How to load a CSV into a table in the Lakehouse

João Barros 04 de July de 2026 4 min read

Having your data in a CSV file is the most common starting point in any data project, but to analyse it with SQL or Power BI you first need to turn it into a table. Loading a CSV into a table in the Microsoft Fabric Lakehouse does exactly that: in a few clicks you get a Delta table ready to query, without writing a single line of code. The process is quick, and the example below is simple enough for anyone to follow. This guide is ideal for anyone starting out with Fabric who wants the most direct path from file to analysis.

Prerequisites

An account with access to Microsoft Fabric and an active capacity (trial or paid).
A workspace where you have edit permissions.
A Lakehouse already created — if you don't have one yet, create a Lakehouse item in the workspace.
A small CSV file to test with, for example vendas.csv, with the column names in the first row.

Step 1: Open the Lakehouse

Go into your workspace in Microsoft Fabric and open the Lakehouse where you want to store the data. On the left you'll see the Lakehouse explorer, split into two areas: Tables, which holds the Delta tables you can query with SQL, and Files, which holds raw files such as your CSV. Understanding this split is half the battle: first we put the file in Files, then we turn it into a table in Tables.

Step 2: Upload the CSV to the Files area

Right-click the Files section and choose Upload, then Upload files. Select your vendas.csv and confirm. The file now appears in the Files area, stored in OneLake, but it is not a table yet — it's just the original file.

Tip: make sure the CSV has a header row with clear column names, ideally without spaces or accents. This helps Fabric detect the schema correctly and avoids errors in the next step.

Step 3: Use "Load to Tables"

With the file visible in Files, right-click vendas.csv and choose Load to Tables, then New table. Type a name for the table, for example vendas, confirm that the separator is the comma and that the option to use the first row as the header is enabled. Click Load. The New table option creates a brand-new table; if you pick Existing table instead, you append the data to a table that already exists.

Behind the scenes, Fabric reads the CSV, infers the type of each column (text, number, date) and creates a Delta table in the Tables area. The Delta format is open and optimised for analytical queries, which is why it is the default format of a Lakehouse.

Step 4: Query the table with SQL

Once the table exists, you can query it without leaving Fabric. In the top-right corner, switch from Lakehouse mode to the SQL analytics endpoint and open a New SQL query. Try a simple query to see the first rows:

SELECT TOP 10 *
FROM vendas;

Prefer to work in a notebook? With PySpark you get the same result from the table:

df = spark.read.table("vendas")
display(df.limit(10))

Check the result

In the Tables area of the Lakehouse explorer you should now see the vendas table with the Delta table icon. Expand it to confirm that the columns and their types were detected as you expected. The SELECT query above should return the first rows of your CSV. If a numeric value showed up as text, or a date was misread, go back to the file, fix the header or the data and load it again — it's quick and free. This small test confirms the data is accessible and ready for the next step of your project.

Conclusion

In four steps you turned a plain CSV into a Delta table ready to analyse with SQL, Power BI or notebooks. A good next step is to stop uploading files by hand: with a Dataflow Gen2 or a pipeline you can automate the load whenever the data changes. What's the first everyday file you'll bring into the Lakehouse?