Common rejections

Encoding problem in attribute

An encoding fault damages every field at once, which is why Merchant Center reports it against attributes that look perfectly innocent in your editor. The fault is in how the file was written, not in the data.

Scan your feed for free

What Google requires

A consistently encoded file, ideally UTF-8

Google reads UTF-8, UTF-16 and Latin-1, but the file has to be one of them consistently, and what it declares has to match what it is. The failure mode is a file exported as Latin-1, reprocessed as UTF-8 somewhere in the pipeline, so every accented character, every e with an acute in a French product name, arrives as mojibake or as the replacement character.

What happens if you do not

Corrupt titles and descriptions, or a file that will not parse

Mild corruption shows as products with garbage characters in their visible titles, which shoppers see. Severe corruption breaks parsing: rows shift, quotes never close, and the import drops products without anyone deciding to remove them. A byte order mark glued to the first header cell is the special case that makes the first column unmappable while everything else looks fine.

How to fix it by hand

Fix the export, verify the bytes

Set the export to UTF-8 without BOM and keep every step of the pipeline, FTP, scripts, spreadsheets, from re-encoding it.
Check the actual bytes rather than the rendering: a file can display fine in an editor that guesses generously and still fail a strict parser.
If accents are already doubly damaged, re-export from the source rather than repairing text by hand.

Which of our checks apply

7 checks apply, read from the scanner's registry

The import gate decodes the file the way a strict parser does and names what it found: the encoding used, a removed BOM, replacement characters already baked in. Rows that cannot be parsed are counted and named, never dropped in silence.

encoding_not_utf8note
encoding_bom_removednote
encoding_replacement_charsat risk
encoding_binarystops the scan
header_control_charsat risk
quote_unclosedat risk
row_wrong_widthat risk

The quicker way

Let the scan find them

The scan reads your file byte by byte before anything else does, and tells you whether the encoding will survive Google's parser. The scan to identify errors is free and shows every finding before you pay anything. You only pay for the fix.

Scan your feed