Excel to JSON reads the first worksheet in an XLSX or XLS workbook, treats its first row as field names, and converts later rows into an object array. Blank cells usually omit a property, so the same key can be present in one record and absent in another.
Headers containing `>` paths, such as profile>name or items>0>sku, are reconstructed as nested objects and arrays. This matches the flattening convention used by JSON to Excel; ordinary headers remain first-level properties.