1. Identify the kind of failure
If a CSV will not parse, fix that first. A broken quote can change where several records appear to begin and end, so counts and ID checks would be misleading. The CSV import checker stops on malformed quoting rather than guessing at a repair.
For each example below, choose Clear, paste the complete snippet, leave First record contains headings checked, select the specified delimiter and choose Check CSV. The downloadable ZIP includes each snippet and a Read Me with expected results. Use the deliberately broken files only for this exercise.
| Symptom | First check |
|---|---|
| No results; unclosed or unexpected quote message | Inspect quotation marks in the indicated record. |
| Expected N columns; found M | Look for a delimiter inside an unquoted value or a missing/extra delimiter. |
| Several headings appear in one cell | Choose the source’s actual delimiter. |
| Record number differs from line number | Look for a quoted field with a line break. |
| Invalid UTF-8 or garbled characters | Re-export from the original source using UTF-8. |
2. Put commas inside a quoted field
With Comma (,) selected, this source has two headings but three values in the data record:
ContactID,Name
C001,Rivera, SamExpected: 1 record, 2 expected columns, 1 to fix. If the intended name is Rivera, Sam, enclose the whole name in straight double quotes:
ContactID,Name
C001,"Rivera, Sam"Expected: 1 record, 2 columns, 0 to fix, 0 to review. The parsed name contains the comma, but not the surrounding quotes. Do not delete every comma to eliminate the warning; the commas between fields are essential.
An extra trailing comma creates an extra empty field too. In a three-column file, C001,Alex, correctly supplies an empty third field; C001,Alex supplies only two fields.
3. Escape quotation marks inside a quoted value
Quotation marks inside a value must be represented without prematurely closing that field. This example closes the name too early:
ContactID,Name
C002,"Jo "JJ" Park"The checker stops with an unexpected-character error. Double each literal quotation mark inside the quoted field:
ContactID,Name
C002,"Jo ""JJ"" Park"Expected: one record, two columns and no findings. The preview should read Jo "JJ" Park. Use straight ASCII quotation marks for CSV syntax; curly quotes from a word processor are ordinary characters, not field delimiters.
An opening quote without a closing quote, such as C002,"Jo Park, produces an unclosed-field error. This strict checker also rejects spaces before an opening quote or after a closing quote. Prefer correcting the source export over a global search-and-replace that could alter names or notes. Claris describes doubled quotation marks in its CSV format notes.
4. Confirm the delimiter before interpreting counts
A file extension does not guarantee a comma separator. For this fixture, choose Semicolon (;):
ContactID;Name;City
C003;Alex Morgan;ParisExpected: one data record, three columns and no findings. With Comma selected, the same text is read as a one-column file and can show no findings. That is why you must compare the preview headings against the intended structure, even when the report is clear.
Auto-detect is a starting point. The tool supports comma, semicolon, tab and pipe, and examines a bounded sample to choose a likely separator. Override it when you know how the source was exported. If the source has no headings, turn off First record contains headings; the first record still establishes the expected column count.
5. Distinguish records from physical lines
A quoted CSV value can span physical lines. Try this with Comma selected:
ContactID,Note
C004,"Call Monday
Ask about renewal"
C005,ReadyExpected: 2 data records, 2 columns and no findings. The header is record 1. C004 is record 2, starting on line 2; C005 is record 3, starting on line 4. The line break inside the quotation marks belongs to C004’s Note.
When locating a finding, use the reported starting line and inspect the full quoted field. Do not split a large CSV at arbitrary physical lines; that can cut a record in half. Split by records with a CSV-aware exporter.
FileMaker exports have an additional detail: in-field carriage returns can be represented as vertical tabs, and repeating values as group separators. The checker flags those control characters for inspection; they are not automatically evidence of corruption. See Claris CSV export behavior.
6. Re-export encoding problems from the original
The file chooser accepts UTF-8, with or without an initial byte-order mark. It rejects invalid UTF-8 and NUL characters. Do not rename an Excel workbook from .xlsx to .csv; use its export command. If the file is UTF-16 or a legacy encoding, export a new UTF-8 copy from a program that reads the original correctly.
Names such as José and Nguyễn make useful spot checks against the source. A file can contain valid UTF-8 while already holding incorrectly decoded text. A clear checker report cannot prove those names are correct, and re-saving garbled text does not recover the original characters.
7. Recheck structure, then review the values
After repairing the source, reload it and confirm headings, column count and representative values. Then choose an ID column if it must be unique. Review leading zeros and dates separately using the six-contact walkthrough.
A clean structural result does not check existing FileMaker records or field rules. Test on a copy and inspect mapping, import action, auto-enter options and the resulting values before importing real data. Claris explains those choices in its import mapping guide.
Use Download report · TXT to keep findings, not to obtain a repaired CSV. The report includes excerpts; avoid sharing real customer values unnecessarily. Source contents stay in the browser tab. Reloading starts fresh, while any files you downloaded remain on your device.