Supplier data · Updated 16 November 2026
How to prepare a supplier CSV for import
A CSV can look tidy in a spreadsheet while still having repeated headers, a short row or columns in the wrong order. Work on a copy of the supplier file and get the destination's current import template before editing. The examples here are invented inventory data.
Start with the file structure
Use CSV Import Checker to find repeated headers, quoting problems and unexpected field counts. These checks do not verify prices, stock levels or your importer's business rules.
Check a CSV before importing1. Find the duplicate header and short row
Download the original sample CSV (2 deliberate issues). Use Open a CSV file in the checker, or paste the example below. These samples contain invented data and are for practice.
Paste this comma-separated example into the checker and select Check CSV:
SKU,Product,Price,Price
A101,"Mug, blue",12.00,8.00
A102,Notebook,6.50
A103,"Frame, small",15.00,9.50
It reports two issues: Price repeats in columns 3 and 4, and line 3 has only three fields instead of four. The comma inside "Mug, blue" belongs to the product name because that field is quoted.
Select Download check report (.txt) to keep the issue locations and suggested fixes. The report should say Issues found. It records the check; it does not repair the source file.
2. Correct the data using the supplier's meaning
Download the corrected sample CSV to compare it with your edits.
For this example, assume the supplier confirms the fourth column is Cost and the Notebook's missing cost is 4.00. Make those corrections in a copy:
SKU,Product,Price,Cost
A101,"Mug, blue",12.00,8.00
A102,Notebook,6.50,4.00
A103,"Frame, small",15.00,9.50
Check the corrected CSV again. It should pass the header and field-count checks with four columns and three data rows. Select Download corrected CSV to save the checked text for the next step. For a real file, confirm missing values with the supplier; do not invent them or insert zero just to make the row pass. A blank field is appropriate only when its meaning and the destination allow it.
Download a fresh check report after rechecking: its status should now be Passed. Editing the CSV invalidates the previous download until you run the check again.
3. Map columns to the destination
Open CSV Column Mapper. Paste the corrected CSV and use this illustrative destination header:
SKU,Title,Price
Select Set up mapping, then map SKU → SKU, Product → Title and Price → Price. The Cost column is deliberately excluded. Select Build mapped CSV; the first data row should be:
A101,"Mug, blue",12.00
Use the exact header from your actual destination rather than assuming these three example fields are sufficient. Review every selection, including any Use empty values choice. Download the full CSV: the preview shows only the first five data rows. The mapper accepts UTF-8 comma-separated CSV input up to 2 MiB.
Review the mapping summary and save its report
After building, review Mapping summary. This example should show 3 input and 3 output data rows, four source columns and three destination columns. Product → Title is renamed; source column 4, Cost, is omitted. The table also identifies moved columns and any source reused in more than one destination. Counts exclude the header, and data-row order is preserved.
Select Download mapping report (.txt) to save csv-mapping-report.txt beside mapped.csv. The report describes the prepared export and includes headings, mapping choices and counts, but no product-row contents. Review headings before sharing. Changing the source, destination, column choices or loading a mapping clears the report and CSV until you build again. Download both again after changes.
To practise empty-value mapping, use Title,SKU,Optional as the destination header. Map Product → Title, SKU → SKU and choose Use empty values for Optional. Expect three output rows, two omitted source columns (Price and Cost), and three inserted blank cells. These counts do not include blanks already present in source cells. Only use an empty destination when your importer accepts it. Restore SKU,Title,Price and its original mapping before continuing with the example below.
This is a mapping record, not a formula-safety or importer-acceptance certificate. Check the exact exported CSV in the next steps.
Check required values without changing the export
Tick Required value for each destination column your importer requires, then select Build mapped CSV. For the corrected example, select SKU, Title and Price: expect Passed, with no missing values across three data rows. Unticked columns are not checked. With no selected columns, or with only a header and no data rows, the result says Not checked.
Download the required-field practice CSV. Map SKU to SKU, Product to Title and Price to Price, and mark all three required. Expect 3 missing values across 3 data rows: SKU in data row 2, Title in data row 3 (spaces only), and Price in data row 4. Their CSV record numbers and source/mapped line numbers are 3, 4 and 5. The first row's price of 0 is present and should not be flagged. 0.00 also counts as present; this is a presence check, not a price-range rule.
The on-page findings and mapping report identify the destination column, data row, CSV record and start line in both the source and mapped CSV. Data rows exclude the header; records include it. Source and mapped line numbers can differ when multiline fields are omitted or reordered. All rows are checked; the page and report list up to 100 findings and show the full total.
Checks do not trim, fill, remove or otherwise alter values: the mapped CSV is identical with the same mapping whether required checks are on or off. Downloads remain available when issues are found so you can save and correct the file. Confirm missing values with the supplier, fix a copy, then rebuild and download a fresh CSV and report. A passing presence check does not verify numeric ranges, unique identifiers, formula safety or importer acceptance.
Changing a required choice clears the previous result until you rebuild. Required choices start off with a new setup, source CSV, destination text or sample. Save mapping includes your required-field choices; Load mapping restores them for the destination columns, even when source columns are reordered. Older version 1 mappings still load, with every required choice off and a reminder to review them. Return to the corrected sample and check your required fields before continuing.
Find duplicate SKUs without changing your data
In CSV Column Mapper, tick Check duplicates for SKU and build the CSV. You can select other destination columns too; each is checked separately, not as a combined key. Every occurrence of a duplicate is flagged, including the first. The page and mapping report identify its data row, CSV record, source/mapped start lines and the first row of its group, without including the SKU value.
Both comparison options start off. Ignore case compares lowercase text, so ABC and abc match. Ignore leading/trailing whitespace compares trimmed text, so ABC and ABC match; internal whitespace is preserved. Neither option changes the exported values. Leading zeros remain significant: 0012 and 12 are always different. Blank and whitespace-only values are skipped; select Required value separately to report them.
For practice, paste this source, use SKU,Title,Price as the destination, map Product to Title, and select duplicates for SKU:
SKU,Product,Price
0012,Mug,0
12,Notebook,0.00
ABC,Frame,15
abc,Stand,6
" ABC ",Tray,8
0012,Cup,12
With both options off, expect 1 duplicate group and 2 occurrences: data rows 1 and 6 (CSV records/lines 2 and 7). Ignore case alone gives 2 groups and 4 occurrences (rows 1, 3, 4, 6). Ignore whitespace alone gives 2 groups and 4 occurrences (rows 1, 3, 5, 6). Both on gives 2 groups and 5 occurrences (rows 1, 3, 4, 5, 6). Row 2, 12, never joins the 0012 group. These are invented practice records, not recommendations to discard stock.
All rows are checked. The report lists up to 100 duplicate occurrences, ordered by destination column then data row, and shows full group/occurrence/affected-row totals. If no columns are selected, no rows exist, or selected columns have only blank values, the status is Not checked. Review duplicates with the supplier; do not automatically merge or delete them. Downloaded CSV bytes stay identical for the same mapping regardless of comparison options. Rebuild and download a fresh report after changes.
Check price and stock formats
Use the Numeric rule dropdown for each destination: choose Price: non-negative decimal, Stock: non-negative whole number or No numeric rule. These rules are optional and start off. Prices accept digits with an optional dot decimal, including 0, 0.00 and 12.50. Stock accepts whole-number digits, including 0; both 2.5 and 2.0 are flagged because decimal notation is not accepted.
Negative or positive signs (including -0 and +1), currency symbols, thousands separators, decimal commas, scientific notation, and internal spaces are flagged. Outer whitespace is ignored for validation only. Blanks and whitespace-only values are skipped; choose Required value as well when missing values are not allowed. SKU columns should normally keep No numeric rule so text identifiers are not subjected to price or stock formats.
Try this invented CSV with destination SKU,Price,Stock. Select Price for the Price column and Stock for the Stock column:
SKU,Price,Stock
A101,0,0
A102,-1,2
A103,12.50,2.5
A104,£5,3
A105,,
A106," "," "
Expect 3 invalid values across 3 data rows: Price in data row 2 (CSV record/line 3), Stock in data row 3 (record/line 4), and Price in data row 4 (record/line 5). Eight nonblank values are checked and four blanks are skipped. The zero price and zero stock pass. If both columns are also required, the separate required-field section reports four missing values across data rows 5 and 6.
The page and downloadable mapping report identify the rule, column and source/mapped line location without including cell contents. All rows are checked; up to 100 numeric findings are listed with full totals. No rules, no rows or only blank selected values produce Not checked. Changing a rule invalidates the old export and report until you rebuild. Numeric rules never round, fill, trim or rewrite the CSV: the download remains byte-for-byte identical for the same mapping. Confirm corrections with the supplier, rebuild and download a fresh report. Passing these format rules does not verify currency, maximum amounts, decimal precision, real stock availability or importer acceptance.
Save the mapping for your next supplier file
- Once all source choices are selected, choose Save mapping to download
utilityshelf-column-mapping.json. The version 4 file contains source and destination headings, column selections, required and duplicate choices for each destination, both duplicate comparison options and a numeric rule for each destination, with no product rows. - Next time, open or paste the new supplier CSV, select Load mapping, and choose that JSON file. The destination headings, required choices, duplicate settings and numeric rules are restored. A source column moving does not move checks to a different destination.
- Review every source choice, required/duplicate checkbox, comparison option and numeric rule, then select Build mapped CSV. Loading a mapping does not automatically create an export.
Columns may change order: the mapper follows their exact header names. If a header is added, removed, renamed or changed in case or spacing, loading stops and asks you to set up a fresh mapping. Duplicate or blank source headers cannot be used in reusable mappings. This protects against silently selecting a different column. Header matches do not prove that a supplier has kept the same field meanings or units. Invalid mapping files (including missing, malformed or mismatched check settings) are rejected; no previous export or report remains available. Files saved by this version require the updated mapper; older version 1, 2 and 3 files remain supported. All older files start with numeric rules off. Version 1 starts with all checks off; version 2 restores required choices but starts duplicate checks and comparison options off. New setups, source CSVs, destination text and samples reset check settings.
Review problem rows together
After building the CSV with your chosen checks, select Show problem rows. The combined table brings together missing required values, duplicate occurrences and invalid price or stock formats. Each entry shows the data row, CSV record, source and mapped start lines, source and destination columns, a quoted value preview and the reason. Duplicate groups include their first occurrence. A row can have several issues; totals distinguish issues from affected rows.
All data rows are checked. The table shows the first 100 issues in data-row and destination-column order, with an explicit notice if more exist. Previews longer than 200 characters are shortened; quoted spaces and escaped line breaks distinguish whitespace from an empty cell. The mapping report includes the same combined locations and reasons, but omits cell values. Hiding the table never filters the CSV. Changing inputs or rules clears the old results; rebuild and download a fresh CSV and report after correcting a copy of the source. With no checks selected, the view says Not checked, not passed.
4. Review formula-like values in the export
Open the downloaded file in CSV Formula Checker. A value such as "=SUM(2,3)" can become a formula when opened in a spreadsheet even though the CSV quotes are correct. A flag may also be legitimate data, such as a negative amount.
The checker flags prefixes; it does not sanitise the file. Review the intended values and your spreadsheet's text-import settings. Do not assume that adding quotes makes a CSV safe. See OWASP's CSV formula-injection guidance for application-specific limitations.
Select Download check report (.txt) to save the formula review. The mapped corrected example has three columns and three data rows: expect 0 fields to review out of 12 checked, including the header. For a deliberate warning, change the Notebook title in a practice copy to =SUM(2) and rescan; the report should identify record 3, line 3, column 2.
The two checker reports use Passed, Issues found or Check incomplete. An incomplete check, for example from a missing closing quote, is not a pass. Reports list up to 100 findings and state when more were found. They can include header names or field previews, so review them before sharing. Keep reports alongside the exact CSV you checked and rerun after changes.
5. Check the final saved file
- Run CSV Import Checker on the exact export you intend to import, especially after making corrections.
- Confirm required fields, currencies, decimal separators, identifiers and stock values against the destination's specification.
- Keep identifiers such as
0012as text if opening the file in a spreadsheet, so leading zeros are not lost. - Use the destination's preview or a small test import first and compare its results with your source records.
Choose or paste the file in each tool separately; the tools do not transfer data between tabs. A passing structure check and no formula-prefix flags do not guarantee a successful import.