ETL tools—ETL stands for extract, transform, and load—move data from source systems, like sales CRMs, marketing tools, event logs, and a host of other sources. In other words, they extract data from these systems, transform it to standardize it and make it suitable for a database, and load it into that database.
In modern architecture, ETL tools are sometimes called ELT, or extract-load-transform tools. These new tools have been rebranded this way because they do very little transformation before writing the data into the warehouse. In essence, they simply copy data from a source system and paste it into another.
ELT, however, is a slight misnomer for two reasons. First, there’s still a little bit of transformation between the E and the L, like standardizing date formats. These transformations are typically cosmetic though.
Second, most ELT tools don’t actually transform data once it’s loaded into your warehouse. That job is often done by separate transformation tools.
Proper ETL tools don’t just move data from one place to another; they do it regularly, automatically, and durably. If, for example, we connect an ETL tool to Facebook Ads so that we can write ad conversion data to my data warehouse, the ETL tool shouldn’t conduct a one-off sync from one to the other. It will maintain an active connection to Facebook and regularly push new data from Facebook into our warehouse. Also, if our warehouse goes down for some reason, the ETL tool should make sure it “catches up” once it’s back online.
These capabilities are critical for ensuring that the data in our warehouse is reliable, and are just as important to any ETL tool as the ability to read from one place and write to another. Databases are only as valuable as the data in them.
Learn how to choose ETL tools for your analytics stack.
ETL stands for extract, transform, and load—move data from source systems, like sales CRMs, marketing tools, event logs, and a host of other sources.
Related terms:Transformation tools, data cleaning