CSV → Excel / verification guide

Keep leading zeros and long IDs when moving CSV to Excel

An identifier is text, not a number to calculate. Changing the display format after double-clicking a CSV cannot restore zeros or digits that were already lost. Import the original CSV with identifier columns as text, then verify the saved workbook.

Display difference or changed value?

OriginalExample after importCauseRedo
00123123Leading zeros treated as a numberRe-import the original as text
9007199254740993A rounded valueLong number coerced to numericRe-import the original as text
Date-like codeDate displayAutomatic type conversionChoose and verify the column type

If the source is already rounded, do not guess a replacement.

Keep only these columns as text

Keep customer_id and sku as text. Select amount as a number only after checking it. Quoting alone does not always force Excel to keep a type.

customer_id,sku,amount
00123,9007199254740993,1200
00007,000045,980

The expected result is 2 rows, 3 columns, text IDs, and a total of 2180.

Create the handoff with Web-Tool

  1. Save the fixture and keep the source CSV unchanged.
  2. Open Table to Excel, load the file, and inspect the imported table and types.
  3. Keep identifier columns as text and choose only checked numeric columns.
  4. Save the .xlsx and verify cell types, values, row count, and column count with an independent reader.

This guide's files and your input stay in the browser. Customer IDs and source files are not sent to a URL, analytics, or an external server.

The Excel-only alternative

  1. Use Data → From Text/CSV to load the original.
  2. In the transform step, set customer_id and sku to text.
  3. If automatic conversion changed a value, restart from the original rather than correcting the display afterward.
  4. Verify the saved cells for the Excel version and recipient system you actually use.

Cases that cannot be restored

If zeros or digits were lost in the source file, do not reconstruct them by guesswork. State which Excel version and features were actually checked, and ask the recipient when the condition is unknown.

Use the JSON & CSV workbench or character and UTF-8 byte counter for local checks without sending the value.