Check a Contact CSV for Personal Data Clues
Admin ·
Before sharing a CSV, identify columns that may hold contact details or identifiers. Pattern counts can direct a review while leaving individual records in the original file.
Compare four signals in the contact example
Select Load example and Scan columns. The scan covers three data records and six columns with 18 nonempty cells. Column 1 has three email-like values. Column 2 has two formatted phone-like values and a phone header clue. Column 3 has three IP literals.
Column 4 shows a Name header clue with zero pattern matches. The tool does not classify personal names by value. Column 6 includes an address-like substring in a note, which stays outside the whole-cell email pattern.
RFC 4180, CSV records and quotationRFC 5322 mailbox syntax, basis for the practical email subsetRFC 4291, IPv6 address textual representation
Read the denominator and confidence label
A Strong signal requires at least three matching cells covering 75% of nonempty cells, or an agreeing header plus two matches covering half. A weaker whole-cell match is a Possible signal. A recognized header can show a Header clue without any matches.
Turn off First record is a header and scan the same text. The scan now sees four data records and no header clues. The phone column becomes Possible signal with two matches out of four. Nonempty counts are measured after trimming surrounding whitespace.
RFC 4180, CSV records and quotationITU-T E.164 international telephone numbering plan
Verify the report omits names and cell contents
Download the report. It identifies columns by position and records counts and confidence. Match those positions against your local copy before deciding which columns to remove, mask or retain.
The rules cover common ASCII email syntax, formatted 10-to-15-digit phone-like values, isolated IPv4 and IPv6 literals, and three-two-four hyphenated US SSN shapes with basic excluded ranges. Labels can also suggest names, addresses or personal identifiers. Patterns can miss embedded or unusual values, and matches can be synthetic.
RFC 5322 mailbox syntax, basis for the practical email subsetITU-T E.164 international telephone numbering planRFC 4291, IPv6 address textual representationUS Social Security Administration, excluded SSN number ranges
Common questions
Will this scan confirm that an address or identifier exists?
It checks the supplied forms and selected headers locally. It does not query a service or match a value to a person.
Is a clean report enough to share the file?
A clean report only describes the selected rules and records. Review other identifiers and sensitive context separately.
Sources
Put it into practice.
- CSV PII Column Detector
- CSV Header Mapper to JSON Schema
- CSV Viewer and HTML Table Exporter
- CSV and TSV Validator