What is ETL?
Definition
ETL stands for extract, transform, load. Data is retrieved from a source, prepared or cleaned, and then written to a destination such as a data warehouse.
In ELT, the order changes: data is loaded before transformations run in the destination.
Why it matters
Preparing data before it reaches the destination can make a shared dataset easier to use consistently. For example, an import process might standardise date formats, remove duplicate records and map source fields to agreed names. Reports can then work from that prepared structure instead of repeating the same cleanup.
The transformation rules also become a dependency. A changed source field or an incorrect exclusion can affect every downstream report. Keep the rules understandable, check representative outputs and decide how failed or partial runs are handled. It is worth establishing how a historical period would be reprocessed if a rule needs correcting, especially when the original input is not retained indefinitely.