Why explore ETL solutions? A data integration tool enables companies to centralize their data across all teams into one single location where data is sanitized, properly formatted, and stored for analytics purposes. With the ubiquity of data warehouses like Snowflake, BigQuery, Redshift, etc., ETL tools have now become a necessity in companies that rely on data to move the business forward — basically every business that hits a certain scale.
Among ETL tools available are two leading options: Fivetran and Matillion.
Fivetran is a managed ELT solution built for data teams to get data pipelines up and running quickly. It’s primarily used automatically sync data from SaaS tools to a warehouse. Fivetran started in 2012 and has many mature integrations out of the box. Fivetran enables users to transform data as well.
Fivetran also has household customer logos using their product and have thousands of customers using their solution.
Matillion is a self-hosted ETL product. The platform has a variety of pre-built integrations to move data from popular SaaS tools to a range of data stores. The product also allows for joins, aggregations, and filters to set up data pipelines tailored to your use case.
Founded in 2011, Matillion has two primary product offerings: Matillion Data Loader and Matillion ETL. Matillion Data Loader is a no-code product that ingests data into a warehouse with either batch updates or CDC. Matillion ETL also loads data into a warehouse but allows for transformations as well.
Matillion has hundreds of customers, many of which are household names.
Feature Comparison Overview
Connectors is the best starting place when evaluating an ETL solution. If your company has requirements involving specific SaaS tools or data stores that are offered by one ETL product and not the other, the rest of this article isn’t really going to convince you to use one tool vs. the other. You should go straight to the the tool that meets your business needs.
Fivetran offers about 3 times the number of connectors as Matillion as well as the major data stores. Not all connectors support CDC and Fivetran also builds all of their connectors in-house.
Matillion’s ETL product has about 100 connectors and all the major data stores while the Data Loader product has about 40 connectors. Not all connectors support CDC. Matillion builds all of their connectors in-house.
Fivetran has data partitioning, incremental load support, change data capture, and automatic scaling.
Matillion on the other hand has some different key features they’ve built to improve their platform’s scalability. It incorporates parallel processing, incremental load support, and change data capture.
Fivetran is pretty commonly commended for their support. They have support channels in place for when things break. In fact, the company claims that their support agent response time is within an hour, but this depends which plan you are on (the more costly the plan, the faster the support SLAs). On some of the lower tier plans, you’ll have to just submit a form and wait for a reply as opposed to directly messaging an agent.
Docs are available. They're thorough and managed internally by Fivetran.
With the exception of the most expensive tier, Matillion offers “standard” support. This boils down support 8 hours a day and 5 days a week with 4 customer support contacts. For the highest tier, Matillion offers “Mission Critical” as well as “Mission Critical Plus” with longer support hours and faster response times.
Documentation & Support
Matillion also has documentation that is managed internally — user guides, FAQs, you name it.
Unique to Matillion, the company offers a wide range of training resources to get started with the product. The content includes topics such as, orchestration workflows, basics on data integrations, advanced transformation techniques, etc.
Every company has custom data architectures and, therefore, unique data integration needs. A lot of tools don’t enable teams to address those, which results in a lot of investment in building and maintaining additional in-house scripts.
Fivetran offers customizability for connectors via cloud functions. Fivetran will take care of the transformation and works well with dbt on this.
Customers can code custom functions to edit connectors. And because Fivetran supports multiple serverless platforms like Azure functions, Google Cloud functions, and AWS Lambda, functions can be written using C#, F#, Go, Java, Node.js, or Python.
Matillion doesn’t have customizability functionality.
Fivetran does not transform data prior to loading, but rather allows for some common normalizations. It focuses more on support for post-load transformations SQL. Fivetran works well with dbt and it’s common for customers to have a Fivetran <> dbt pairing to get the most of our these data pipeline-focused tools.
Matillion has a low-code, drag and drop interface that you have to adhere to. You can do common transformations such as merges, filters, and joins without an understanding of how SQL and dbt work. However, this tends to be restrictive because the knowledge of the these tools allow you to use transformations that are very bespoke to your use case.
Preference for using a low-code GUI vs. SQL / dbt boils down to how technical you are.
Fivetran has row-based pricing. Specifically it charges based on monthly active rows. An active row is a row with data that is created, updated, or deleted in a given month. Even if the row is updated 200 times in a month, that row is only counted once.
Fivetran tends to be more affordable at low data volumes but as soon as this scales up, the product becomes quite expensive. In fact, a common reason companies look for Fivetran alternatives is because of how expensive Fivetran becomes. The pricing is not very predictable because it’s hard to know how many rows will be active in a given month. Fivetran is certainly considered a premium product with a premium price point.
Fivetran does have a free plan available and has a 14 day free trial on all plans, including the enterprise plan.
Matillion has a credit consumption based model. Customers must purchase credits that range from $2.00 - $2.70 per credit and these credits allow them to use the product. The credit model allows customers to pay upfront or pay as you go.
For Matillion Data Loader, credits are consumed when a row is batch updated or a row has an incremental update via CDC.
For Matillion ETL, credits are consumed when an instance is running and when more users than originally accounted for log into Matillion.
The Matillion ETL product is the more commonly used product. So for many customers, the pricing model is based on server hours, which will depend on the instance size customers run.
Matillion has a 30 day free trial with access to all features and 500 free credits.
Fivetran is a data ingestion tool alone, meaning it only allows for data to go from SaaS tools to your warehouse. If you’re looking for a Reverse ETL tool that moves data from your warehouse such as Snowflake back to your SaaS tools, you will have to find an additional solution for this.
Matillion does offer reverse ETL meaning it enables you have one-way syncs in either direction. However, you will be restricted to their low-code / no-code GUI interface as opposed to a SQL-based reverse ETL solution like some alternative options offer.
Both Fivetran and Matillion are high quality options for your company’s ETL solution. When it comes down to it, the decision for most likely comes down to the connectors available, pricing, and the low-code vs. SQL & dbt UI/UX preference.
What both tools lack
If you’re looking for a two-way ETL solution, you should explore Bracket. Unlike Matillion which has 2 one-way syncing solutions, Bracket’s bidirectional ETL product keeps SaaS tools and databases in sync. For example, an update in Salesforce will be reflected in Postgres and vice-versa, without any merge conflicts or infinite loop issues that you would run into by selecting a solution like Matillion that just stacks 2 one-way data pipelines on top of each other.
With a solution like Bracket, you also get unified logging, up to real-time data refreshes, and don’t have to worry about data inconsistency issues that you would have with Matillion or the combination of Fivetran and an additional third party vendor.
You can think of Bracket as a Heroku Connect alternative but for any SaaS and database / warehouse combination.
While Matillion and Fivetran are used primarily for analytics purposes, Bracket is used to power customer experience. User data is stored both in your web app as well as SaaS tools like Salesforce. So when a customer interacts with the web app, that data will be sent to business teams in tools like Salesforce to kick off any internal processes to support, market, or sell to the customer. Any data or process that needs to be completed to help onboard the customer can be entered into Salesforce (or HubSpot, Google Sheets, etc.) which will then show up or set off triggers in the customer portal.
Interested in learning more about Bracket? Check out our website here or go to our web app and get started!