The CSV format appears simple, but encoding, separator and value agreements vary from producer to producer. A prior check shall avoid late corrections.
Controlled items
- readable encoding, separator and number of columns;
- absent, empty or duplicated headers;
- lines too short or too long;
- missing values and identical lines;
- mix of numbers, dates, booleans and text in a column;
- cells starting with characters that can be interpreted as formulas.
Before a trade import
- Define the required columns and their expected type.
- Choose an explicit rule for empty fields.
- Normalize dates, decimals and identifiers.
- Decide whether duplicates are legitimate mistakes or events.
Analyse without storage
DataCheck processes the file in memory, does not run any cell and does not store content after query.
Validate my CSV fileInterpret the report
Atypical value is not necessarily false. The report indicates technical inconsistencies; the business decision remains to be made with the rules of the data set.
See also the method for detecting duplicates and CSV cleaning guide.
Updated on 19 July 2026.