How to Import Excel to Snowflake (3 Methods)

Snowflake doesn't read .xlsx files natively with COPY INTO. The standard path is converting to CSV or Parquet, staging the file, and loading it. There's also a Python-based approach using Snowpark that can read Excel directly. Here are three methods.

Method 1: Convert to CSV, Stage, and COPY INTO

The standard Snowflake loading workflow: convert the Excel file to CSV, stage it, and load.

-- Create a file format
CREATE OR REPLACE FILE FORMAT csv_format
  TYPE = 'CSV'
  FIELD_OPTIONALLY_ENCLOSED_BY = '"'
  SKIP_HEADER = 1
  NULL_IF = ('', 'NULL');
 
-- Create a stage (or use a named external stage)
CREATE OR REPLACE STAGE my_stage
  FILE_FORMAT = csv_format;

Upload the file with SnowSQL:

# PUT the file into the stage
snowsql -q "PUT file://./sales.csv @my_stage AUTO_COMPRESS=TRUE"

Then load:

COPY INTO sales
FROM @my_stage/sales.csv.gz
FILE_FORMAT = csv_format
ON_ERROR = 'CONTINUE';

For the Snowflake web UI (Snowsight), you can skip the stage step for small files:

1. Open the target database and table
2. Click Load Data
3. Select your CSV file and configure options
4. Run the load

Using Parquet Instead of CSV

Parquet preserves types and compresses better. Convert in Python:

import pandas as pd
df = pd.read_excel("data.xlsx")
df.to_parquet("data.parquet", index=False)

Then stage and load:

CREATE OR REPLACE FILE FORMAT parquet_format TYPE = 'PARQUET';
 
-- After PUT file://./data.parquet @my_stage
COPY INTO sales
FROM (
  SELECT $1:id::INT, $1:product::VARCHAR, $1:amount::DECIMAL(10,2), $1:sale_date::DATE
  FROM @my_stage/data.parquet
)
FILE_FORMAT = parquet_format;

Method 2: Python with snowflake-connector-python

Python reads Excel directly and can write to Snowflake using the write_pandas helper:

import pandas as pd
import snowflake.connector
from snowflake.connector.pandas_tools import write_pandas
 
conn = snowflake.connector.connect(
    user="myuser",
    password="mypass",
    account="myorg-myaccount",
    database="mydb",
    schema="public",
    warehouse="compute_wh"
)
 
df = pd.read_excel("report.xlsx", sheet_name="Revenue")
df.columns = [c.strip().upper().replace(" ", "_") for c in df.columns]
 
# Create the table first
cursor = conn.cursor()
cursor.execute("""
    CREATE TABLE IF NOT EXISTS revenue (
        ID INTEGER,
        PRODUCT VARCHAR,
        AMOUNT DECIMAL(10,2),
        SALE_DATE DATE
    )
""")
 
# write_pandas stages a Parquet file internally and uses COPY INTO
success, nchunks, nrows, _ = write_pandas(conn, df, "REVENUE")
print(f"Loaded {nrows} rows in {nchunks} chunks")

write_pandas handles the staging and COPY internally. It converts the DataFrame to Parquet, uploads to a temporary stage, runs COPY INTO, and cleans up.

For Snowpark (Snowflake's Python runtime that runs inside Snowflake), you can read Excel files from a stage using the openpyxl library, which is pre-installed:

from snowflake.snowpark import Session
import pandas as pd
 
session = Session.builder.configs(connection_params).create()
 
# Upload Excel to a stage first
session.sql("PUT file://./data.xlsx @my_stage").collect()
 
# Read in Snowpark (runs inside Snowflake)
# Download from stage and parse
import io
stream = session.file.get_stream("@my_stage/data.xlsx")
df = pd.read_excel(io.BytesIO(stream.read()))
snowpark_df = session.create_dataframe(df)
snowpark_df.write.save_as_table("my_table")

Method 3: Mako

Mako imports Excel files into Snowflake without manual staging:

1. Connect to your Snowflake account in Mako
2. Click Import and select the .xlsx file
3. Choose the sheet and review the type preview
4. Select the target database, schema, and table
5. Run the import

Mako handles the conversion and staging internally. For one-off imports, this avoids the CSV-stage-COPY pipeline.

Type Mapping: Excel to Snowflake

Snowflake column names are case-insensitive by default (stored as uppercase). If your Excel headers are lowercase, they'll be uppercased in the table. Use quoted identifiers if you need case preservation.

Common Gotchas

Case sensitivity. Snowflake uppercases unquoted identifiers. A column named sale_date in your CSV becomes SALE_DATE in the table. This matters when using write_pandas -- make sure DataFrame column names match the table's column names (uppercase).

Warehouse must be running. COPY INTO and write_pandas require an active warehouse. If your warehouse is suspended, the load will fail or auto-resume it (which incurs costs).

File format mismatches. The most common load failures are from format issues: wrong delimiter, missing quotes around fields with commas, or encoding mismatches. Always test with a small sample first.

Date formatting. Snowflake parses dates in YYYY-MM-DD format by default. For other formats, set the file format option:

CREATE FILE FORMAT csv_dates
  TYPE = 'CSV'
  DATE_FORMAT = 'MM/DD/YYYY'
  SKIP_HEADER = 1;

Stage cleanup. Files uploaded with PUT remain in the stage until you remove them. Clean up after loading:

REMOVE @my_stage/sales.csv.gz;

Costs. Snowflake charges for active warehouse time during loads. For small Excel files, the minimum 1-minute charge applies regardless of how fast the load finishes. Batch multiple files into a single load to minimize cost.

Mako connects to Snowflake, PostgreSQL, MySQL, MongoDB, BigQuery, and ClickHouse with AI-powered autocomplete. Try it free at mako.ai.