After you double-click a CSV file, the table initially looks tidy. Later, you discover that an identifier became a number and a date was interpreted differently. The spreadsheet application classified the text values automatically. A controlled import gives you a chance to inspect the preview and specify column types before the data becomes the basis for further work.
Preserve the original source
Save the received CSV unchanged. If useful, open a copy in a text editor to inspect the actual characters. This helps distinguish a defect in the source from a change introduced during import. Avoid accidentally saving it back with a different encoding.
Record the expected columns and a few known check values. An attendee identifier such as 00417 is useful, along with a confirmed date and a name containing an accented character. A date such as 03/04/2026 is ambiguous without an agreed order. Ask what it means rather than guessing from your own regional settings.
Decide what kind of information each column contains. A quantity is a number; a postal code or item identifier is often text. A value consisting entirely of digits is not necessarily something that should be used in arithmetic.
Use the import preview deliberately
Use the application's text or CSV import workflow. Excel provides a From Text/CSV route, while LibreOffice Calc offers text-import options. The precise interface varies by version and platform.
Check the delimiter and character encoding first. Each heading should sit above the intended column, and accented characters should display correctly. Inspect a row containing a comma within a description as well. If it creates an extra column, field enclosure is not being handled correctly.
Explicitly set identifier columns to text. For dates, choose the order used by the source, or initially import them as text if their meaning remains unresolved. Decimal and grouping separators must also match the chosen interpretation. A value written as 1,250 can mean very different things under different conventions.
In tools that create automatic transformation steps, changing the display format later may be insufficient. Check whether an earlier step has already converted values to numbers or dates. The required type handling must occur before original characters are lost.
Check beyond the first row
After loading, compare your known check values with the unchanged source. Also inspect record count, final records, empty fields, and unusually long identifiers. A preview may show only a sample, with uncommon cases occurring farther down.
Only then build calculations or joins. If you discover a mistake, repeating a clean import is usually more reliable than manually repairing the few conspicuous cells. For recurring deliveries, document the verified import choices or save them as an understandable query.
A new file from the same supplier may still add a column or change its date representation. Keep the short verification steps even for repeat imports. Automation is most useful when it exposes such differences rather than silently passing them downstream.