CSV Desk focuses on everyday file problems: text that looks wrong, identifiers that change, and imports that need a second look.
Who is behind CSV Desk?
CSV Desk is published by JAZIA, a data analyst with a background in mathematics and experience working with academic, app and technology industry data. Contact JAZIA with a question or correction.
How the content is produced
These guides and practice datasets were developed with AI assistance. Each guide states its verification scope. The publisher’s professional background is separate from the specific checks recorded for a tutorial; a publishing credit does not mean every procedure has been personally tested by JAZIA.
Examples you can inspect
Our practice files are small, original datasets. They contain fictional records, explicit edge cases and expected values. You can download and inspect them without uploading your own spreadsheet.
Two kinds of verification
File-level checks compare decoded records, exact strings, field boundaries and file hashes. Application checks follow a named Excel import path and read the resulting cells. One does not substitute for the other.
The original eight guides include recorded checks from Excel for Mac 16.112.2: text imports, a semicolon import in Power Query, worksheet formulas, exact-row duplicate removal, and CSV export followed by a fresh import. Each article identifies the path and limits of its checks. Formula tests use entered sample values and do not establish every possible import path. Two additional guides were tested in Excel for Mac 16.113.2 on October 6, 2026: fixed-format date validation and exact two-way list comparison. Those tests wrote literal inputs into worksheet cells and recorded recalculated outputs; they do not certify new CSV-import paths. Two further guides were tested in the same Mac version on October 12, 2026: category lookups with duplicate and blank-label checks, and group totals with row coverage and missing-amount checks. Native range images show English worksheet content, and changes to inputs were recalculated and restored. Windows procedures remain documentation-based guidance, not observations from Windows Excel. The recorded application tests used a Chinese-language interface. These pages present English instructions and reference tables; they do not claim that an English-interface test was performed.
Two file-level tutorials were added on October 22, 2026: quoted CSV record boundaries and header-based combination. Their downloadable Python helpers and 37-assertion suite were run on macOS with Python 3.9.6. These are local file tests, not new Excel application tests. An earlier unsuccessful Excel Text Import Wizard multiline import is disclosed in the quoting guide.
On October 29, 2026, two further file-level guides added record-based splitting and keyed field-change comparison. Their shared practice suite passed 59 checks with Python 3.14.9 on macOS. The reports preserve exact decoded strings and disclose file-size and comparison limits. No new Excel or Windows application test is claimed.
On November 10, 2026, the quoting guide added a verified Power Query load in Excel for Mac 16.113.2. The three loaded note strings, including the embedded CRLF, matched the original fixture. The guide includes a native range capture and a separate dated test record, with the observed header layout and limits.
Sources and limitations
Microsoft documentation supports product-specific guidance. Each article links its sources and distinguishes them from our observations. A successful import on one version is not a promise that every version uses the same menus or defaults.
Use the files for practice
You may copy and adapt our fictional sample datasets for practice. Keep backups of your own data and compare results before replacing an original file.