Compare two CSV/Excel files by SKU, even with different column names
Sorting a spreadsheet changes row positions. It should not change which business record a row represents. Match records by a unique identifier first, then compare the fields you care about.
This guide uses synthetic data only. It describes a local browser tool and an Excel alternative. The tool interface is primarily Chinese.
A small inventory example
The old export has columns 商品编码 (product code) and 库存 (stock): A001=30, A002=50, A003=15, A005=9. The new export has SKU and 可用库存 (available stock), in a different order: A003=15, A001=26, A004=10, A005=9.
Expected result: one added record (A004), one removed record (A002), one changed record (A001, 30 → 26), and two unchanged records. Comparing one stock field produces three change-detail rows, excluding the CSV header. With more mapped fields, an added or removed record can produce multiple detail rows.
Download synthetic old CSV · Download synthetic new CSV
Use the visible controls
- Open the tool and click 先看示例 (try the example), or choose the old file (选择旧表文件) and the new file (选择新表文件). Check the first five preview rows. Select the appropriate XLSX sheet or CSV encoding.
- Under 主键 (record key), select 商品编码 in the old table and SKU in the new table. A key must be complete and unique within each table. For warehouse-level inventory, use two key fields in the same corresponding order on both sides. Do not use stock values as identifiers.
- Map 库存 to 可用库存. Matching headers are suggested; different headers need a manual choice. Each field can be used once. Select at least one non-key field for comparison.
- Click 开始对账 (compare), inspect the record counts, filter changes and search for a SKU. Matching ignores row order, but values use strict character comparison.
- Click 下载 diff-report.csv. Export always includes all changes, regardless of the current filter. Then use 清除文件和数据 to clear the page.
An Excel alternative
If your Excel version supports XLOOKUP/XMATCH, first validate unique text keys. Use an exact-match lookup to retrieve the new stock beside each old record, then compare old and new values; separately check the new keys against the old table to identify additions. Power Query can also merge by the identifier. These approaches require care with duplicates, types, blanks and case rules; they do not automatically reproduce this tool's strict text comparison. See the Chinese guide for the simple example formulas.
Limits and data interpretation
- CSV and XLSX only. Each file: up to 5 MiB, 10,000 data rows and 100 columns. Oversized inputs are rejected as a whole. XLSX expansion is limited to 40 MiB.
- No formula evaluation, macros, encrypted workbooks, formatting preservation or merged-cell expansion. A selected sheet containing formulas is rejected; paste as values first.
- CSV text keeps leading zeros and whitespace. 00123 differs from 123, and 1 differs from 1.0. Numeric text is preserved from the XLSX file, but identifier formatting already lost in Excel cannot be recovered. Dates become UTC ISO strings.
- The UTF-8 BOM report contains status, key, old field, new field, old value and new value. Formula-like values receive a leading apostrophe for safer spreadsheet import. Import identifier columns as text in Excel to avoid its own type conversion.
Privacy
Parsing, comparison and export happen in your browser. File contents, names, headers, keys, cells and search terms are not sent to a server. This build has no analytics, business event reporting or session replay. Optional feedback only creates text you may copy yourself; it is not submitted or counted. Clearing or refreshing discards page data but does not delete reports you downloaded.