A product code can look like a number without being a quantity.
The receiving team needs the same identifier the sender supplied. A tidy column does not establish that the characters survived the handoff.
Microsoft documents conversions that remove leading zeros and limit numeric precision to fifteen digits. It recommends text for values that must retain their original representation. Microsoft: keeping leading zeros and large numbers.
Start from the original CSV. Choose the import type before loading the data. Compare the exported identifiers with the original values before handing the file onward.
The procedure below uses synthetic product codes. The comparison table defines acceptance targets. It does not report a live Excel experiment or a production migration.
Decide Which Columns Carry Identity
Mark the identifier columns before opening the file in Excel. Product codes, warehouse codes, and account identifiers belong on that list when their exact characters define the record.
Keep quantities separate. A quantity supports arithmetic under an agreed unit. An identifier distinguishes one record from another. The fictional code 00042 does not become more useful when someone subtracts it from another code.
Write a short rule for each identifier column: preserve every character, preserve case, preserve leading zeros, and reject unexpected blanks. Decide whether spaces and punctuation are meaningful with the source owner. Do not remove them as a cosmetic cleanup.
CSV does not carry a column's datatype declaration by itself. The W3C tabular-data primer describes accompanying metadata for that purpose. Ask for a column specification when the file's meaning is unclear.
For this exercise, every value in product_code is text. The numeric quantity_units column remains separate. This is a local exercise rule, not a universal definition of product identifiers.
Separate Stored Text From Display Formatting
Microsoft distinguishes importing codes as text from applying custom number formats. Its leading-zero guidance limits the custom-format approach to workbook display and codes below sixteen digits. Microsoft: leading zeros and custom formats.
Treat the stored identifier as the acceptance target. A display rule that makes the number forty-two appear as five digits does not prove which original identifier arrived. The source might have contained 42, 042, or 00042.
For this proposed review, inspect the imported type and exact value together. Do not accept a screenshot as the only evidence. A screenshot cannot establish that an exported field contains the agreed characters.
Formatting cannot recover digits that were destroyed during conversion. Return to the original file before attempting recovery. Microsoft explains this precision limit and the need to enter long codes as text in its leading-zero guidance.
Do not infer a missing prefix from neighboring rows. A pattern is a reason to investigate, not permission to manufacture an identifier.
Record the Excel Version and Import Route
The instructions here concern desktop Excel. Record the operating system, product edition, installed version, locale, and import route in the handoff worksheet. These details identify the procedure someone must reproduce.
Microsoft lists Automatic Data Conversions for Microsoft 365 and Excel 2024 on Windows and Mac. Its documentation places the controls under File > Options > Data on Windows and Excel > Preferences > Edit on Mac. Microsoft: automatic data conversions.
Do not assume that Excel 2021, 2019, or 2016 has those controls. Those editions are outside the feature's documented applicability. If the control is absent, record that limitation and use an explicit import procedure your edition supports.
Power Query availability has a different boundary. Microsoft documents it for Excel 2016 or later on Windows and Microsoft 365 for Mac. It lists Text/CSV among the supported sources. Microsoft: Power Query data sources by version.
For the Mac editor procedure below, Microsoft specifies Microsoft 365 for Mac version 16.69 (23010700) or later. Its documented path begins at Data > Get Data (Power Query). Microsoft: importing and shaping data on Mac.
These are documented product boundaries checked on October 28, 2026. Confirm the controls in the installed version. This article does not verify web, mobile, or perpetual Mac editions against the same procedure.
Set Conversion Controls Before Direct Opening
In a supported desktop edition, review the Automatic Data Conversions options before opening a CSV. Disable conversions that conflict with the identifier specification.
The documented controls cover leading zeros, long numeric strings, digits surrounding E, and continuous date-like letter-and-number codes. The date-like setting does not cover every punctuated form; JAN-1 may still become a date. These settings do not directly govern Power Query. Microsoft: conversion controls and limitations.
Use the controls as one condition in the review. Keep the source comparison as a separate condition. A person can use a different import route from the one the worksheet describes.
If the team standardizes on Power Query, document that route explicitly. Avoid a vague instruction to open the file carefully. Record the named query and its intended input columns so another operator can identify the same workflow.
Assign Text Inside Power Query
Keep the original CSV unchanged. Create a workbook for the exercise, then import the source through the appropriate desktop menu.
On Windows, choose Data > From Text/CSV. Select Transform Data from the preview to inspect the query before loading. Microsoft documents this route in its text and CSV import instructions.
On supported Microsoft 365 for Mac, select the Text/CSV source through Get Data (Power Query). Use Transform Data to inspect the import in the editor. Microsoft: Mac Power Query procedure.
Power Query can infer types for CSV input and add a Changed Type step. Inspect that step before accepting the result. Microsoft documents automatic detection and explicit column types in its Power Query datatype reference.
For this exercise, replace any numeric or date conversion of product_code with Text at the earliest type-setting step. Re-evaluate from the untouched source. Do not add a text conversion after a step that already changed the identifier.
Select the identifier column, set its datatype to Text, and inspect every synthetic value before loading. Keep the quantity column under its own numeric rule. The column-header datatype menu is documented in the Power Query datatype reference.
Before using Close & Load, record the source filename and the relevant applied steps. A later refresh review should use the same column rules. Stop if the incoming header changes or the identifier column disappears.
Use a Small Round-Trip Worksheet
Create an acceptance sample before handling a business file. The values below are invented product codes. They deliberately include characters that deserve inspection.
product_code,quantity_units
00042,3
78345678901234567,8
47E6,2
MAR2,5
MAR-2,1
Use an unchanged copy as the comparison source. Keep a second copy for the Excel exercise. Give the output a different filename so the original remains available.
The following table contains required outcomes, not measured application behavior. Keep both target columns unchanged. Add separate columns for the observed imported value, imported type, exported value, and pass or fail result.
| Before import: original code | After import: required text value | After export: required field value | Review focus |
|---|---|---|---|
00042 |
00042 |
00042 |
Leading zeros |
78345678901234567 |
78345678901234567 |
78345678901234567 |
Every digit preserved |
47E6 |
47E6 |
47E6 |
Letter retained as part of code |
MAR2 |
MAR2 |
MAR2 |
Continuous date-like code |
MAR-2 |
MAR-2 |
MAR-2 |
Punctuation retained |
For each row, record the observed value, its type in the imported column, and the exported field. Mark a mismatch as a failed comparison. Do not replace the target with the observed value to make the worksheet pass.
Check all five records. The sample is deliberately small enough to inspect completely. A successful sample only demonstrates the conditions exercised. It does not establish that every future source file follows the same rules.
For a business handoff, extend the comparison to the agreed identifier set. Include blanks, duplicates, and unexpected lengths in the exception report. Keep identifier integrity separate from decisions about whether two business records should be merged.
Inspect the Export as a File
Save the working workbook separately before exporting. Select the intended worksheet and use Save As with the required CSV format. Microsoft notes that text export saves the current worksheet and can omit unsupported workbook features. Microsoft: exporting text and CSV files.
Inspect the exported file through a text viewer or a CSV reader that preserves field strings. Do not rely only on reopening it through the same spreadsheet route. That would leave the review dependent on the same interpretation settings.
Compare parsed field values, not the appearance of quotation marks around fields. Keep the required delimiter, encoding, and header order on the worksheet. Confirm those conditions separately from identifier equality.
For the synthetic exercise, require five data records and five distinct codes. Require the exact five original code strings. Require quantities of three, eight, two, five, and one beside the corresponding codes.
Equal record counts alone are insufficient. Two different codes could have become the same value while the row count stayed unchanged. Compare the identifiers and their associated quantities together.
If a receiving workflow uses Google Sheets automation, run a separate receiving check there. This Excel procedure establishes no claim about Google's conversion behavior.
Recover From the Earliest Intact Source
When an observed code differs, stop the handoff. Keep the original, the working workbook, and the failed export clearly labeled. Record where the first mismatch appears.
If the original CSV contains the correct characters, discard the damaged working result and repeat the import with explicit text handling. Preserve the failed comparison as evidence of why the procedure changed.
If the original CSV is already damaged, request a fresh export from the source owner. Explain which codes failed and what authoritative value is needed. A reformatted workbook cannot establish the missing information.
If no intact source is available, mark the affected identifiers unresolved. Ask the owner to verify them against an authoritative record. Do not invent trailing digits, leading zeros, or a presumed date code from a familiar pattern.
Record the corrected delivery as a replacement with its own comparison result. This evidence can support later data quality work, while keeping the recovery task focused on the exported identifiers.
Leave a Repeatable Handoff Record
Put the accepted procedure in the team's operating documentation. Include the source specification, Excel edition and version, import route, text columns, and export format.
Attach the completed round-trip worksheet. Identify who checked the values and which file that person accepted. Keep a pending review distinct from a completed handoff.
Repeat the synthetic exercise when the import route, query steps, product edition, or source layout changes. Name the changed condition so the next reviewer knows what the check covers.
Preserve the identifier before using it.
Sources
- Microsoft: Keeping leading zeros and large numbers. Text, display formatting, and numeric precision. Checked 2026.10.28.
- Microsoft: Set automatic data conversions. Desktop applicability, settings, and limitations. Checked 2026.10.28.
- Microsoft: Import or export text files. Windows import and worksheet export procedures. Checked 2026.10.28.
- Microsoft Learn: Data types in Power Query. Type detection and column controls. Checked 2026.10.28.
- Microsoft: Import and shape data in Excel for Mac. Microsoft 365 Mac editor requirements and import route. Checked 2026.10.28.
- Microsoft: Power Query data sources in Excel versions. Platform and source availability. Checked 2026.10.28.
- W3C: CSV on the Web: A Primer. Column types and accompanying metadata. Checked 2026.10.28.