varies across the catalog: lite connectors cover fewer objects than their standard counterparts, so check the specific tables you need before assuming parity.
Sync frequency is 15 minutes on Free and Standard and 1 minute on Enterprise and above, which matters if your use case is operational rather than analytical.
Transformations After the dbt Labs Merger
Fivetran completed its merger with dbt Labs on June 1, 2026, under the joint positioning of building the data infrastructure for trusted AI agents. By September 16, 2026, the combined company was shipping joint capabilities at dbt Summit 2026 aimed at making enterprise data agent-ready.
Practically, the merger means dbt is no longer an external dependency you glue on. Fivetran’s transformation model was already post-load: data lands in the destination first, then dbt-compatible models transform it in place. Now the company that runs those models is the company that loaded the data.
The billing stays separate, though. Transformation runs accrue on their own meter, 5,000 free per month and per-run pricing after that, so a heavy dbt project adds a line to the bill that ingestion pricing never mentions.
For AI infrastructure, Fivetran is how structured data reaches the warehouse your BI and AI workloads read from, whether that is RAG over warehouse tables or agents that need governed context. The company calls this its Context Layer pitch, and it has at least one named proof point: Oakbrook Finance selected Fivetran in June 2026 to support real-time lending decisions on AI-ready data.
Fivetran also ships dbt Wizard, an AI agent for working with dbt projects, and in May 2026 became steward of Great Expectations, the open source data-quality project.
Security and Compliance
Fivetran holds SOC 2 Type II, SOC 1 Type II, ISO 27001, ISO 27701, and PCI DSS Level 1, aligns with GDPR, and is certified under the EU-US Data Privacy Framework. HITRUST i1 certification supports HIPAA workloads.
That org-level posture is as strong as anything in the category, and it is a real reason enterprises pick Fivetran over open-source alternatives they would have to certify themselves.
Even observability is tiered. Sync and platform logs can land in your warehouse for free through the Fivetran Platform Connector, but shipping them to external services like Datadog, Splunk, or CloudWatch requires Enterprise or Business Critical.
The strongest controls are gated by plan. Custom roles, VPN tunnels, and choice of cloud provider arrive at Enterprise. PCI DSS Level 1 coverage, customer-managed encryption keys, and private networking are listed only under Business Critical, the plan billing $1,067 per million MAR at the 1M threshold.
If you are a regulated-industry buyer, read that as the real price list: the controls your auditors treat as table stakes sit behind the most expensive tier, and no lower plan gets you there.
How Does Fivetran Compare to Competitors?
My verdict holds against the field: nothing matches Fivetran’s combination of connector breadth and zero-maintenance operation. But every rival wins a niche, and some of those niches might be yours.
- Airbyte wins on open-source control: the core platform is free to self-host, you can build custom connectors, and Cloud plans start at $20 per month on a credit system. The trade is sync frequency, since its Standard cloud tier syncs hourly at best
- Hevo Data wins on pricing legibility: events-based billing where every insert, update, or delete counts as one event, plus a free plan covering 1M events per month. Streaming pipelines require its Professional tier at $750 to $849 per month
- Stitch wins on cheap, simple replication: from $100 per month for 5 to 300 million rows, a fit for small teams with modest volumes and no appetite for MAR math. Its higher tiers require annual billing and a sales call, so it stays a small-team tool
- Matillion wins on in-warehouse transformation depth: a visual, low-code pipeline builder on credit-based pricing tied to task hours rather than row counts, though that credit math takes its own learning
- dlt wins the code-first free path: a Python library under Apache 2.0 for engineering teams that want full ownership instead of a managed black box. Its managed layer, dltHub, starts at $12,000 per month, so the free path is the point
The pattern across all five: you leave Fivetran to gain control, predictability, or a lower bill, and you give up catalog breadth and hands-off maintenance to get it. That trade is exactly what Fivetran charges for.
How We Test Data Tools at Panoply
Every Panoply review starts with our own research and testing. We set the tool up ourselves, run a real task end to end, compare it against the strongest rivals in its category, and weigh sustained user sentiment, looking for complaints that repeat across time rather than one-off rants. For data movement tools, that means working the billing mechanics the way your finance team eventually will, because in this category the pricing model is half the product.
We do not invent benchmarks, and we do not repeat claims we cannot confirm. Prices current as of September 2026.
Fivetran Review: Should You Move Your Data With It?
Yes, if you fit the profile. Fivetran is the right choice for teams standardizing on a cloud warehouse who value engineering time over software spend, for SQL and dbt-comfortable teams who want transformations and ingestion under one roof, and for enterprises that need Oracle, SAP, or Db2 replication handled without a dedicated pipelines engineer.
Our rating: 4.3/5 (good). The automation and connector breadth are world-class; the score gives up ground on pricing unpredictability and tier-gated security. No rival matches the breadth, and no rival makes you watch your bill this closely either.
Skip it if you run a tight fixed budget against high-churn data, because MAR billing punishes exactly that shape of workload. Skip it if your stack sprawls past 20 connectors without governance, since the per-connection bases and unpooled discounts compound. And skip it if you want deep pipeline customization or flat, legible pricing: Airbyte serves the first need better and Hevo the second.
The next action costs nothing: start on the free 500,000 MAR tier, sync your real sources, and measure your actual MAR for a month. That number, not any review, tells you what Fivetran will cost you. And if it stays under the free cap, you may never pay at all.
FAQ
What does Fivetran actually cost?
There is no flat number. You pay per million Monthly Active Rows at a rate set by plan, $500 on Standard at the 1M threshold, falling per million as volume grows, plus a $5 base charge on every Standard plan connection syncing between 1 and 1M MAR. Transformations and Activations bill on separate meters. A free tier covers 500,000 MAR per month.
What is a Monthly Active Row?
A MAR is a distinct row inserted, updated, or deleted in your destination during a calendar month, tracked by primary key. A row updated ten times in a month counts once. Initial historical syncs are free, and since the 2026 pricing update, deletes count as active rows too.
Is Fivetran an ETL or ELT tool?
ELT. Fivetran loads raw data into your warehouse first, then transforms it in place with dbt-compatible models. That post-load pattern is the product’s core design, and the dbt Labs merger deepened it rather than changing it, since the company that loads your data now also owns the transformation layer that reshapes it.
What happened between Fivetran and dbt Labs?
The two companies completed a merger on June 1, 2026, positioning the combined business as data infrastructure for trusted AI agents. Transformations are now first-class inside Fivetran, and joint capabilities were already shipping by the September 2026 dbt Summit. Transformation runs still bill on their own meter.
Does Fivetran have a free plan?
Yes, and it is usable. The free tier covers 500,000 MAR per month with 15-minute syncs and 5,000 free transformation runs, and initial historical syncs do not count against the cap. It is the right place to measure your real MAR before committing to a paid plan.
