# ELT (Extract, Load, Transform)

URL: https://softwaredictionary.org/terms/elt
Category: Databases
Last updated: 2026-10-06
Pronunciation: ee-el-TEE

In short: ELT is a data integration approach that extracts data from sources, loads it raw into a data warehouse, and then transforms it there, usually with SQL.

## What is ELT?

ELT stands for extract, load, transform. Data is extracted from sources such as application databases, SaaS tools and event logs, loaded into the target exactly as it is, and only then transformed into clean, analysis-ready tables inside the target itself. The target is usually a cloud data warehouse such as BigQuery, Snowflake or Amazon Redshift, or a lakehouse, a data lake that supports warehouse-style tables.

The approach spread in the 2010s, when cloud warehouses made storage cheap and computing power easy to scale. Ingestion tools copy the raw data in on a schedule, and transformations are written as SQL queries that build cleaned, joined and aggregated tables in layers, often managed with a tool such as dbt. Because the raw copy is kept, a fixed bug or a new business question can be handled by rerunning the SQL, without extracting the data again.

Think of grocery shopping: ETL is like washing and chopping vegetables at the market before bringing them home, while ELT brings everything home, puts it in the fridge, and prepares whatever each recipe needs later. The trade-off is that raw data, including personal or sensitive fields, lands in the warehouse, so access controls and masking matter, and heavy transformations add to the warehouse bill.

ELT vs ETL comes down to the order of the steps and where the work happens. In ETL, a separate system transforms the data before it reaches the target, so only cleaned data is stored; in ELT, the target stores the raw data and transforms it with its own computing power. ETL still fits when data must be filtered or anonymized before it leaves its source, or when the target can't handle heavy processing, and many real pipelines mix both.

## Key takeaways

- ELT loads raw data first and transforms it inside the target.
- It relies on the computing power of cloud data warehouses.
- Transformations are usually written in SQL, often organized with a tool such as dbt.
- Keeping the raw data lets you rebuild results without extracting again.
- ETL transforms before loading; ELT transforms after loading.

## Example: The transform step of ELT, inside the warehouse

```sql
-- Extract + Load: an ingestion tool copies raw rows into raw.orders as-is

-- Transform: build a clean table inside the warehouse with SQL
CREATE OR REPLACE TABLE analytics.daily_revenue AS
SELECT
    CAST(created_at AS DATE)    AS order_date,
    UPPER(TRIM(country_code))   AS country,
    SUM(total_cents) / 100.0    AS revenue
FROM raw.orders
WHERE status = 'paid'
GROUP BY 1, 2;

-- Logic changed? Rerun the query: the raw data is still there
```

## Frequently asked questions

**What is the difference between ELT and ETL?**

Both move data from sources into a target for analysis. ETL transforms the data on the way, before loading it, while ELT loads the raw data first and transforms it inside the target, usually a cloud data warehouse, with SQL.

**Is ELT replacing ETL?**

For analytics on cloud warehouses, ELT has become the most common approach. ETL is still used when data must be cleaned or stripped of sensitive fields before it is stored, and many pipelines combine the two.

**What is dbt?**

dbt, short for data build tool, is a widely used tool for the transform step of ELT. Each table is written as a SQL `SELECT` query, and dbt runs them in the right order inside the warehouse, along with data tests and documentation.

---

Software Dictionary: https://softwaredictionary.org/ · https://softwaredictionary.org/llms.txt