If your workspace has Genie spaces enabled, you can upload Excel (.xlsx) directly to Genie and analyze it there; it’s designed for quick validation and NLQ over uploaded files and UC tables.

Otherwise, you can use the following approaches:

Option A: Use the “Create or modify a table” UI with a clean CSV

• Open the Add or upload data flow and choose Create or modify a table. Select your CSV file, then pick an active compute to preview (SQL warehouse or serverless). Group clusters are not supported for this preview step.
• In the format options for CSV, set these correctly:
  • Turn on/off First row contains the header to match your file. Header settings must be consistent across files you upload together.
  • Set Column delimiter to the actual separator (comma is default; only one character is allowed and backslash is not supported).
  • If schema inference caused casting problems, disable Automatically detect column types so all columns are treated as STRING; you can transform after table creation.
  • If any field contains embedded newlines, enable Rows span multiple lines before previewing.
• Check and fix column names:
  • Column names must not include commas, backslashes, or Unicode characters; avoid special characters (spaces are commonly problematic—rename to underscores before creating or rename later in a notebook/workflow).
• Confirm size constraints:
  • The table upload UI supports up to 10 files with a combined size under 2 GB in a single operation.

Option B: Upload the file to a Unity Catalog volume, then create a table from it (robust and flexible)

• Upload your CSV to a UC volume (any format is allowed for volumes; UI uploads are limited to 5 GB per file).
• From Catalog Explorer, click the file and use Create table to convert it into a managed table with the same preview and configuration options (headers, delimiter, types) as the upload page.