Import a LinkedIn CSV without silently changing its fields

The short answer

Preserve the original CSV, inspect its encoding, delimiter and actual heading record, then import through a preview that lets you control column types. Keep identifiers and uncertain dates as text until their meaning is checked. Compare the imported records with the original; opening a file successfully does not prove that its values survived.

What you’ll take away

A CSV record can span several visible lines.
Automatic date and number conversion can change identifiers.
The local inspector reports structure, not archive completeness.

Keep a byte-identical original before opening a spreadsheet

Work on a separate copy. Note the source, download time and file size privately. A saved spreadsheet is a derivative, even if it has the same visible table. Keep the untouched source so you can resolve a disputed date or damaged character without guessing what it used to say.

This guide concerns importing a file you already have. It does not request account access. Use the fictional download to learn the controls before selecting a private archive. The example is constructed for this exercise and is not a LinkedIn account export or a sample of real connections.

Inspect records rather than counting line breaks

In common comma-separated CSV, quoted fields can contain commas and line breaks; a quote inside a quoted field is doubled. Splitting the text at every comma or newline breaks such records. RFC 4180 documents common conventions, while real files also use other dialects.

Our example includes Designer, Writer in one field and Lead followed by a line break and Writer in another. It has six logical records: a notice, a seven-field heading record and four data records. Choosing record 1 as the header is wrong for this file. Your own file may have a different layout; inspect it locally before selecting options.

Example setting	Value	Reason
Heading record	2	Record 1 is an explicit fictional notice
Exact-key column	3	Fictional profile reference
Email column	4	Optional presence check only

Use the local structure check before importing

Load the example, then run the check with the displayed settings. Expect four data records, seven fields, zero width mismatches, one extra identical parsed record and three blank email cells. One formula-like cell is flagged at record 6, field 5. These results describe the constructed input, not typical LinkedIn data.

The inspector accepts comma-separated UTF-8, an optional UTF-8 BOM, quoted fields and CRLF, LF or CR record endings. It rejects malformed quotes and files outside its published bounds instead of returning a shortened success. A one-column warning can indicate a delimiter mismatch. It does not guess another encoding or repair a semicolon-separated file.

MAKE IT YOUR OWN

Inspect a CSV locally

Your inputs stay on this page and are not sent to us. Copying or downloading happens only when you choose it.

Local structure checks for comma-separated UTF-8 with CSV quotation rules. Cell values, headings and filenames are excluded from the report and are not sent to us. The file is not changed or saved. A formula flag is a review prompt, not a safety verdict.

Fictional CSV (473 bytes)Examples, source and checks (ZIP)Instructions and limits

The fictional CSV includes =1+1 as a harmless formula example. Inspect as text first; a spreadsheet may calculate the expression.

Choose types in the import preview

In Excel, use the text/CSV import controls rather than relying on a double-click and default interpretation. Microsoft documents preview and delimiter selection; the Text Import Wizard also offers a Text column format. Controls differ between editions. The key is to inspect the preview and prevent unwanted conversion before loading.

For example, an illustrative identifier 00123 must remain five characters if that is how the source identifies a record. Keep ambiguous 03/04/2026 as text until the source date convention is established. Do not resolve uncertainty by whichever calendar date your computer happens to choose. Names and URLs also deserve a text-preserving import.

Select the working copy in your spreadsheet’s import controls.
Choose the actual delimiter and inspect quoted multiline fields.
Set identifiers and uncertain date fields to text using the available column controls.
Load the table, then compare representative values and record counts with the source.

Resolve damaged characters at the source

Character encoding maps stored bytes to characters; changing a font cannot repair a wrong decoding. A UTF-8 validation pass establishes that the bytes can be decoded that way, not that every name is correct. Text already damaged and saved as valid UTF-8 can still be wrong.

If accented names look wrong, retain the source and check the exporter’s encoding plus the importer’s file-origin setting. An actual replacement character in this inspector produces a warning. Do not apply blanket name replacements: they can corrupt legitimate text. Resolve an important spelling against an authorized source or leave the issue explicit.

Accept an import only after a small reconciliation

Check the heading width, logical data-record count, one quoted comma, one multiline field, one leading-zero identifier if present, and a known date representation. Preserve blanks as blanks. If you export a derivative back to CSV, inspect that actual file again; a spreadsheet preview alone is insufficient.

The example’s SHA-256 is available in its expected-results JSON and the report. Matching it identifies the bytes used for this exercise; it does not certify a private file’s truth or permissions. The ZIP includes source, tests and an offline example runner. No real account data is required to reproduce the exercise.