6 min read

Best ETL tools: eight options compared by pricing unit and open-source choice

ETL stands for extract, transform, load: pulling data out of your business systems, reshaping it, and loading it into a central store such as a data warehouse. Many modern tools actually do ELT (extract, load, then transform inside the warehouse), but buyers still search for “ETL tools,” so we use the familiar name here.

The biggest difference between these tools isn’t features. It’s the unit you pay by. One charges by rows that change each month, another by credits, another by events, another by compute hours. Two tools with the same headline price can produce very different bills for the same data, so we’ve put the pricing unit at the center of this comparison.

The eight at a glance#

ToolOwner or statusPricing unitPublished entry priceOpen-source option
FivetranMerged with dbt Labs (June 2026)Monthly active rows (MAR)Free plan up to 500,000 MARdbt Core integration; core product is not open source
AirbyteIndependentCreditsStandard from $20 per monthYes
MatillionIndependentMatillion CreditsFree Developer plan to try; paid plans via salesNo
Qlik Talend CloudQlik (owns Talend)Subscribed usage capacityQuote onlyWe couldn’t confirm a current open-source edition
StitchA Qlik productRows per monthStandard from $100 per monthNo
HevoIndependentEventsStarter 299permonth,or299 per month, or 265 billed annuallyNo
AWS GlueAmazon Web ServicesDPU-hours (data processing units)$0.44 per DPU-hour in the page’s exampleNo
InformaticaSalesforceInformatica Processing Units (IPUs)Quote only; free Cloud Data Integration offerNo

How we compared them#

  1. Connector claims: how many sources the vendor says it supports, quoted only where stated.
  2. Pricing unit: what you’re billed for, and whether a starting price is published.
  3. Open-source option: whether there’s a version you can run yourself without a license fee.

Vendor pages checked 2026-11-11.

The tools#

Fivetran#

Pricing modelBest suited to
Usage-based on monthly active rows, with a free planTeams that want fully managed pipelines with little upkeep

Fivetran completed its merger with dbt Labs on 2026-06-01; dbt is a framework for transforming data inside the warehouse, so loading and transforming now sit under one company. The pricing page offers a Free plan up to 500,000 MAR for connections and 5,000 monthly model runs for transformations. The Standard plan lists unlimited users, 15-minute syncs and 700+ fully managed connectors, and the page says annual contracts save up to 22%. A 5basechargeappliestostandardconnectionswithbetween1and1millionMAR.Thepage′sworkedexampleputsfourmarketingconnectorsforacompanyof1to200employeesatamedianof5 base charge applies to standard connections with between 1 and 1 million MAR. The page's worked example puts four marketing connectors for a company of 1 to 200 employees at a median of 511.15 per month.

Trade-off: MAR pricing tracks change volume, so a busy source system can raise the bill even when your team does nothing new.

Airbyte#

Pricing modelBest suited to
Credit packages on Airbyte Cloud, or self-hosted open sourceTeams that want open source or the option to run pipelines in their own cloud

Airbyte’s home page describes open-source data integration for 700+ sources, run in your own cloud network or on Airbyte Cloud. Its pricing page offers a 14-day trial with 400 credits and no card, a Standard plan from 20permonth(5credits,roughly1GBofsyncs),andaPlusplanfrom20 per month (5 credits, roughly 1 GB of syncs), and a Plus plan from 189 per month for 40 credits that adds SSO (single sign-on) and 15-minute syncs, with larger credit packages above that.

Trade-off: self-hosting saves license fees but moves the upkeep onto your team.

Matillion#

Pricing modelBest suited to
Consumption-based credits across three plansData teams building transformations visually or in SQL and Python

Matillion’s pricing page describes three plans, starting with Developer for individuals and Teams for scaling businesses, and a credit system where you pay for the work pipelines actually do. The Developer plan, for individuals, includes one developer user, unlimited projects, pre-built connectors and a low-code canvas plus SQL and Python, and can be tried free. Its AI pipeline assistant, Maia, is available on all plans.

Trade-off: paid plan prices come from sales, so credit consumption needs modeling up front.

Qlik Talend Cloud#

Pricing modelBest suited to
Subscribed usage capacity by editionLarger organizations that want integration, data quality and governance together

Talend is now part of Qlik, and its integration product is sold as Qlik Talend Cloud. The pricing page lists Starter, Standard, Premium and Enterprise editions, all through sales. You subscribe to a usage capacity measured by data volume moved, job executions and execution duration, with a dashboard and alerts as you approach the limit. A client-managed (self-hosted) version is also offered, and there’s a free trial.

Trade-off: three usage measures in one capacity make forecasting harder than a single-unit model.

Stitch#

Pricing modelBest suited to
Rows per month, by planSmaller teams that want simple, published pricing

Stitch now describes itself as “A Qlik product.” Its pricing page shows Standard from 100permonth,withasliderfrom5to300millionrowspermonthandmonthlyorannualbilling.Advancedis100 per month, with a slider from 5 to 300 million rows per month and monthly or annual billing. Advanced is 1,500 per month for 100 million rows and three destinations, and Premium is $3,000 per month for 1 billion rows and five destinations, both billed annually. There’s a free trial on Standard.

Trade-off: the pricing is clear, but with Qlik also selling Talend Cloud, ask about the product roadmap before committing long term.

Hevo#

Pricing modelBest suited to
Event-based usage tiersMid-sized teams that want published prices and a no-code setup

Hevo’s pipeline pricing page lists Starter at 299permonthpaidmonthly,or299 per month paid monthly, or 265 per month paid annually (12% off), with tiers by millions of events, where each insert, update or delete in the destination counts as an event. A higher plan starts at $750 per month. Starter includes hourly scheduling, up to 10 users and dbt integration. Its site describes loading data from 190+ sources.

Trade-off: event counts rise with updates, not just new records, so frequently edited data costs more to sync.

AWS Glue#

Pricing modelBest suited to
Pay per DPU-hour of processingTeams already on AWS with engineers who write code

AWS Glue is Amazon’s serverless data integration service. Its pricing page bills processing in DPU-hours. The page’s own example: a 15-minute job using 6 DPUs at 0.44perDPU−hourcosts0.44 per DPU-hour costs 0.66. There’s no license fee; you pay for compute while jobs run, plus charges for related features like the data catalog.

Trade-off: low unit prices, but setting it up is engineering work, not a point-and-click connector library.

Informatica#

Pricing modelBest suited to
Informatica Processing Units (IPUs) across its cloud platformLarge organizations with complex data management needs

Informatica is now part of Salesforce. Its pricing page explains that IPUs give access to eligible cloud services, with consumption measured per service, and it offers a free Cloud Data Integration option. The platform goes well beyond ETL into data quality, master data and governance.

Trade-off: broad and powerful, but more platform than most small teams need.

Picking a shortlist#

Start with the unit. If your sources change constantly, compare MAR (Fivetran), events (Hevo) and rows (Stitch) against your real volumes, not the vendor’s example. If you have engineers and want control, Airbyte’s open source or AWS Glue keep license costs down. If governance matters as much as moving data, look at Qlik Talend Cloud and Informatica.

Then check where the data lands. Most of these tools load into warehouses such as Snowflake or Databricks (our Snowflake vs Databricks comparison explains the difference). If what you really need is app-to-app syncing rather than analytics, an iPaaS platform is the better fit, and our API integration guide explains the distinction.