Paste the sample below and select comma as its delimiter. Set the expected column count to two, then validate it. The final record contains three fields and should be reported as a structure error. Correct that record and validate again. This small test confirms that the selected settings describe your data before you inspect a larger file.
Configure header handling, empty-field allowances and optional type checks to match the source. Review each reported row and column against the original record, particularly when a quoted field spans several physical lines. Re-run validation after making corrections elsewhere. A successful check means the enabled rules passed; it does not prove that the dataset satisfies your organization’s business rules or that every value is factually correct.
Keep the selected checks with your report so another reviewer understands what a successful validation actually covered.
name,age
Alex,30
Jordan,25,extra
Validation parses quoting and separators and can check expected columns and permitted empty fields. Set the options deliberately: permissive settings may accept data another importer rejects. Expected column counts help expose accidental separators or missing fields. A green result describes the selected checks, not universal compatibility or the correctness of every business value.
No. Optional type checks infer likely column types using headers and values. A valid-looking date, email address or number may still be wrong for your application. Leading zeros and identifiers can need string treatment. Compare the inferred expectations with your actual schema before deciding that a reported mismatch should be corrected.
Read the reported row, column, severity and message, then inspect the corresponding source record. Correct the data in an editor and validate it again. Keep a copy of the original so a mistaken correction can be recovered. The validator reports issues; it does not automatically repair a spreadsheet or update a database.
The file upload allows 100 MiB. Browser memory holds input, parsed values, and diagnostics, so complexity can limit smaller files. No latency guarantee applies. Test settings with an excerpt containing quoted multiline fields and unusual records before running the complete dataset and interpreting its report.
Processing runs in your browser. Choosing Share uploads the selected tool state; anyone with the link can open it. Signing in to save sends content to your account library and records save history. Review the selection before either action. Closing the page does not delete those copies. Copying locally does not create a server share.